Merge remote-tracking branch 'origin/master' into ui_remove_qt_spinner_text

cereal/log.capnp

@@ -1174,6 +1174,8 @@ struct ModelDataV2 {
 
   struct Action {
     desiredCurvature @0 :Float32;
+    desiredAcceleration @1 :Float32;
+    shouldStop @2 :Bool;
   }
 }
 
@@ -2284,6 +2286,7 @@ struct LiveDelayData {
   status @2 :Status;
 
   lateralDelayEstimate @3 :Float32;
+  lateralDelayEstimateStd @5 :Float32;
   points @4 :List(Float32);
 
   enum Status {

selfdrive/controls/lib/drive_helpers.py

@@ -1,7 +1,7 @@
 import numpy as np
 from cereal import log
 from opendbc.car.vehicle_model import ACCELERATION_DUE_TO_GRAVITY
-from openpilot.common.realtime import DT_CTRL
+from openpilot.common.realtime import DT_CTRL, DT_MDL
 
 MIN_SPEED = 1.0
 CONTROL_N = 17
@@ -19,6 +19,9 @@ def clamp(val, min_val, max_val):
   clamped_val = float(np.clip(val, min_val, max_val))
   return clamped_val, clamped_val != val
 
+def smooth_value(val, prev_val, tau, dt=DT_MDL):
+  alpha = 1 - np.exp(-dt/tau) if tau > 0 else 1
+  return alpha * val + (1 - alpha) * prev_val
 
 def clip_curvature(v_ego, prev_curvature, new_curvature, roll):
   # This function respects ISO lateral jerk and acceleration limits + a max curvature
@@ -43,3 +46,19 @@ def get_speed_error(modelV2: log.ModelDataV2, v_ego: float) -> float:
     vel_err = np.clip(modelV2.temporalPose.trans[0] - v_ego, -MAX_VEL_ERR, MAX_VEL_ERR)
     return float(vel_err)
   return 0.0
+
+
+def get_accel_from_plan(speeds, accels, t_idxs, action_t=DT_MDL, vEgoStopping=0.05):
+  if len(speeds) == len(t_idxs):
+    v_now = speeds[0]
+    a_now = accels[0]
+    v_target = np.interp(action_t, t_idxs, speeds)
+    a_target = 2 * (v_target - v_now) / (action_t) - a_now
+    v_target_1sec = np.interp(action_t + 1.0, t_idxs, speeds)
+  else:
+    v_target = 0.0
+    v_target_1sec = 0.0
+    a_target = 0.0
+  should_stop = (v_target < vEgoStopping and
+                 v_target_1sec < vEgoStopping)
+  return a_target, should_stop

selfdrive/controls/lib/longitudinal_planner.py

@@ -11,7 +11,7 @@ from openpilot.selfdrive.modeld.constants import ModelConstants
 from openpilot.selfdrive.controls.lib.longcontrol import LongCtrlState
 from openpilot.selfdrive.controls.lib.longitudinal_mpc_lib.long_mpc import LongitudinalMpc
 from openpilot.selfdrive.controls.lib.longitudinal_mpc_lib.long_mpc import T_IDXS as T_IDXS_MPC
-from openpilot.selfdrive.controls.lib.drive_helpers import CONTROL_N, get_speed_error
+from openpilot.selfdrive.controls.lib.drive_helpers import CONTROL_N, get_speed_error, get_accel_from_plan
 from openpilot.selfdrive.car.cruise import V_CRUISE_MAX, V_CRUISE_UNSET
 from openpilot.common.swaglog import cloudlog
 
@@ -48,27 +48,11 @@ def limit_accel_in_turns(v_ego, angle_steers, a_target, CP):
   return [a_target[0], min(a_target[1], a_x_allowed)]
 
 
-def get_accel_from_plan(speeds, accels, action_t=DT_MDL, vEgoStopping=0.05):
-  if len(speeds) == CONTROL_N:
-    v_now = speeds[0]
-    a_now = accels[0]
-
-    v_target = np.interp(action_t, CONTROL_N_T_IDX, speeds)
-    a_target = 2 * (v_target - v_now) / (action_t) - a_now
-    v_target_1sec = np.interp(action_t + 1.0, CONTROL_N_T_IDX, speeds)
-  else:
-    v_target = 0.0
-    v_target_1sec = 0.0
-    a_target = 0.0
-  should_stop = (v_target < vEgoStopping and
-                 v_target_1sec < vEgoStopping)
-  return a_target, should_stop
-
-
 class LongitudinalPlanner:
   def __init__(self, CP, init_v=0.0, init_a=0.0, dt=DT_MDL):
     self.CP = CP
     self.mpc = LongitudinalMpc(dt=dt)
+    self.mpc.mode = 'acc'
     self.fcw = False
     self.dt = dt
     self.allow_throttle = True
@@ -106,7 +90,7 @@ class LongitudinalPlanner:
     return x, v, a, j, throttle_prob
 
   def update(self, sm):
-    self.mpc.mode = 'blended' if sm['selfdriveState'].experimentalMode else 'acc'
+    self.mode = 'blended' if sm['selfdriveState'].experimentalMode else 'acc'
 
     if len(sm['carControl'].orientationNED) == 3:
       accel_coast = get_coast_accel(sm['carControl'].orientationNED[1])
@@ -129,7 +113,7 @@ class LongitudinalPlanner:
     # No change cost when user is controlling the speed, or when standstill
     prev_accel_constraint = not (reset_state or sm['carState'].standstill)
 
-    if self.mpc.mode == 'acc':
+    if self.mode == 'acc':
       accel_clip = [ACCEL_MIN, get_max_accel(v_ego)]
       steer_angle_without_offset = sm['carState'].steeringAngleDeg - sm['liveParameters'].angleOffsetDeg
       accel_clip = limit_accel_in_turns(v_ego, steer_angle_without_offset, accel_clip, self.CP)
@@ -176,8 +160,17 @@ class LongitudinalPlanner:
     self.v_desired_filter.x = self.v_desired_filter.x + self.dt * (self.a_desired + a_prev) / 2.0
 
     action_t =  self.CP.longitudinalActuatorDelay + DT_MDL
-    output_a_target, self.output_should_stop = get_accel_from_plan(self.v_desired_trajectory, self.a_desired_trajectory,
+    output_a_target_mpc, output_should_stop_mpc = get_accel_from_plan(self.v_desired_trajectory, self.a_desired_trajectory, CONTROL_N_T_IDX,
                                                                         action_t=action_t, vEgoStopping=self.CP.vEgoStopping)
+    output_a_target_e2e = sm['modelV2'].action.desiredAcceleration
+    output_should_stop_e2e = sm['modelV2'].action.shouldStop
+
+    if self.mode == 'acc':
+      output_a_target = output_a_target_mpc
+      self.output_should_stop = output_should_stop_mpc
+    else:
+      output_a_target = min(output_a_target_mpc, output_a_target_e2e)
+      self.output_should_stop = output_should_stop_e2e or output_should_stop_mpc
 
     for idx in range(2):
       accel_clip[idx] = np.clip(accel_clip[idx], self.prev_accel_clip[idx] - 0.05, self.prev_accel_clip[idx] + 0.05)

selfdrive/locationd/lagd.py

@@ -21,8 +21,12 @@ MIN_OKAY_WINDOW_SEC = 25.0
 MIN_RECOVERY_BUFFER_SEC = 2.0
 MIN_VEGO = 15.0
 MIN_ABS_YAW_RATE = np.radians(1.0)
+MAX_YAW_RATE_SANITY_CHECK = 1.0
 MIN_NCC = 0.95
 MAX_LAG = 1.0
+MAX_LAG_STD = 0.1
+MAX_LAT_ACCEL = 2.0
+MAX_LAT_ACCEL_DIFF = 0.6
 
 
 def masked_normalized_cross_correlation(expected_sig: np.ndarray, actual_sig: np.ndarray, mask: np.ndarray, n: int):
@@ -124,13 +128,23 @@ class BlockAverage:
       self.block_idx = (self.block_idx + 1) % self.num_blocks
       self.valid_blocks = min(self.valid_blocks + 1, self.num_blocks)
 
-  def get(self) -> tuple[float, float]:
+  def get(self) -> tuple[float, float, float, float]:
     valid_block_idx = [i for i in range(self.valid_blocks) if i != self.block_idx]
     valid_and_current_idx = valid_block_idx + ([self.block_idx] if self.idx > 0 else [])
 
-    valid_mean = float(np.mean(self.values[valid_block_idx], axis=0).item()) if len(valid_block_idx) > 0 else float('nan')
-    current_mean = float(np.mean(self.values[valid_and_current_idx], axis=0).item()) if len(valid_and_current_idx) > 0 else float('nan')
-    return valid_mean, current_mean
+    if len(valid_block_idx) > 0:
+      valid_mean = float(np.mean(self.values[valid_block_idx], axis=0).item())
+      valid_std = float(np.std(self.values[valid_block_idx], axis=0).item())
+    else:
+      valid_mean, valid_std = float('nan'), float('nan')
+
+    if len(valid_and_current_idx) > 0:
+      current_mean = float(np.mean(self.values[valid_and_current_idx], axis=0).item())
+      current_std = float(np.std(self.values[valid_and_current_idx], axis=0).item())
+    else:
+      current_mean, current_std = float('nan'), float('nan')
+
+    return valid_mean, valid_std, current_mean, current_std
 
 
 class LateralLagEstimator:
@@ -139,7 +153,8 @@ class LateralLagEstimator:
   def __init__(self, CP: car.CarParams, dt: float,
                block_count: int = BLOCK_NUM, min_valid_block_count: int = BLOCK_NUM_NEEDED, block_size: int = BLOCK_SIZE,
                window_sec: float = MOVING_WINDOW_SEC, okay_window_sec: float = MIN_OKAY_WINDOW_SEC, min_recovery_buffer_sec: float = MIN_RECOVERY_BUFFER_SEC,
-               min_vego: float = MIN_VEGO, min_yr: float = MIN_ABS_YAW_RATE, min_ncc: float = MIN_NCC):
+               min_vego: float = MIN_VEGO, min_yr: float = MIN_ABS_YAW_RATE, min_ncc: float = MIN_NCC,
+               max_lat_accel: float = MAX_LAT_ACCEL, max_lat_accel_diff: float = MAX_LAT_ACCEL_DIFF):
     self.dt = dt
     self.window_sec = window_sec
     self.okay_window_sec = okay_window_sec
@@ -151,6 +166,8 @@ class LateralLagEstimator:
     self.min_vego = min_vego
     self.min_yr = min_yr
     self.min_ncc = min_ncc
+    self.max_lat_accel = max_lat_accel
+    self.max_lat_accel_diff = max_lat_accel_diff
 
     self.t = 0.0
     self.lat_active = False
@@ -159,10 +176,13 @@ class LateralLagEstimator:
     self.desired_curvature = 0.0
     self.v_ego = 0.0
     self.yaw_rate = 0.0
+    self.yaw_rate_std = 0.0
+    self.pose_valid = False
 
     self.last_lat_inactive_t = 0.0
     self.last_steering_pressed_t = 0.0
     self.last_steering_saturated_t = 0.0
+    self.last_pose_invalid_t = 0.0
     self.last_estimate_t = 0.0
 
     self.calibrator = PoseCalibrator()
@@ -181,17 +201,27 @@ class LateralLagEstimator:
 
     liveDelay = msg.liveDelay
 
-    valid_mean_lag, current_mean_lag = self.block_avg.get()
-    if self.block_avg.valid_blocks >= self.min_valid_block_count and not np.isnan(valid_mean_lag):
-      liveDelay.status = log.LiveDelayData.Status.estimated
-      liveDelay.lateralDelay = valid_mean_lag
+    valid_mean_lag, valid_std, current_mean_lag, current_std = self.block_avg.get()
+    if self.block_avg.valid_blocks >= self.min_valid_block_count and not np.isnan(valid_mean_lag) and not np.isnan(valid_std):
+      if valid_std > MAX_LAG_STD:
+        liveDelay.status = log.LiveDelayData.Status.invalid
+      else:
+        liveDelay.status = log.LiveDelayData.Status.estimated
     else:
       liveDelay.status = log.LiveDelayData.Status.unestimated
+
+    if liveDelay.status == log.LiveDelayData.Status.estimated:
+      liveDelay.lateralDelay = valid_mean_lag
+    else:
       liveDelay.lateralDelay = self.initial_lag
-    if not np.isnan(current_mean_lag):
+
+    if not np.isnan(current_mean_lag) and not np.isnan(current_std):
       liveDelay.lateralDelayEstimate = current_mean_lag
+      liveDelay.lateralDelayEstimateStd = current_std
     else:
       liveDelay.lateralDelayEstimate = self.initial_lag
+      liveDelay.lateralDelayEstimateStd = 0.0
+
     liveDelay.validBlocks = self.block_avg.valid_blocks
     if debug:
       liveDelay.points = self.block_avg.values.flatten().tolist()
@@ -212,7 +242,9 @@ class LateralLagEstimator:
     elif which == "livePose":
       device_pose = Pose.from_live_pose(msg)
       calibrated_pose = self.calibrator.build_calibrated_pose(device_pose)
-      self.yaw_rate = calibrated_pose.angular_velocity.z
+      self.yaw_rate = calibrated_pose.angular_velocity.yaw
+      self.yaw_rate_std = calibrated_pose.angular_velocity.yaw_std
+      self.pose_valid = msg.angularVelocityDevice.valid and msg.posenetOK and msg.inputsOK
     self.t = t
 
   def points_enough(self):
@@ -222,23 +254,30 @@ class LateralLagEstimator:
     return self.points.num_okay >= int(self.okay_window_sec / self.dt)
 
   def update_points(self):
+    la_desired = self.desired_curvature * self.v_ego * self.v_ego
+    la_actual_pose = self.yaw_rate * self.v_ego
+
+    fast = self.v_ego > self.min_vego
+    turning = np.abs(self.yaw_rate) >= self.min_yr
+    sensors_valid = self.pose_valid and np.abs(self.yaw_rate) < MAX_YAW_RATE_SANITY_CHECK and self.yaw_rate_std < MAX_YAW_RATE_SANITY_CHECK
+    la_valid = np.abs(la_actual_pose) <= self.max_lat_accel and np.abs(la_desired - la_actual_pose) <= self.max_lat_accel_diff
+    calib_valid = self.calibrator.calib_valid
+
     if not self.lat_active:
       self.last_lat_inactive_t = self.t
     if self.steering_pressed:
       self.last_steering_pressed_t = self.t
     if self.steering_saturated:
       self.last_steering_saturated_t = self.t
+    if not sensors_valid or not la_valid:
+      self.last_pose_invalid_t = self.t
 
-    la_desired = self.desired_curvature * self.v_ego * self.v_ego
-    la_actual_pose = self.yaw_rate * self.v_ego
-
-    fast = self.v_ego > self.min_vego
-    turning = np.abs(self.yaw_rate) >= self.min_yr
-    has_recovered = all( # wait for recovery after !lat_active, steering_pressed, steering_saturated
+    has_recovered = all( # wait for recovery after !lat_active, steering_pressed, steering_saturated, !sensors/la_valid
       self.t - last_t >= self.min_recovery_buffer_sec
-      for last_t in [self.last_lat_inactive_t, self.last_steering_pressed_t, self.last_steering_saturated_t]
+      for last_t in [self.last_lat_inactive_t, self.last_steering_pressed_t, self.last_steering_saturated_t, self.last_pose_invalid_t]
     )
-    okay = self.lat_active and not self.steering_pressed and not self.steering_saturated and fast and turning and has_recovered
+    okay = self.lat_active and not self.steering_pressed and not self.steering_saturated and \
+           fast and turning and has_recovered and calib_valid and sensors_valid and la_valid
 
     self.points.update(self.t, la_desired, la_actual_pose, okay)
 
@@ -288,8 +327,9 @@ def retrieve_initial_lag(params_reader: Params, CP: car.CarParams):
         if last_CP.carFingerprint != CP.carFingerprint:
           raise Exception("Car model mismatch")
 
-        lag, valid_blocks = ld.lateralDelayEstimate, ld.validBlocks
+        lag, valid_blocks, status = ld.lateralDelayEstimate, ld.validBlocks, ld.status
         assert valid_blocks <= BLOCK_NUM, "Invalid number of valid blocks"
+        assert status != log.LiveDelayData.Status.invalid, "Lag estimate is invalid"
         return lag, valid_blocks
     except Exception as e:
       cloudlog.error(f"Failed to retrieve initial lag: {e}")

selfdrive/locationd/test/test_lagd.py

@@ -3,7 +3,7 @@ import numpy as np
 import time
 import pytest
 
-from cereal import messaging
+from cereal import messaging, log
 from openpilot.selfdrive.locationd.lagd import LateralLagEstimator, retrieve_initial_lag, masked_normalized_cross_correlation, \
                                                BLOCK_NUM_NEEDED, BLOCK_SIZE, MIN_OKAY_WINDOW_SEC
 from openpilot.selfdrive.test.process_replay.migration import migrate, migrate_carParams
@@ -23,8 +23,8 @@ def process_messages(mocker, estimator, lag_frames, n_frames, vego=20.0, rejecti
 
   for i in range(n_frames):
     t = i * estimator.dt
-    desired_la = np.cos(t)
-    actual_la = np.cos(t - lag_frames * estimator.dt)
+    desired_la = np.cos(t) * 0.1
+    actual_la = np.cos(t - lag_frames * estimator.dt) * 0.1
 
     # if sample is masked out, set it to desired value (no lag)
     rejected = random.uniform(0, 1) < rejection_threshold
@@ -41,7 +41,9 @@ def process_messages(mocker, estimator, lag_frames, n_frames, vego=20.0, rejecti
       (t, "livePose", mocker.Mock(orientationNED=ZeroMock(),
                                 velocityDevice=ZeroMock(),
                                 accelerationDevice=ZeroMock(),
-                                angularVelocityDevice=ZeroMock(z=actual_yr))),
+                                angularVelocityDevice=ZeroMock(z=actual_yr, valid=True),
+                                posenetOK=True, inputsOK=True)),
+      (t, "liveCalibration", mocker.Mock(rpyCalib=[0, 0, 0], calStatus=log.LiveCalibrationData.Status.calibrated)),
     ]
     for t, w, m in msgs:
       estimator.handle_log(t, w, m)
@@ -111,6 +113,7 @@ class TestLagd:
         assert msg.liveDelay.status == 'estimated'
         assert np.allclose(msg.liveDelay.lateralDelay, lag_frames * DT, atol=0.01)
         assert np.allclose(msg.liveDelay.lateralDelayEstimate, lag_frames * DT, atol=0.01)
+        assert np.allclose(msg.liveDelay.lateralDelayEstimateStd, 0.0, atol=0.01)
         assert msg.liveDelay.validBlocks == BLOCK_NUM_NEEDED
 
   def test_estimator_masking(self, mocker):
@@ -119,6 +122,7 @@ class TestLagd:
     process_messages(mocker, estimator, lag_frames, (int(MIN_OKAY_WINDOW_SEC / DT) + BLOCK_SIZE) * 2, rejection_threshold=0.4)
     msg = estimator.get_msg(True)
     assert np.allclose(msg.liveDelay.lateralDelayEstimate, lag_frames * DT, atol=0.01)
+    assert np.allclose(msg.liveDelay.lateralDelayEstimateStd, 0.0, atol=0.01)
 
   @pytest.mark.skipif(PC, reason="only on device")
   @pytest.mark.timeout(60)

selfdrive/modeld/fill_model_msg.py

@@ -56,7 +56,7 @@ def fill_lane_line_meta(builder, lane_lines, lane_line_probs):
   builder.rightProb = lane_line_probs[2]
 
 def fill_model_msg(base_msg: capnp._DynamicStructBuilder, extended_msg: capnp._DynamicStructBuilder,
-                   net_output_data: dict[str, np.ndarray], v_ego: float, delay: float,
+                   net_output_data: dict[str, np.ndarray], action: log.ModelDataV2.Action,
                    publish_state: PublishState, vipc_frame_id: int, vipc_frame_id_extra: int,
                    frame_id: int, frame_drop: float, timestamp_eof: int, model_execution_time: float,
                    valid: bool) -> None:
@@ -71,7 +71,8 @@ def fill_model_msg(base_msg: capnp._DynamicStructBuilder, extended_msg: capnp._D
   driving_model_data.frameIdExtra = vipc_frame_id_extra
   driving_model_data.frameDropPerc = frame_drop_perc
   driving_model_data.modelExecutionTime = model_execution_time
-  driving_model_data.action.desiredCurvature = float(net_output_data['desired_curvature'][0,0])
+
+  driving_model_data.action = action
 
   modelV2 = extended_msg.modelV2
   modelV2.frameId = vipc_frame_id
@@ -89,17 +90,17 @@ def fill_model_msg(base_msg: capnp._DynamicStructBuilder, extended_msg: capnp._D
   fill_xyzt(modelV2.orientationRate, ModelConstants.T_IDXS, *net_output_data['plan'][0,:,Plan.ORIENTATION_RATE].T)
 
   # temporal pose
-  temporal_pose = modelV2.temporalPose
-  temporal_pose.trans = net_output_data['sim_pose'][0,:ModelConstants.POSE_WIDTH//2].tolist()
-  temporal_pose.transStd = net_output_data['sim_pose_stds'][0,:ModelConstants.POSE_WIDTH//2].tolist()
-  temporal_pose.rot = net_output_data['sim_pose'][0,ModelConstants.POSE_WIDTH//2:].tolist()
-  temporal_pose.rotStd = net_output_data['sim_pose_stds'][0,ModelConstants.POSE_WIDTH//2:].tolist()
+  #temporal_pose = modelV2.temporalPose
+  #temporal_pose.trans = net_output_data['sim_pose'][0,:ModelConstants.POSE_WIDTH//2].tolist()
+  #temporal_pose.transStd = net_output_data['sim_pose_stds'][0,:ModelConstants.POSE_WIDTH//2].tolist()
+  #temporal_pose.rot = net_output_data['sim_pose'][0,ModelConstants.POSE_WIDTH//2:].tolist()
+  #temporal_pose.rotStd = net_output_data['sim_pose_stds'][0,ModelConstants.POSE_WIDTH//2:].tolist()
 
   # poly path
   fill_xyz_poly(driving_model_data.path, ModelConstants.POLY_PATH_DEGREE, *net_output_data['plan'][0,:,Plan.POSITION].T)
 
-  # lateral planning
-  modelV2.action.desiredCurvature = float(net_output_data['desired_curvature'][0,0])
+  # action
+  modelV2.action = action
 
   # times at X_IDXS of edges and lines aren't used
   LINE_T_IDXS: list[float] = []

selfdrive/modeld/modeld.py

@@ -21,14 +21,15 @@ from opendbc.car.car_helpers import get_demo_car_params
 from openpilot.common.swaglog import cloudlog
 from openpilot.common.params import Params
 from openpilot.common.filter_simple import FirstOrderFilter
-from openpilot.common.realtime import config_realtime_process
+from openpilot.common.realtime import config_realtime_process, DT_MDL
 from openpilot.common.transformations.camera import DEVICE_CAMERAS
 from openpilot.common.transformations.model import get_warp_matrix
 from openpilot.system import sentry
 from openpilot.selfdrive.controls.lib.desire_helper import DesireHelper
+from openpilot.selfdrive.controls.lib.drive_helpers import get_accel_from_plan, smooth_value
 from openpilot.selfdrive.modeld.parse_model_outputs import Parser
 from openpilot.selfdrive.modeld.fill_model_msg import fill_model_msg, fill_pose_msg, PublishState
-from openpilot.selfdrive.modeld.constants import ModelConstants
+from openpilot.selfdrive.modeld.constants import ModelConstants, Plan
 from openpilot.selfdrive.modeld.models.commonmodel_pyx import DrivingModelFrame, CLContext
 
 
@@ -40,6 +41,26 @@ POLICY_PKL_PATH = Path(__file__).parent / 'models/driving_policy_tinygrad.pkl'
 VISION_METADATA_PATH = Path(__file__).parent / 'models/driving_vision_metadata.pkl'
 POLICY_METADATA_PATH = Path(__file__).parent / 'models/driving_policy_metadata.pkl'
 
+LAT_SMOOTH_SECONDS = 0.1
+LONG_SMOOTH_SECONDS = 0.3
+
+
+def get_action_from_model(model_output: dict[str, np.ndarray], prev_action: log.ModelDataV2.Action,
+                          lat_action_t: float, long_action_t: float,) -> log.ModelDataV2.Action:
+    plan = model_output['plan'][0]
+    desired_accel, should_stop = get_accel_from_plan(plan[:,Plan.VELOCITY][:,0],
+                                                     plan[:,Plan.ACCELERATION][:,0],
+                                                     ModelConstants.T_IDXS,
+                                                     action_t=long_action_t)
+    desired_accel = smooth_value(desired_accel, prev_action.desiredAcceleration, LONG_SMOOTH_SECONDS)
+
+    desired_curvature = model_output['desired_curvature'][0, 0]
+    desired_curvature = smooth_value(desired_curvature, prev_action.desiredCurvature, LAT_SMOOTH_SECONDS)
+
+    return log.ModelDataV2.Action(desiredCurvature=float(desired_curvature),
+                                  desiredAcceleration=float(desired_accel),
+                                  shouldStop=bool(should_stop))
+
 class FrameMeta:
   frame_id: int = 0
   timestamp_sof: int = 0
@@ -147,7 +168,7 @@ class ModelState:
     # TODO model only uses last value now
     self.full_prev_desired_curv[0,:-1] = self.full_prev_desired_curv[0,1:]
     self.full_prev_desired_curv[0,-1,:] = policy_outputs_dict['desired_curvature'][0, :]
-    self.numpy_inputs['prev_desired_curv'][:] = self.full_prev_desired_curv[0, self.temporal_idxs]
+    self.numpy_inputs['prev_desired_curv'][:] = 0*self.full_prev_desired_curv[0, self.temporal_idxs]
 
     combined_outputs_dict = {**vision_outputs_dict, **policy_outputs_dict}
     if SEND_RAW_PRED:
@@ -221,7 +242,10 @@ def main(demo=False):
   cloudlog.info("modeld got CarParams: %s", CP.brand)
 
   # TODO this needs more thought, use .2s extra for now to estimate other delays
-  steer_delay = CP.steerActuatorDelay + .2
+  # TODO Move smooth seconds to action function
+  lat_delay = CP.steerActuatorDelay + .2 + LAT_SMOOTH_SECONDS
+  long_delay = CP.longitudinalActuatorDelay + LONG_SMOOTH_SECONDS
+  prev_action = log.ModelDataV2.Action()
 
   DH = DesireHelper()
 
@@ -263,7 +287,7 @@ def main(demo=False):
     is_rhd = sm["driverMonitoringState"].isRHD
     frame_id = sm["roadCameraState"].frameId
     v_ego = max(sm["carState"].vEgo, 0.)
-    lateral_control_params = np.array([v_ego, steer_delay], dtype=np.float32)
+    lateral_control_params = np.array([v_ego, lat_delay], dtype=np.float32)
     if sm.updated["liveCalibration"] and sm.seen['roadCameraState'] and sm.seen['deviceState']:
       device_from_calib_euler = np.array(sm["liveCalibration"].rpyCalib, dtype=np.float32)
       dc = DEVICE_CAMERAS[(str(sm['deviceState'].deviceType), str(sm['roadCameraState'].sensor))]
@@ -306,7 +330,10 @@ def main(demo=False):
       modelv2_send = messaging.new_message('modelV2')
       drivingdata_send = messaging.new_message('drivingModelData')
       posenet_send = messaging.new_message('cameraOdometry')
-      fill_model_msg(drivingdata_send, modelv2_send, model_output, v_ego, steer_delay,
+
+      action = get_action_from_model(model_output, prev_action, lat_delay + DT_MDL, long_delay + DT_MDL)
+      prev_action = action
+      fill_model_msg(drivingdata_send, modelv2_send, model_output, action,
                      publish_state, meta_main.frame_id, meta_extra.frame_id, frame_id,
                      frame_drop_ratio, meta_main.timestamp_eof, model_execution_time, live_calib_seen)
 

selfdrive/modeld/models/driving_policy.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:98f0121ccb6f850077b04cc91bd33d370fc6cbdc2bd35f1ab55628a15a813f36
-size 15966721
+oid sha256:84061882148485dabfdd23ee7e0925284a4370d0ae5be804d6099ae847352891
+size 15578328

selfdrive/modeld/models/driving_vision.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:897f80d0388250f99bba69b6a8434560cc0fd83157cbeb0bc134c67fe4e64624
-size 34882971
+oid sha256:f222d2c528f1763828de01bb55e8979b1e4056e1dbb41350f521d2d2bb09d177
+size 46265585

selfdrive/modeld/parse_model_outputs.py

@@ -88,6 +88,12 @@ class Parser:
     self.parse_mdn('pose', outs, in_N=0, out_N=0, out_shape=(ModelConstants.POSE_WIDTH,))
     self.parse_mdn('wide_from_device_euler', outs, in_N=0, out_N=0, out_shape=(ModelConstants.WIDE_FROM_DEVICE_WIDTH,))
     self.parse_mdn('road_transform', outs, in_N=0, out_N=0, out_shape=(ModelConstants.POSE_WIDTH,))
+    self.parse_mdn('lane_lines', outs, in_N=0, out_N=0, out_shape=(ModelConstants.NUM_LANE_LINES,ModelConstants.IDX_N,ModelConstants.LANE_LINES_WIDTH))
+    self.parse_mdn('road_edges', outs, in_N=0, out_N=0, out_shape=(ModelConstants.NUM_ROAD_EDGES,ModelConstants.IDX_N,ModelConstants.LANE_LINES_WIDTH))
+    self.parse_mdn('lead', outs, in_N=ModelConstants.LEAD_MHP_N, out_N=ModelConstants.LEAD_MHP_SELECTION,
+                   out_shape=(ModelConstants.LEAD_TRAJ_LEN,ModelConstants.LEAD_WIDTH))
+    for k in ['lead_prob', 'lane_lines_prob']:
+      self.parse_binary_crossentropy(k, outs)
     self.parse_categorical_crossentropy('desire_pred', outs, out_shape=(ModelConstants.DESIRE_PRED_LEN,ModelConstants.DESIRE_PRED_WIDTH))
     self.parse_binary_crossentropy('meta', outs)
     return outs
@@ -95,17 +101,10 @@ class Parser:
   def parse_policy_outputs(self, outs: dict[str, np.ndarray]) -> dict[str, np.ndarray]:
     self.parse_mdn('plan', outs, in_N=ModelConstants.PLAN_MHP_N, out_N=ModelConstants.PLAN_MHP_SELECTION,
                    out_shape=(ModelConstants.IDX_N,ModelConstants.PLAN_WIDTH))
-    self.parse_mdn('lane_lines', outs, in_N=0, out_N=0, out_shape=(ModelConstants.NUM_LANE_LINES,ModelConstants.IDX_N,ModelConstants.LANE_LINES_WIDTH))
-    self.parse_mdn('road_edges', outs, in_N=0, out_N=0, out_shape=(ModelConstants.NUM_ROAD_EDGES,ModelConstants.IDX_N,ModelConstants.LANE_LINES_WIDTH))
-    self.parse_mdn('sim_pose', outs, in_N=0, out_N=0, out_shape=(ModelConstants.POSE_WIDTH,))
-    self.parse_mdn('lead', outs, in_N=ModelConstants.LEAD_MHP_N, out_N=ModelConstants.LEAD_MHP_SELECTION,
-                   out_shape=(ModelConstants.LEAD_TRAJ_LEN,ModelConstants.LEAD_WIDTH))
     if 'lat_planner_solution' in outs:
       self.parse_mdn('lat_planner_solution', outs, in_N=0, out_N=0, out_shape=(ModelConstants.IDX_N,ModelConstants.LAT_PLANNER_SOLUTION_WIDTH))
     if 'desired_curvature' in outs:
       self.parse_mdn('desired_curvature', outs, in_N=0, out_N=0, out_shape=(ModelConstants.DESIRED_CURV_WIDTH,))
-    for k in ['lead_prob', 'lane_lines_prob']:
-      self.parse_binary_crossentropy(k, outs)
     self.parse_categorical_crossentropy('desire_state', outs, out_shape=(ModelConstants.DESIRE_PRED_WIDTH,))
     return outs
 

selfdrive/test/longitudinal_maneuvers/plant.py

@@ -107,6 +107,7 @@ class Plant:
     position = log.XYZTData.new_message()
     position.x = [float(x) for x in (self.speed + 0.5) * np.array(ModelConstants.T_IDXS)]
     model.modelV2.position = position
+    model.modelV2.action.desiredAcceleration = float(self.acceleration + 0.1)
     velocity = log.XYZTData.new_message()
     velocity.x = [float(x) for x in (self.speed + 0.5) * np.ones_like(ModelConstants.T_IDXS)]
     velocity.x[0] = float(self.speed) # always start at current speed

selfdrive/test/process_replay/migration.py

@@ -13,7 +13,7 @@ from opendbc.car.gm.values import GMSafetyFlags
 from openpilot.selfdrive.modeld.constants import ModelConstants
 from openpilot.selfdrive.modeld.fill_model_msg import fill_xyz_poly, fill_lane_line_meta
 from openpilot.selfdrive.test.process_replay.vision_meta import meta_from_encode_index
-from openpilot.selfdrive.controls.lib.longitudinal_planner import get_accel_from_plan
+from openpilot.selfdrive.controls.lib.longitudinal_planner import get_accel_from_plan, CONTROL_N_T_IDX
 from openpilot.system.manager.process_config import managed_processes
 from openpilot.tools.lib.logreader import LogIterable
 
@@ -109,7 +109,7 @@ def migrate_longitudinalPlan(msgs):
     if msg.which() != 'longitudinalPlan':
       continue
     new_msg = msg.as_builder()
-    a_target, should_stop = get_accel_from_plan(msg.longitudinalPlan.speeds, msg.longitudinalPlan.accels)
+    a_target, should_stop = get_accel_from_plan(msg.longitudinalPlan.speeds, msg.longitudinalPlan.accels, CONTROL_N_T_IDX)
     new_msg.longitudinalPlan.aTarget, new_msg.longitudinalPlan.shouldStop = float(a_target), bool(should_stop)
     ops.append((index, new_msg.as_reader()))
   return ops, [], []

selfdrive/test/process_replay/model_replay.py

@@ -64,6 +64,7 @@ def generate_report(proposed, master, tmp, commit):
   ModelV2_Plots = zl([
                      (lambda x: get_idx_if_non_empty(x.velocity.x, 0), "velocity.x"),
                      (lambda x: get_idx_if_non_empty(x.action.desiredCurvature), "desiredCurvature"),
+                     (lambda x: get_idx_if_non_empty(x.action.desiredAcceleration), "desiredAcceleration"),
                      (lambda x: get_idx_if_non_empty(x.leadsV3[0].x, 0), "leadsV3.x"),
                      (lambda x: get_idx_if_non_empty(x.laneLines[1].y, 0), "laneLines.y"),
                      (lambda x: get_idx_if_non_empty(x.meta.desireState, 3), "desireState.laneChangeLeft"),

selfdrive/test/process_replay/ref_commit

@@ -1 +1 @@
-3d3f875dc211ab0ca8a8e564ecf6dd3f52fcf7d9
+fcc771b9ceb487a61035885acbd84e592fb316bf

selfdrive/ui/translations/main_de.ts

@@ -68,23 +68,23 @@
     </message>
     <message>
         <source>Hidden Network</source>
-        <translation type="unfinished"></translation>
+        <translation>Verborgenes Netzwerk</translation>
     </message>
     <message>
         <source>CONNECT</source>
-        <translation type="unfinished">CONNECT</translation>
+        <translation>VERBINDEN</translation>
     </message>
     <message>
         <source>Enter SSID</source>
-        <translation type="unfinished">SSID eingeben</translation>
+        <translation>SSID eingeben</translation>
     </message>
     <message>
         <source>Enter password</source>
-        <translation type="unfinished">Passwort eingeben</translation>
+        <translation>Passwort eingeben</translation>
     </message>
     <message>
         <source>for &quot;%1&quot;</source>
-        <translation type="unfinished">für &quot;%1&quot;</translation>
+        <translation>für &quot;%1&quot;</translation>
     </message>
 </context>
 <context>
@@ -117,31 +117,31 @@
     <name>DeveloperPanel</name>
     <message>
         <source>Joystick Debug Mode</source>
-        <translation type="unfinished"></translation>
+        <translation>Joystick Debug-Modus</translation>
     </message>
     <message>
         <source>Longitudinal Maneuver Mode</source>
-        <translation type="unfinished"></translation>
+        <translation>Längsmanöver-Modus</translation>
     </message>
     <message>
         <source>openpilot Longitudinal Control (Alpha)</source>
-        <translation type="unfinished"></translation>
+        <translation>openpilot Längsregelung (Alpha)</translation>
     </message>
     <message>
         <source>WARNING: openpilot longitudinal control is in alpha for this car and will disable Automatic Emergency Braking (AEB).</source>
-        <translation type="unfinished"></translation>
+        <translation>WARNUNG: Die openpilot Längsregelung befindet sich für dieses Fahrzeug im Alpha-Stadium und deaktiviert das automatische Notbremsen (AEB).</translation>
     </message>
     <message>
         <source>On this car, openpilot defaults to the car&apos;s built-in ACC instead of openpilot&apos;s longitudinal control. Enable this to switch to openpilot longitudinal control. Enabling Experimental mode is recommended when enabling openpilot longitudinal control alpha.</source>
-        <translation type="unfinished"></translation>
+        <translation>Bei diesem Fahrzeug verwendet openpilot standardmäßig den eingebauten Tempomaten anstelle der openpilot Längsregelung. Aktiviere diese Option, um auf die openpilot Längsregelung umzuschalten. Es wird empfohlen, den experimentellen Modus zu aktivieren, wenn die openpilot Längsregelung (Alpha) aktiviert wird.</translation>
     </message>
     <message>
         <source>Enable ADB</source>
-        <translation type="unfinished"></translation>
+        <translation>ADB aktivieren</translation>
     </message>
     <message>
         <source>ADB (Android Debug Bridge) allows connecting to your device over USB or over the network. See https://docs.comma.ai/how-to/connect-to-comma for more info.</source>
-        <translation type="unfinished"></translation>
+        <translation>ADB (Android Debug Bridge) ermöglicht die Verbindung zu deinem Gerät über USB oder Netzwerk. Siehe https://docs.comma.ai/how-to/connect-to-comma für weitere Informationen.</translation>
     </message>
 </context>
 <context>
@@ -280,11 +280,11 @@
     </message>
     <message>
         <source>Pair Device</source>
-        <translation type="unfinished"></translation>
+        <translation>Gerät koppeln</translation>
     </message>
     <message>
         <source>PAIR</source>
-        <translation type="unfinished"></translation>
+        <translation>KOPPELN</translation>
     </message>
 </context>
 <context>
@@ -309,36 +309,38 @@
     <name>FirehosePanel</name>
     <message>
         <source>🔥 Firehose Mode 🔥</source>
-        <translation type="unfinished"></translation>
+        <translation>🔥 Firehose-Modus 🔥</translation>
     </message>
     <message>
         <source>openpilot learns to drive by watching humans, like you, drive.
 
 Firehose Mode allows you to maximize your training data uploads to improve openpilot&apos;s driving models. More data means bigger models, which means better Experimental Mode.</source>
-        <translation type="unfinished"></translation>
+        <translation>openpilot lernt das Fahren, indem es Menschen wie dir beim Fahren zuschaut.
+
+Der Firehose-Modus ermöglicht es dir, deine Trainingsdaten-Uploads zu maximieren, um die Fahrmodelle von openpilot zu verbessern. Mehr Daten bedeuten größere Modelle, was zu einem besseren Experimentellen Modus führt.</translation>
     </message>
     <message>
         <source>Firehose Mode: ACTIVE</source>
-        <translation type="unfinished"></translation>
+        <translation>Firehose-Modus: AKTIV</translation>
     </message>
     <message>
         <source>ACTIVE</source>
-        <translation type="unfinished"></translation>
+        <translation>AKTIV</translation>
     </message>
     <message>
         <source>For maximum effectiveness, bring your device inside and connect to a good USB-C adapter and Wi-Fi weekly.&lt;br&gt;&lt;br&gt;Firehose Mode can also work while you&apos;re driving if connected to a hotspot or unlimited SIM card.&lt;br&gt;&lt;br&gt;&lt;br&gt;&lt;b&gt;Frequently Asked Questions&lt;/b&gt;&lt;br&gt;&lt;br&gt;&lt;i&gt;Does it matter how or where I drive?&lt;/i&gt; Nope, just drive as you normally would.&lt;br&gt;&lt;br&gt;&lt;i&gt;Do all of my segments get pulled in Firehose Mode?&lt;/i&gt; No, we selectively pull a subset of your segments.&lt;br&gt;&lt;br&gt;&lt;i&gt;What&apos;s a good USB-C adapter?&lt;/i&gt; Any fast phone or laptop charger should be fine.&lt;br&gt;&lt;br&gt;&lt;i&gt;Does it matter which software I run?&lt;/i&gt; Yes, only upstream openpilot (and particular forks) are able to be used for training.</source>
-        <translation type="unfinished"></translation>
+        <translation>Für maximale Effektivität bring dein Gerät jede Woche nach drinnen und verbinde es mit einem guten USB-C-Adapter und WLAN.&lt;br&gt;&lt;br&gt;Der Firehose-Modus funktioniert auch während der Fahrt, wenn das Gerät mit einem Hotspot oder einer ungedrosselten SIM-Karte verbunden ist.&lt;br&gt;&lt;br&gt;&lt;br&gt;&lt;b&gt;Häufig gestellte Fragen&lt;/b&gt;&lt;br&gt;&lt;br&gt;&lt;i&gt;Spielt es eine Rolle, wie oder wo ich fahre?&lt;/i&gt; Nein, fahre einfach wie gewohnt.&lt;br&gt;&lt;br&gt;&lt;i&gt;Werden im Firehose-Modus alle meine Segmente hochgeladen?&lt;/i&gt; Nein, wir wählen selektiv nur einen Teil deiner Segmente aus.&lt;br&gt;&lt;br&gt;&lt;i&gt;Welcher USB-C-Adapter ist gut?&lt;/i&gt; Jedes Schnellladegerät für Handy oder Laptop sollte ausreichen.&lt;br&gt;&lt;br&gt;&lt;i&gt;Spielt es eine Rolle, welche Software ich nutze?&lt;/i&gt; Ja, nur das offizielle Upstream‑openpilot (und bestimmte Forks) kann für das Training verwendet werden.</translation>
     </message>
     <message numerus="yes">
         <source>&lt;b&gt;%n segment(s)&lt;/b&gt; of your driving is in the training dataset so far.</source>
-        <translation type="unfinished">
-            <numerusform></numerusform>
-            <numerusform></numerusform>
+        <translation>
+            <numerusform>&lt;b&gt;%n Segment&lt;/b&gt; deiner Fahrten ist bisher im Trainingsdatensatz.</numerusform>
+            <numerusform>&lt;b&gt;%n Segmente&lt;/b&gt; deiner Fahrten sind bisher im Trainingsdatensatz.</numerusform>
         </translation>
     </message>
     <message>
         <source>&lt;span stylesheet=&apos;font-size: 60px; font-weight: bold; color: #e74c3c;&apos;&gt;INACTIVE&lt;/span&gt;: connect to an unmetered network</source>
-        <translation type="unfinished"></translation>
+        <translation>&lt;span stylesheet=&apos;font-size: 60px; font-weight: bold; color: #e74c3c;&apos;&gt;INAKTIV&lt;/span&gt;: Verbinde dich mit einem ungedrosselten Netzwerk</translation>
     </message>
 </context>
 <context>
@@ -411,48 +413,49 @@ Firehose Mode allows you to maximize your training data uploads to improve openp
     <name>OffroadAlert</name>
     <message>
         <source>Immediately connect to the internet to check for updates. If you do not connect to the internet, openpilot won&apos;t engage in %1</source>
-        <translation type="unfinished"></translation>
+        <translation>Stelle sofort eine Internetverbindung her, um nach Updates zu suchen. Wenn du keine Verbindung herstellst, kann openpilot in %1 nicht mehr aktiviert werden.</translation>
     </message>
     <message>
         <source>Connect to internet to check for updates. openpilot won&apos;t automatically start until it connects to internet to check for updates.</source>
-        <translation type="unfinished"></translation>
+        <translation>Verbinde dich mit dem Internet, um nach Updates zu suchen. openpilot startet nicht automatisch, bis eine Internetverbindung besteht und nach Updates gesucht wurde.</translation>
     </message>
     <message>
         <source>Unable to download updates
 %1</source>
-        <translation type="unfinished"></translation>
+        <translation>Updates konnten nicht heruntergeladen werden
+%1</translation>
     </message>
     <message>
         <source>Taking camera snapshots. System won&apos;t start until finished.</source>
-        <translation type="unfinished"></translation>
+        <translation>Kamera-Snapshots werden aufgenommen. Das System startet erst, wenn dies abgeschlossen ist.</translation>
     </message>
     <message>
         <source>An update to your device&apos;s operating system is downloading in the background. You will be prompted to update when it&apos;s ready to install.</source>
-        <translation type="unfinished"></translation>
+        <translation>Ein Update für das Betriebssystem deines Geräts wird im Hintergrund heruntergeladen. Du wirst aufgefordert, das Update zu installieren, sobald es bereit ist.</translation>
     </message>
     <message>
         <source>Device failed to register. It will not connect to or upload to comma.ai servers, and receives no support from comma.ai. If this is an official device, visit https://comma.ai/support.</source>
-        <translation type="unfinished"></translation>
+        <translation>Gerät konnte nicht registriert werden. Es wird keine Verbindung zu den comma.ai-Servern herstellen oder Daten hochladen und erhält keinen Support von comma.ai. Wenn dies ein offizielles Gerät ist, besuche https://comma.ai/support.</translation>
     </message>
     <message>
         <source>NVMe drive not mounted.</source>
-        <translation type="unfinished"></translation>
+        <translation>NVMe-Laufwerk nicht gemounted.</translation>
     </message>
     <message>
         <source>Unsupported NVMe drive detected. Device may draw significantly more power and overheat due to the unsupported NVMe.</source>
-        <translation type="unfinished"></translation>
+        <translation>Nicht unterstütztes NVMe-Laufwerk erkannt. Das Gerät kann dadurch deutlich mehr Strom verbrauchen und überhitzen.</translation>
     </message>
     <message>
         <source>openpilot was unable to identify your car. Your car is either unsupported or its ECUs are not recognized. Please submit a pull request to add the firmware versions to the proper vehicle. Need help? Join discord.comma.ai.</source>
-        <translation type="unfinished"></translation>
+        <translation>openpilot konnte dein Auto nicht identifizieren. Dein Auto wird entweder nicht unterstützt oder die Steuergeräte (ECUs) werden nicht erkannt. Bitte reiche einen Pull Request ein, um die Firmware-Versionen für das richtige Fahrzeug hinzuzufügen. Hilfe findest du auf discord.comma.ai.</translation>
     </message>
     <message>
         <source>openpilot detected a change in the device&apos;s mounting position. Ensure the device is fully seated in the mount and the mount is firmly secured to the windshield.</source>
-        <translation type="unfinished"></translation>
+        <translation>openpilot hat eine Änderung der Montageposition des Geräts erkannt. Stelle sicher, dass das Gerät vollständig in der Halterung sitzt und die Halterung fest an der Windschutzscheibe befestigt ist.</translation>
     </message>
     <message>
         <source>Device temperature too high. System cooling down before starting. Current internal component temperature: %1</source>
-        <translation type="unfinished"></translation>
+        <translation>Gerätetemperatur zu hoch. Das System kühlt ab, bevor es startet. Aktuelle interne Komponententemperatur: %1</translation>
     </message>
 </context>
 <context>
@@ -474,23 +477,23 @@ Firehose Mode allows you to maximize your training data uploads to improve openp
     <name>OnroadAlerts</name>
     <message>
         <source>openpilot Unavailable</source>
-        <translation type="unfinished"></translation>
+        <translation>openpilot nicht verfügbar</translation>
     </message>
     <message>
         <source>TAKE CONTROL IMMEDIATELY</source>
-        <translation type="unfinished"></translation>
+        <translation>ÜBERNIMM SOFORT DIE KONTROLLE</translation>
     </message>
     <message>
         <source>Reboot Device</source>
-        <translation type="unfinished"></translation>
+        <translation>Gerät neu starten</translation>
     </message>
     <message>
         <source>Waiting to start</source>
-        <translation type="unfinished"></translation>
+        <translation>Warten auf Start</translation>
     </message>
     <message>
         <source>System Unresponsive</source>
-        <translation type="unfinished"></translation>
+        <translation>System reagiert nicht</translation>
     </message>
 </context>
 <context>
@@ -513,7 +516,7 @@ Firehose Mode allows you to maximize your training data uploads to improve openp
     </message>
     <message>
         <source>Please connect to Wi-Fi to complete initial pairing</source>
-        <translation type="unfinished"></translation>
+        <translation>Bitte verbinde dich mit WLAN, um die Koppelung abzuschließen.</translation>
     </message>
 </context>
 <context>
@@ -547,15 +550,15 @@ Firehose Mode allows you to maximize your training data uploads to improve openp
     </message>
     <message>
         <source>24/7 LTE connectivity</source>
-        <translation type="unfinished"></translation>
+        <translation>24/7 LTE-Verbindung</translation>
     </message>
     <message>
         <source>1 year of drive storage</source>
-        <translation type="unfinished"></translation>
+        <translation>Fahrdaten-Speicherung für 1 Jahr</translation>
     </message>
     <message>
         <source>Remote snapshots</source>
-        <translation type="unfinished"></translation>
+        <translation>Remote-Snapshots</translation>
     </message>
 </context>
 <context>
@@ -606,7 +609,7 @@ Firehose Mode allows you to maximize your training data uploads to improve openp
     </message>
     <message>
         <source>now</source>
-        <translation type="unfinished"></translation>
+        <translation>jetzt</translation>
     </message>
 </context>
 <context>
@@ -637,16 +640,17 @@ Firehose Mode allows you to maximize your training data uploads to improve openp
     </message>
     <message>
         <source>Unable to mount data partition. Partition may be corrupted. Press confirm to erase and reset your device.</source>
-        <translation type="unfinished"></translation>
+        <translation>Datenpartition konnte nicht gemounted werden. Die Partition ist möglicherweise beschädigt. Drücke Bestätigen, um das Gerät zu löschen und zurückzusetzen.</translation>
     </message>
     <message>
         <source>Resetting device...
 This may take up to a minute.</source>
-        <translation type="unfinished"></translation>
+        <translation>Gerät wird zurückgesetzt...
+Dies kann bis zu einer Minute dauern.</translation>
     </message>
     <message>
         <source>System reset triggered. Press confirm to erase all content and settings. Press cancel to resume boot.</source>
-        <translation type="unfinished"></translation>
+        <translation>System-Reset ausgelöst. Drücke Bestätigen, um alle Inhalte und Einstellungen zu löschen. Drücke Abbrechen, um den Startvorgang fortzusetzen.</translation>
     </message>
 </context>
 <context>
@@ -673,11 +677,11 @@ This may take up to a minute.</source>
     </message>
     <message>
         <source>Developer</source>
-        <translation type="unfinished"></translation>
+        <translation>Entwickler</translation>
     </message>
     <message>
         <source>Firehose</source>
-        <translation type="unfinished"></translation>
+        <translation>Firehose</translation>
     </message>
 </context>
 <context>
@@ -752,11 +756,11 @@ This may take up to a minute.</source>
     </message>
     <message>
         <source>No custom software found at this URL.</source>
-        <translation type="unfinished"></translation>
+        <translation>Keine benutzerdefinierte Software unter dieser URL gefunden.</translation>
     </message>
     <message>
         <source>Something went wrong. Reboot the device.</source>
-        <translation type="unfinished"></translation>
+        <translation>Etwas ist schiefgelaufen. Starte das Gerät neu.</translation>
     </message>
     <message>
         <source>Select a language</source>
@@ -764,15 +768,15 @@ This may take up to a minute.</source>
     </message>
     <message>
         <source>Choose Software to Install</source>
-        <translation type="unfinished"></translation>
+        <translation>Wähle die zu installierende Software</translation>
     </message>
     <message>
         <source>openpilot</source>
-        <translation type="unfinished">openpilot</translation>
+        <translation>openpilot</translation>
     </message>
     <message>
         <source>Custom Software</source>
-        <translation type="unfinished"></translation>
+        <translation>Benutzerdefinierte Software</translation>
     </message>
 </context>
 <context>
@@ -923,23 +927,23 @@ This may take up to a minute.</source>
     </message>
     <message>
         <source>failed to check for update</source>
-        <translation type="unfinished"></translation>
+        <translation>Update-Prüfung fehlgeschlagen</translation>
     </message>
     <message>
         <source>up to date, last checked %1</source>
-        <translation type="unfinished"></translation>
+        <translation>Auf dem neuesten Stand, zuletzt geprüft am %1</translation>
     </message>
     <message>
         <source>DOWNLOAD</source>
-        <translation type="unfinished"></translation>
+        <translation>HERUNTERLADEN</translation>
     </message>
     <message>
         <source>update available</source>
-        <translation type="unfinished"></translation>
+        <translation>Update verfügbar</translation>
     </message>
     <message>
         <source>never</source>
-        <translation type="unfinished"></translation>
+        <translation>nie</translation>
     </message>
 </context>
 <context>
@@ -1000,11 +1004,11 @@ This may take up to a minute.</source>
     </message>
     <message>
         <source>Welcome to openpilot</source>
-        <translation type="unfinished"></translation>
+        <translation>Willkommen bei openpilot</translation>
     </message>
     <message>
         <source>You must accept the Terms and Conditions to use openpilot. Read the latest terms at &lt;span style=&apos;color: #465BEA;&apos;&gt;https://comma.ai/terms&lt;/span&gt; before continuing.</source>
-        <translation type="unfinished"></translation>
+        <translation>Du musst die Nutzungsbedingungen akzeptieren, um openpilot zu verwenden. Lies die aktuellen Bedingungen unter &lt;span style=&apos;color: #465BEA;&apos;&gt;https://comma.ai/terms&lt;/span&gt;, bevor du fortfährst.</translation>
     </message>
 </context>
 <context>
@@ -1071,51 +1075,51 @@ This may take up to a minute.</source>
     </message>
     <message>
         <source>Aggressive</source>
-        <translation type="unfinished"></translation>
+        <translation>Aggressiv</translation>
     </message>
     <message>
         <source>Standard</source>
-        <translation type="unfinished"></translation>
+        <translation>Standard</translation>
     </message>
     <message>
         <source>Relaxed</source>
-        <translation type="unfinished"></translation>
+        <translation>Entspannt</translation>
     </message>
     <message>
         <source>Driving Personality</source>
-        <translation type="unfinished"></translation>
+        <translation>Fahrstil</translation>
     </message>
     <message>
         <source>End-to-End Longitudinal Control</source>
-        <translation type="unfinished"></translation>
+        <translation>Ende-zu-Ende Längsregelung</translation>
     </message>
     <message>
         <source>openpilot longitudinal control may come in a future update.</source>
-        <translation type="unfinished"></translation>
+        <translation>Die openpilot Längsregelung könnte in einem zukünftigen Update verfügbar sein.</translation>
     </message>
     <message>
         <source>An alpha version of openpilot longitudinal control can be tested, along with Experimental mode, on non-release branches.</source>
-        <translation type="unfinished"></translation>
+        <translation>Eine Alpha-Version der openpilot Längsregelung kann zusammen mit dem Experimentellen Modus auf non-stable Branches getestet werden.</translation>
     </message>
     <message>
         <source>Enable the openpilot longitudinal control (alpha) toggle to allow Experimental mode.</source>
-        <translation type="unfinished"></translation>
+        <translation>Aktiviere den Schalter für openpilot Längsregelung (Alpha), um den Experimentellen Modus zu erlauben.</translation>
     </message>
     <message>
         <source>Standard is recommended. In aggressive mode, openpilot will follow lead cars closer and be more aggressive with the gas and brake. In relaxed mode openpilot will stay further away from lead cars. On supported cars, you can cycle through these personalities with your steering wheel distance button.</source>
-        <translation type="unfinished"></translation>
+        <translation>Standard wird empfohlen. Im aggressiven Modus folgt openpilot vorausfahrenden Fahrzeugen enger und ist beim Gasgeben und Bremsen aggressiver. Im entspannten Modus hält openpilot mehr Abstand zu vorausfahrenden Fahrzeugen. Bei unterstützten Fahrzeugen kannst du mit der Abstandstaste am Lenkrad zwischen diesen Fahrstilen wechseln.</translation>
     </message>
     <message>
         <source>The driving visualization will transition to the road-facing wide-angle camera at low speeds to better show some turns. The Experimental mode logo will also be shown in the top right corner.</source>
-        <translation type="unfinished"></translation>
+        <translation>Die Fahrvisualisierung wechselt bei niedrigen Geschwindigkeiten auf die nach vorne gerichtete Weitwinkelkamera, um Kurven besser darzustellen. Das Logo des Experimentellen Modus wird außerdem oben rechts angezeigt.</translation>
     </message>
     <message>
         <source>Always-On Driver Monitoring</source>
-        <translation type="unfinished"></translation>
+        <translation>Dauerhaft aktive Fahrerüberwachung</translation>
     </message>
     <message>
         <source>Enable driver monitoring even when openpilot is not engaged.</source>
-        <translation type="unfinished"></translation>
+        <translation>Fahrerüberwachung auch aktivieren, wenn openpilot nicht aktiv ist.</translation>
     </message>
 </context>
 <context>
@@ -1157,15 +1161,15 @@ This may take up to a minute.</source>
     <name>WiFiPromptWidget</name>
     <message>
         <source>Open</source>
-        <translation type="unfinished"></translation>
+        <translation>Öffnen</translation>
     </message>
     <message>
         <source>Maximize your training data uploads to improve openpilot&apos;s driving models.</source>
-        <translation type="unfinished"></translation>
+        <translation>Maximiere deine Trainingsdaten-Uploads, um die Fahrmodelle von openpilot zu verbessern.</translation>
     </message>
     <message>
         <source>&lt;span style=&apos;font-family: &quot;Noto Color Emoji&quot;;&apos;&gt;🔥&lt;/span&gt; Firehose Mode &lt;span style=&apos;font-family: Noto Color Emoji;&apos;&gt;🔥&lt;/span&gt;</source>
-        <translation type="unfinished"></translation>
+        <translation>&lt;span style=&apos;font-family: &quot;Noto Color Emoji&quot;;&apos;&gt;🔥&lt;/span&gt; Firehose-Modus &lt;span style=&apos;font-family: Noto Color Emoji;&apos;&gt;🔥&lt;/span&gt;</translation>
     </message>
 </context>
 <context>

system/ui/lib/application.py

@@ -50,9 +50,11 @@ class GuiApplication:
     self._set_styles()
     self._load_fonts()
 
-  def load_texture_from_image(self, file_name: str, width: int, height: int):
+  def load_texture_from_image(self, file_name: str, width: int, height: int, alpha_premultiply = False):
     """Load and resize a texture, storing it for later automatic unloading."""
     image = rl.load_image(file_name)
+    if alpha_premultiply:
+      rl.image_alpha_premultiply(image)
     rl.image_resize(image, width, height)
     texture = rl.load_texture_from_image(image)
     # Set texture filtering to smooth the result

system/ui/spinner.py

@@ -11,12 +11,11 @@ from openpilot.system.ui.text import wrap_text
 # Constants
 PROGRESS_BAR_WIDTH = 1000
 PROGRESS_BAR_HEIGHT = 20
-ROTATION_TIME_SECONDS = 1.0  # Time for one full circle
-MARGIN_H = 50
-MARGIN_V = 200
+DEGREES_PER_SECOND = 360.0  # one full rotation per second
+MARGIN_H = 100
 TEXTURE_SIZE = 360
-FONT_SIZE = 96
-LINE_HEIGHT = 104
+FONT_SIZE = 88
+LINE_HEIGHT = 96
 DARKGRAY = (55, 55, 55, 255)
 
 
@@ -27,10 +26,11 @@ def clamp(value, min_value, max_value):
 class SpinnerRenderer:
   def __init__(self):
     self._comma_texture = gui_app.load_texture_from_image(os.path.join(BASEDIR, "selfdrive/assets/img_spinner_comma.png"), TEXTURE_SIZE, TEXTURE_SIZE)
-    self._spinner_texture = gui_app.load_texture_from_image(os.path.join(BASEDIR, "selfdrive/assets/img_spinner_track.png"), TEXTURE_SIZE, TEXTURE_SIZE)
+    self._spinner_texture = gui_app.load_texture_from_image(os.path.join(BASEDIR, "selfdrive/assets/img_spinner_track.png"), TEXTURE_SIZE, TEXTURE_SIZE,
+                                                            alpha_premultiply=True)
     self._rotation = 0.0
-    self._progress: int | None = 0
-    self._wrapped_lines = []
+    self._progress: int | None = None
+    self._wrapped_lines: list[str] = []
     self._lock = threading.Lock()
 
   def set_text(self, text: str) -> None:
@@ -40,36 +40,30 @@ class SpinnerRenderer:
         self._wrapped_lines = []
       else:
         self._progress = None
-        self._wrapped_lines = wrap_text(text, FONT_SIZE, gui_app.width - MARGIN_H * 2)
+        self._wrapped_lines = wrap_text(text, FONT_SIZE, gui_app.width - MARGIN_H)
 
   def render(self):
-    center = rl.Vector2(gui_app.width / 2.0, gui_app.height / 2.0)
-
-    # Display progress bar or text based on user input
     with self._lock:
       progress = self._progress
-      wrapped_lines = self._wrapped_lines.copy()
-
-    if progress is not None:
-      y_pos = rl.get_screen_height() - MARGIN_V - PROGRESS_BAR_HEIGHT
-      bar = rl.Rectangle(center.x - PROGRESS_BAR_WIDTH / 2.0, y_pos, PROGRESS_BAR_WIDTH, PROGRESS_BAR_HEIGHT)
-      rl.draw_rectangle_rounded(bar, 1, 10, DARKGRAY)
+      wrapped_lines = self._wrapped_lines
 
-      bar.width *= progress / 100.0
-      rl.draw_rectangle_rounded(bar, 1, 10, rl.WHITE)
+    if wrapped_lines:
+      # Calculate total height required for spinner and text
+      spacing = 50
+      total_height = TEXTURE_SIZE + spacing + len(wrapped_lines) * LINE_HEIGHT
+      center_y = (gui_app.height - total_height) / 2.0 + TEXTURE_SIZE / 2.0
     else:
-      center.y -= MARGIN_V
-      y_pos = center.y + TEXTURE_SIZE / 2 + MARGIN_V / 2
-      for i, line in enumerate(wrapped_lines):
-        text_size = rl.measure_text_ex(gui_app.font(), line, FONT_SIZE, 1.0)
-        rl.draw_text_ex(gui_app.font(), line, rl.Vector2(center.x - text_size.x / 2, y_pos + i * LINE_HEIGHT),
-                        FONT_SIZE, 1.0, rl.WHITE)
+      # Center spinner vertically
+      spacing = 150
+      center_y = gui_app.height / 2.0
+    y_pos = center_y + TEXTURE_SIZE / 2.0 + spacing
 
-    # Draw rotating spinner and static comma logo
-    fps = rl.get_fps()
-    if fps > 0:
-      degrees_per_frame = 360.0 / (ROTATION_TIME_SECONDS * fps)
-      self._rotation = (self._rotation + degrees_per_frame) % 360.0
+    center = rl.Vector2(gui_app.width / 2.0, center_y)
+    spinner_origin = rl.Vector2(TEXTURE_SIZE / 2.0, TEXTURE_SIZE / 2.0)
+    comma_position = rl.Vector2(center.x - TEXTURE_SIZE / 2.0, center.y - TEXTURE_SIZE / 2.0)
+
+    delta_time = rl.get_frame_time()
+    self._rotation = (self._rotation + DEGREES_PER_SECOND * delta_time) % 360.0
 
     spinner_origin = rl.Vector2(TEXTURE_SIZE / 2.0, TEXTURE_SIZE / 2.0)
     comma_position = rl.Vector2(center.x - TEXTURE_SIZE / 2.0, center.y - TEXTURE_SIZE / 2.0)
@@ -79,6 +73,19 @@ class SpinnerRenderer:
                         spinner_origin, self._rotation, rl.WHITE)
     rl.draw_texture_v(self._comma_texture, comma_position, rl.WHITE)
 
+    # Display progress bar or text based on user input
+    if progress is not None:
+      bar = rl.Rectangle(center.x - PROGRESS_BAR_WIDTH / 2.0, y_pos, PROGRESS_BAR_WIDTH, PROGRESS_BAR_HEIGHT)
+      rl.draw_rectangle_rounded(bar, 1, 10, DARKGRAY)
+
+      bar.width *= progress / 100.0
+      rl.draw_rectangle_rounded(bar, 1, 10, rl.WHITE)
+    elif wrapped_lines:
+      for i, line in enumerate(wrapped_lines):
+        text_size = rl.measure_text_ex(gui_app.font(), line, FONT_SIZE, 0.0)
+        rl.draw_text_ex(gui_app.font(), line, rl.Vector2(center.x - text_size.x / 2, y_pos + i * LINE_HEIGHT),
+                        FONT_SIZE, 0.0, rl.WHITE)
+
 
 class Spinner(Wrapper):
   def __init__(self):

system/ui/text.py

@@ -21,13 +21,12 @@ def wrap_text(text, font_size, max_width):
   font = gui_app.font()
 
   for paragraph in text.split("\n"):
-    indent = re.match(r'^\s*', paragraph).group()
-    content = paragraph[len(indent):]
-
-    if content == "":
-      lines.append("")
+    if not paragraph.strip():
+      # Don't add empty lines first, ensuring wrap_text("") returns []
+      if lines:
+        lines.append("")
       continue
-
+    indent = re.match(r"^\s*", paragraph).group()
     current_line = indent
     for word in paragraph.split():
       test_line = current_line + word + " "
@@ -35,8 +34,8 @@ def wrap_text(text, font_size, max_width):
         current_line = test_line
       else:
         lines.append(current_line)
-        current_line = indent + word + " "
-
+        current_line = word + " "
+    current_line = current_line.rstrip()
     if current_line:
       lines.append(current_line)
 
@@ -56,7 +55,7 @@ class TextWindowRenderer:
       position = rl.Vector2(self._textarea_rect.x + scroll.x, self._textarea_rect.y + scroll.y + i * LINE_HEIGHT)
       if position.y + LINE_HEIGHT < self._textarea_rect.y or position.y > self._textarea_rect.y + self._textarea_rect.height:
         continue
-      rl.draw_text_ex(gui_app.font(), line.rstrip(), position, FONT_SIZE, 0, rl.WHITE)
+      rl.draw_text_ex(gui_app.font(), line, position, FONT_SIZE, 0, rl.WHITE)
     rl.end_scissor_mode()
 
     button_bounds = rl.Rectangle(gui_app.width - MARGIN - BUTTON_SIZE.x, gui_app.height - MARGIN - BUTTON_SIZE.y, BUTTON_SIZE.x, BUTTON_SIZE.y)

system/ui/updater.py

@@ -0,0 +1,197 @@
+#!/usr/bin/env python3
+import sys
+import subprocess
+import threading
+import pyray as rl
+from enum import IntEnum
+
+from openpilot.system.hardware import HARDWARE
+from openpilot.system.ui.lib.application import gui_app, FontWeight
+from openpilot.system.ui.lib.button import gui_button, ButtonStyle
+from openpilot.system.ui.lib.label import gui_text_box, gui_label
+
+# Constants
+MARGIN = 50
+BUTTON_HEIGHT = 160
+BUTTON_WIDTH = 400
+PROGRESS_BAR_HEIGHT = 72
+TITLE_FONT_SIZE = 80
+BODY_FONT_SIZE = 65
+BACKGROUND_COLOR = rl.BLACK
+PROGRESS_BG_COLOR = rl.Color(41, 41, 41, 255)
+PROGRESS_COLOR = rl.Color(54, 77, 239, 255)
+
+
+class Screen(IntEnum):
+  PROMPT = 0
+  WIFI = 1
+  PROGRESS = 2
+
+
+class Updater:
+  def __init__(self, updater_path, manifest_path):
+    self.updater = updater_path
+    self.manifest = manifest_path
+    self.current_screen = Screen.PROMPT
+
+    self.progress_value = 0
+    self.progress_text = "Loading..."
+    self.show_reboot_button = False
+    self.process = None
+    self.update_thread = None
+
+  def install_update(self):
+    self.current_screen = Screen.PROGRESS
+    self.progress_value = 0
+    self.progress_text = "Downloading..."
+    self.show_reboot_button = False
+
+    # Start the update process in a separate thread
+    self.update_thread = threading.Thread(target=self._run_update_process)
+    self.update_thread.daemon = True
+    self.update_thread.start()
+
+  def _run_update_process(self):
+    # TODO: just import it and run in a thread without a subprocess
+    cmd = [self.updater, "--swap", self.manifest]
+    self.process = subprocess.Popen(cmd, stdout=subprocess.PIPE, stderr=subprocess.PIPE,
+                                   text=True, bufsize=1, universal_newlines=True)
+
+    for line in self.process.stdout:
+      parts = line.strip().split(":")
+      if len(parts) == 2:
+        self.progress_text = parts[0]
+        try:
+          self.progress_value = int(float(parts[1]))
+        except ValueError:
+          pass
+
+    exit_code = self.process.wait()
+    if exit_code == 0:
+      HARDWARE.reboot()
+    else:
+      self.progress_text = "Update failed"
+      self.show_reboot_button = True
+
+  def render_prompt_screen(self):
+    # Title
+    title_rect = rl.Rectangle(MARGIN + 50, 250, gui_app.width - MARGIN * 2 - 100, TITLE_FONT_SIZE)
+    gui_label(title_rect, "Update Required", TITLE_FONT_SIZE, font_weight=FontWeight.BOLD)
+
+    # Description
+    desc_text = "An operating system update is required. Connect your device to Wi-Fi for the fastest update experience. \
+                 The download size is approximately 1GB."
+    desc_rect = rl.Rectangle(MARGIN + 50, 250 + TITLE_FONT_SIZE + 75, gui_app.width - MARGIN * 2 - 100, BODY_FONT_SIZE * 3)
+    gui_text_box(desc_rect, desc_text, BODY_FONT_SIZE)
+
+    # Buttons at the bottom
+    button_y = gui_app.height - MARGIN - BUTTON_HEIGHT
+    button_width = (gui_app.width - MARGIN * 3) // 2
+
+    # WiFi button
+    wifi_button_rect = rl.Rectangle(MARGIN, button_y, button_width, BUTTON_HEIGHT)
+    if gui_button(wifi_button_rect, "Connect to Wi-Fi"):
+      self.current_screen = Screen.WIFI
+      return  # Return to avoid processing other buttons after screen change
+
+    # Install button
+    install_button_rect = rl.Rectangle(MARGIN * 2 + button_width, button_y, button_width, BUTTON_HEIGHT)
+    if gui_button(install_button_rect, "Install", button_style=ButtonStyle.PRIMARY):
+      self.install_update()
+      return  # Return to avoid further processing after action
+
+  def render_wifi_screen(self):
+    # Title and back button
+    title_rect = rl.Rectangle(MARGIN + 50, MARGIN, gui_app.width - MARGIN * 2 - 100, 60)
+    gui_label(title_rect, "Wi-Fi Networks", 60, font_weight=FontWeight.BOLD)
+
+    back_button_rect = rl.Rectangle(MARGIN, gui_app.height - MARGIN - BUTTON_HEIGHT, BUTTON_WIDTH, BUTTON_HEIGHT)
+    if gui_button(back_button_rect, "Back"):
+      self.current_screen = Screen.PROMPT
+      return  # Return to avoid processing other interactions after screen change
+
+    # Draw placeholder for WiFi implementation
+    placeholder_rect = rl.Rectangle(
+      MARGIN,
+      title_rect.y + title_rect.height + MARGIN,
+      gui_app.width - MARGIN * 2,
+      gui_app.height - title_rect.height - MARGIN * 3 - BUTTON_HEIGHT
+    )
+
+    # Draw rounded rectangle background
+    rl.draw_rectangle_rounded(
+      placeholder_rect,
+      0.1,
+      10,
+      rl.Color(41, 41, 41, 255)
+    )
+
+    # Draw placeholder text
+    placeholder_text = "WiFi Implementation Placeholder"
+    text_size = rl.measure_text_ex(gui_app.font(), placeholder_text, 80, 1)
+    text_pos = rl.Vector2(
+      placeholder_rect.x + (placeholder_rect.width - text_size.x) / 2,
+      placeholder_rect.y + (placeholder_rect.height - text_size.y) / 2
+    )
+    rl.draw_text_ex(gui_app.font(), placeholder_text, text_pos, 80, 1, rl.WHITE)
+
+    # Draw instructions
+    instructions_text = "Real WiFi functionality would be implemented here"
+    instructions_size = rl.measure_text_ex(gui_app.font(), instructions_text, 40, 1)
+    instructions_pos = rl.Vector2(
+      placeholder_rect.x + (placeholder_rect.width - instructions_size.x) / 2,
+      text_pos.y + text_size.y + 20
+    )
+    rl.draw_text_ex(gui_app.font(), instructions_text, instructions_pos, 40, 1, rl.GRAY)
+
+  def render_progress_screen(self):
+    title_rect = rl.Rectangle(MARGIN + 100, 330, gui_app.width - MARGIN * 2 - 200, 100)
+    gui_label(title_rect, self.progress_text, 90, font_weight=FontWeight.SEMI_BOLD)
+
+    # Progress bar
+    bar_rect = rl.Rectangle(MARGIN + 100, 330 + 100 + 100, gui_app.width - MARGIN * 2 - 200, PROGRESS_BAR_HEIGHT)
+    rl.draw_rectangle_rounded(bar_rect, 0.5, 10, PROGRESS_BG_COLOR)
+
+    # Calculate the width of the progress chunk
+    progress_width = (bar_rect.width * self.progress_value) / 100
+    if progress_width > 0:
+      progress_rect = rl.Rectangle(bar_rect.x, bar_rect.y, progress_width, bar_rect.height)
+      rl.draw_rectangle_rounded(progress_rect, 0.5, 10, PROGRESS_COLOR)
+
+    # Show reboot button if needed
+    if self.show_reboot_button:
+      reboot_rect = rl.Rectangle(MARGIN + 100, gui_app.height - MARGIN - BUTTON_HEIGHT, BUTTON_WIDTH, BUTTON_HEIGHT)
+      if gui_button(reboot_rect, "Reboot"):
+        # Return True to signal main loop to exit before rebooting
+        HARDWARE.reboot()
+        return
+
+  def render(self):
+    if self.current_screen == Screen.PROMPT:
+      self.render_prompt_screen()
+    elif self.current_screen == Screen.WIFI:
+      self.render_wifi_screen()
+    elif self.current_screen == Screen.PROGRESS:
+      self.render_progress_screen()
+
+
+def main():
+  if len(sys.argv) < 3:
+    print("Usage: updater.py <updater_path> <manifest_path>")
+    sys.exit(1)
+
+  updater_path = sys.argv[1]
+  manifest_path = sys.argv[2]
+
+  try:
+    gui_app.init_window("System Update")
+    updater = Updater(updater_path, manifest_path)
+    for _ in gui_app.render():
+      updater.render()
+  finally:
+    # Make sure we clean up even if there's an error
+    gui_app.close()
+
+
+if __name__ == "__main__":
+  main()