Goals:
- Step 1: Keep lane. Control the steering wheel position based on camera feed.
- Step 2: Adaptive cruise control. Utilize vehicle’s existing cruise control to control the vehicle’s speed, while keeping a safe distance to the car ahead.
- Step 3: Speed limit detection. Extend adaptive cruise control by detecting speed limit signs.
- Step 4: Consider turn signals to change lanes. An additional input to the NN of Step 1 animates it to change lanes and take turns instead of just following the current lane.
- Step 5: Navigate with maps. Use Google maps routing as a source for turn signals.
Step 1:
Usage:
- Enable system by long-pressing RES/+ and Phone-Call-Button simultaneously
- Disable by pressing Disable-Cruise-Control-Button, the brake pedal, the clutch pedal
- Disable when steering wheel angle controller notices a persisting deviation between setpoint and actual angle (detecting a forceful manual override)
System inputs:
- RGB image (Camera)
- [Depth image (Camera)]
- Actual steering wheel angle (Vehicle LS-CAN, Data Collection)
- Actual vehicle speed (Vehicle LS-CAN, Data Collection)
- Steering wheel button presses (Vehicle LS-CAN, Engage)
- Brake pedal pressed (Vehicle LS-CAN, Disengage)
- Clutch pedal pressed (Vehicle LS-CAN, Disengage)
System outputs:
- Engange/Disengage actuator
- Steering wheel actuator: rotation speed and direction
Neural Network
Inputs:
- RGB image
- (Depth image)
- (Actual vehicle speed)
Outputs:
- Desired steering wheel angle
Steering wheel angle controller
Inputs:
- System on/off
- Actual steering wheel angle
- Desired steering wheel angle
- Actual vehicle speed (To limit actuator speed)
Outputs:
- Engange/Disengage actuator
- Steering wheel actuator, rotation speed and direction