Adaptive Cruise Control with Model Predictive Control
This page presents the results of the MPC-based Adaptive Cruise Control simulation. The project code and details are available on GitHub.
Performance Metrics
| Metric | Value |
|---|
How to Reproduce
- Clone the GitHub repository.
- Follow the setup instructions in
ACC_w_mpc_py/README.md. - Run
python ACC_w_mpc_py/simulation.pyto generate new plots and themetrics.jsonfile.
About
This project demonstrates the use of Model Predictive Control for Adaptive Cruise Control, simulating different driver styles and analyzing performance metrics.
MPC Formulation
The Model Predictive Control (MPC) formulation for Adaptive Cruise Control optimizes the following cost function:
Cost Function
J = Σk=0N-1 [qv(vk - vref)² + qd(dk - dref)² + qclosemax(0, dsafe - dk)² + qfarmax(0, dk - dmax)² + raak² + rjjk²]
Where:
- vk: Current velocity at time step k
- vref: Reference/desired velocity (15 m/s)
- dk: Current distance to lead vehicle
- dref: Desired following distance (time gap × velocity)
- ak: Acceleration control input
- jk: Jerk (rate of change of acceleration)
- qv, qd, qclose, qfar: State weighting factors
- ra, rj: Control input weighting factors
Driver Style Comparison
Different driver styles are implemented by varying the MPC weights and time gap settings:
| Parameter | Aggressive | Balanced | Conservative |
|---|---|---|---|
| Time Gap | 1.0s (Short) | 2.0s (Medium) | 3.0s (Long) |
| Speed Factor | 1.2x (20% faster) | 1.0x (Default) | 1.0x (Default) |
| qvelocity | 2.0 (High priority) | 1.0 (Balanced) | 0.5 (Low priority) |
| qdistance | 1.0 (Low priority) | 2.0 (Balanced) | 3.0 (High priority) |
| qclose | 5.0 (Moderate penalty) | 10.0 (High penalty) | 20.0 (Very high penalty) |
| qfar | 0.5 (Low penalty) | 1.0 (Balanced) | 2.0 (High penalty) |
| racceleration | 0.05 (Allow aggressive) | 0.1 (Moderate) | 0.2 (Conservative) |
| rjerk | 0.05 (Allow aggressive) | 0.1 (Moderate) | 0.2 (Conservative) |
| Behavior | Closer following, faster acceleration, higher speeds | Moderate following distance, balanced performance | Safe following distance, smooth driving, comfort-focused |
Key Differences:
- Aggressive: Prioritizes speed tracking over distance maintenance, allows more aggressive acceleration and jerk, maintains shorter time gaps
- Balanced: Equal weighting between speed and distance objectives, moderate penalties for safety violations
- Conservative: Prioritizes safety and comfort over speed, maintains longer time gaps, penalizes aggressive control inputs heavily