How Imitation Learning and Human Factors Can Be Combined in a Model Predictive Control Algorithm for Adaptive Motion...

Area of Science:

• Intelligent Transportation Systems (ITSs)
• Robotics and Control Systems
• Artificial Intelligence (AI)

Background:

• Autonomous vehicles (AVs) require sophisticated motion-planning and control for safe and comfortable deployment.
• Existing intelligent transportation systems (ITSs) face challenges in achieving human-like driving behavior for AVs.
• Technological hurdles remain for widespread AV adoption in diverse road scenarios.

Purpose of the Study:

• To develop an effective trajectory-planning and control algorithm for AVs.
• To ensure occupant comfort and safety through advanced motion control.
• To enable human-like driving behavior in AVs via imitation learning.

Main Methods:

• Model Predictive Control (MPC) approach for trajectory planning and vehicle dynamics control.
• Imitation learning from real-world overtaking maneuver data.
• Simulations and Hardware-In-the-Loop (HIL) testing for validation.

Main Results:

• The proposed MPC algorithm effectively generates optimal trajectories for AVs.
• The system demonstrates human-like driving behavior, particularly in overtaking maneuvers.
• Validation through simulations and HIL tests confirms the approach's effectiveness and computational efficiency.

Conclusions:

• The developed MPC-based algorithm is a key enabler for advanced AV motion planning and control.
• The integration of imitation learning allows for more natural and human-like AV operation.
• The approach shows significant promise for the future deployment of autonomous vehicles in various environments.