Driving Authority Transfer Strategy for Shared Control Vehicles Based on Risk Assessment and Take-over Intention Recognition

Driving mode switching is common in autonomous vehicles below L3. Human-machine shared control is a crucial strategy for safely transitioning from autonomous to manual driving. How to identify driver's take-over intention accurately and transfer driving authority smoothly is an intractable problem. To address this issue, this paper proposes an authority transfer strategy based on risk assessment and driver take-over intention recognition. First, the K-means clustering algorithm is used to process driver feature data. A take-over intention identification scheme based on torque threshold is proposed. Lane risk assessment aids in recognizing take-over intention to prevent driver's wrong touch. Then, upon driver's take-over intention is detected, authorities are smoothly allocated to the driver according to a flexible transfer strategy considering driving characteristics. Finally, the proposed intention recognition and authority transfer strategy are validated through a driver-in-the-loop test bench. The results show that, this strategy can accurately identify driver's take-over intention, safely avoiding obstacles and improving vehicle's lateral stability during the authority transfer process.