Autonomous Vehicle Control Through Socially Compliant Human–Robot Interactions With an Application to Eco-Driving

Abstract

The control design of autonomous vehicles (AVs) has mostly focused on achieving specific goals for either individually controlled AVs or cooperatively controlled swarms. However, the impact of autonomous driving on human-driven vehicles (HVs) has often been overlooked in AV controller synthesis, potentially resulting in egoistic AV behavior detrimental to traffic flow. To address this gap, we introduce a framework for socially compliant AV control design, leveraging a psychological metric called social value orientation (SVO). This framework allows AVs to consider their impact on following HVs, enabling socially compliant AV behavior that human drivers can understand and appropriately respond to. We define appropriate utilities for the controlled AV and its following HV, to be maximized in a weighted manner determined by the AV’s SVO. This utility maximization can accommodate various design objectives based on the AV’s goals and the benefits to the following HVs from socially compliant AV controls. An optimal control problem is then formulated and solved using Pontryagin’s minimum principle to maximize the utility function. The developed approach is applied to synthesize socially compliant eco-driving controls for an AV. Leveraging real-world vehicle trajectory data gathered on Highway 55 in Minnesota, our simulation results from a five-vehicle platoon show that altruistic AV driving can improve the average speed of its following HVs by over 5%. Notably, improvements exceeding 45% are achieved in the vicinity of signalized intersections, albeit with a slight reduction in the AV’s payoff.