Natural Speed Planning for Connected Car Using Slack Variable within Social Norm

초록

In recent years, studies on natural behavior planning for autonomous vehicles have been popular in both the academia and the industry. This study proposes receding-horizon control, framework-based vehicle speed planning. Using a sufficiently long prediction horizon, our approach is aimed at a smoothed velocity trajectory, reducing the entire trip time by passing the traffic light properly. Moreover, the proposed method mimics a human driver naturally in terms of driving styles (i.e., conservative, general, and aggressive behaviors), In the view of the social norm, the planned speed from our approach sometimes exceeds the specified speed limit, but it does not violate strict traffic rules. In this manner, the controlled, connected, and automated vehicle behaves as human drivers do in their daily driving. Specifically, we impose a slack variable to the proposed control framework when upcoming traffic information is available, so that the vehicle can pass the traffic signal at the yellow light phase. By relaxing state constraints, the test results in this paper show a reduction in trip time and fuel consumption. © © 2023 KSAE.

키워드