An Optimal Control Framework for Influencing Human Driving Behavior in Mixed-Autonomy Traffic

TL;DR

The proposed framework for achieving human influence is versatility in a wide spectrum of influence objectives and mixed-autonomy configurations, and is validated when applied to the problems of traffic flow optimization and aggressive behavior mitigation.

Abstract

As autonomous vehicles (AVs) become increasingly prevalent, their interaction with human drivers presents a critical challenge. Current AVs lack social awareness, causing behavior that is often awkward or unsafe. To combat this, social AVs, which are proactive rather than reactive in their behavior, have been explored in recent years. With knowledge of robot-human interaction dynamics, a social AV can influence a human driver to exhibit desired behaviors by strategically altering its own behaviors. In this paper, we present a novel framework for achieving human influence. The foundation of our framework lies in an innovative use of control barrier functions to formulate the desired objectives of influence as constraints in an optimal control problem. The computed controls gradually push the system state toward satisfaction of the objectives, e.g. slowing the human down to some desired speed. We demonstrate the proposed framework's feasibility in a variety of scenarios related to car-following and lane changes, including multi-robot and multi-human configurations. In two case studies, we validate the framework's effectiveness when applied to the problems of traffic flow optimization and aggressive behavior mitigation. Given these results, the main contribution of our framework is its versatility in a wide spectrum of influence objectives and mixed-autonomy configurations.

Figures (8)

• Figure 1: A robot car is being tailgated by a human-driven car. How can the robot influence the human to increase following distance or to change lanes?
• Figure 2: Illustrations, human-relative position vs. time graphs, and velocity vs. time graphs for scenarios S1-S3 and SM1-SM3. In illustrations, red boxes are robot cars, blue boxes are human-driven cars, and green boxes are background cars. In graphs, black dotted horizontal lines denote lower or upper bounds on the value via the influence objective, and gray dotted vertical lines denote the time at which a lane change occurs.
• Figure 3: Illustrations, human-relative position vs. time graphs, and velocity vs. time graphs for scenarios M1-M3. In illustrations, red boxes are robot cars, and blue boxes are human-driven cars. In graphs, black dotted horizontal lines denote lower or upper bounds on the value via the influence objective, and gray dotted vertical lines denote the time at which a lane change occurs.
• Figure 4: Visualization of scenario M1, where two adjacent lanes each contain a robot car followed by a human-driven car, and the objective is to influence the human in the right lane to merge in between the two cars in the left lane. Red boxes are robot cars and blue boxes are human-driven cars. Overlaying each box is the car's velocity.
• Figure 5: Average velocity vs. time for traffic flow optimization.

Theorems & Definitions (1)

• Definition 1: Direct influence