| Safety in transportation systems is a global concern with millions of road accidents happening worldwide every year. Despite sophisticated advancement in automotive design and accident severity reduction technologies, traffic fatalities continue to persist and the overwhelming majority has driver behavior as a causative factor. Human drivers are susceptible to distraction, fatigue, substance abuse and inherent limitations in prediction and reaction capabilities. This explains the longstanding interest among the industry and government experts for the development of driver assistance systems, resulting in semi-autonomous vehicles. But further benefits can be reaped if fully autonomous vehicles are used, provided that the overall safety is ensured. A network of cooperative, fully autonomous vehicles can not only decrease the accident rates, but also help in solving other traffic problems such as time and space inefficiency and waste of fuel due to congestion.;We design a framework for cooperative autonomous vehicles that focuses on safely sharing existing roads with human-driven vehicles. Our design approach is inspired by the flocking behavior of birds; one of the primary goals of a driver is collision avoidance and that happens to be a byproduct of the flocking behavior. This project enables cyber-physical vehicles to achieve all basic driving tasks such as lane-driving, braking, and turning using distributed control algorithms. We also model human driver behavior so that the self-driving vehicles can take their surrounding human behavioral factors into account. These hybrid control algorithms create a network of autonomous vehicles with nonlinear switching dynamics. We provide experimental results that illustrate the stability, effectiveness, and safety-awareness of our cooperative autonomous driving algorithms. |
