Fast Safe Mission Plans for Autonomous Vehicles

摘要

Guaranteeing safety is a key problem that needs to be addressed in order to enable the real-world deployment of robots and autonomous cyber-physical systems (CPS). While there is a lot of interest in deploying sensors and predictors that would identify obstacles and unsafe situations, there is little research on how to use such learned systems to plan and execute missions safely and efficiently. Recent research on safe planning and control not only admits to simplistic constraints, most of them assumed to be known a priori, but attempting such synthesis often results in large optimization problems which are often impractical to solve given real time constraints of such systems. In this work we propose a novel combination of sampling-based motion planning with safe control synthesis methods for generating safe high-level plans in real-time. The distinguishing aspect of our work is that it provides a natural framework of incorporating sensor data and the associated prediction about the obstacles to quickly determine the safe mission plan. We showcase this approach with autonomous car scenarios.