Safe Motion Planning for Autonomous Driving
نویسنده
چکیده
Self-driving cars have the potential to revolutionize transportation by making it cheaper, safer, and more efficient. In this thesis we describe a novel motion planning system, which translates high-level navigation goals into low-level actions for controlling a vehicle. Specifically, the motion planning system is responsible for choosing at each time step an appropriate velocity and steering angle, which can then be implemented by the driving hardware or simulator. Our planner is able to compute a safe and efficient trajectory in a dynamic environment while staying within its lane and avoiding obstacles. The planner works as follows: given a road map represented as a graph, an incremental heuristic search method is used to generate an optimal path. A section of this path nearest to the agent is smoothed using spline techniques to generate an arclength parameterized reference path for the agent to follow. A two-level hierarchical state space is produced that encompasses the various possible choices of steering angle. Using a vehicle motion model, the future positions of the car are determined for each action. Those actions that lead to invalid states (those that bring the agent in contact with an obstacle or out of its lane) are ruled out, and the action cost for each state is computed according to its path-following behavior. Finally, the action that leads to the lowest cost state is chosen and executed. The planner was implemented in simulation and tested in various driving scenarios with results comparable or better to those produced by other motion planning systems in terms of speed, safety, path-following behavior and comfort.
منابع مشابه
Toward Safe, Feasible and Human-Like Motion Generation for Urban Autonomous Driving
Motion prediction and motion planning problems are deeply interconnected for urban autonomous driving. In this project, we target the problem of safe, feasible and human-like motion generation. A non-conservatively defensive strategy (NCDS) is proposed, which is defensive under the worst case to guarantee safety, but not conservative so as to degrade driving quality and generate behaviors which...
متن کاملToward Safe and Human-like Motion Generation for Autonomous Driving
1. Project Overview In order to drive safely and efficiently, autonomous vehicles need to understand the behavior of others, predict their future motions and plan desirable motions which are safe and human-like to execute. In this project, we tackled motion prediction and planning together as a motion generation problem to achieve accurate comprehension of others, as well as safe and human-like...
متن کاملOnline Speed Adaptation Using Supervised Learning for High-Speed, Off-Road Autonomous Driving
The mobile robotics community has traditionally addressed motion planning and navigation in terms of steering decisions. However, selecting the best speed is also important – beyond its relationship to stopping distance and lateral maneuverability. Consider a high-speed (35 mph) autonomous vehicle driving off-road through challenging desert terrain. The vehicle should drive slowly on terrain th...
متن کاملOn-Road Trajectory Planning for General Autonomous Driving with Enhanced Tunability
In order to achieve smooth autonomous driving in real-life urban and highway environments, a motion planner must generate trajectories that are locally smooth and responsive (reactive), and at the same time, far-sighted and intelligent (deliberative). Prior approaches achieved both planning qualities for full-speed-range operations at a high computational cost. Moreover, the planning formulatio...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
دوره شماره
صفحات -
تاریخ انتشار 2012