Trajectory Planning for Autonomous Vehicles in Time Varying Environments Using Support Vector Machines

A novel trajectory planning method is proposed in time varying environments for highway driving scenarios. The main objective is to ensure computational efficiency in the approach, while still ensuring collision avoidance with moving obstacles and respecting vehicle constraints such as comfort criteria and roll-over limits. The trajectory planning problem is separated into finding a collision free corridor in space-time domain using a support vector machine (SVM), which means solving a convex optimization problem. After that a time-monotonic path is found in the collision free corridor by solving a simple search problem that can be solved efficiently. The resulting path in space-time domain corresponds to the resulting planned trajectory of the vehicle. The planner is a deterministic search method associated with a cost function that keeps the trajectory kinematically feasible and close to the maximum separating surface, given by the SVM. A kinematic motion model is used to construct motion primitives in the space-time domain representing the non-holonomic behavior of the vehicle and is used to ensure physical constraints on the states of the vehicle such as acceleration, speed, jerk, steer and steer rate. The speed limits include limitations by law and also rollover speed limits. Two highway maneuvers have been used as test scenarios to illustrate the performance of the proposed algorithm.