Path Planning of Autonomous Driving Based on Quadratic Optimization

In order to deal with the scenario that require high flexibility in obstacle avoidance, such as urban roads, this paper proposes an autonomous driving path planning algorithm based on quadratic programming (QP). The algorithm proposes an obstacle avoidance cost based on Frenet frame, which can not only satisfy the characteristics of the positive definite quadratic form of the cost function, but also add the obstacle avoidance cost as a soft constraint, and then adapts an iterative solution strategy. The candidate paths are generated by solving the QP problem, the algorithm will output the optimal path, which satisfy the collision detection. The simulation test shows that the algorithm can deal with nudge, lane change and complex obstacle avoidance scenarios.