| Local path planning is one of the key technologies of navigation in autonomous vehicle.In this paper,we focus on the real-time path planning,the uncertainty of perception environment and the curvature smoothness of the autonomous vehicle in complex scenes.In order to solve the problem that the path generated by 4-neighborhood or 8-neighborhood A * method is not executable for autonomous vehicle,and the problem of real-time can not be satisfied after introducing the motion primitive,this paper improves the path planning algorithm of unmanned vehicle based on A * algorithm.In this paper,we design the motion primitives and the heuristic function,and use the Reedshepps curve to accurately connect the pose between the targets to accelerate the algorithm.The comparison between ANA * algorithm and the ANA * algorithm used in our experiment platform of the unmanned vehicle is carried out.Our method has advantages in speed,smoothness of generated path and so on.For the local path planning problem caused by the lack of lane line and global path in some known environments in practical applications,this paper starts from the reality of the problem that the target point is outside the sensing area,the obstacle detection is unstable.This paper proposes a re-planning method based on sequential information fusion to reduce the influence of inconsistency of frame before and after re-planning,and we solve the problem of too many pieces of backtrack trajectory by using spline interpolation and path optimization.The validity of the method is validated by real vehicle experiment.Aiming at the problem that the multi-objective path optimization can not solve the obstacle avoidance and the parameter is difficult to tune,we proposes a single objective optimization method based on the obstacle constraints,which can optimize the path curvature based on the optimized path security.The single objective optimization method based on obstacle constraints effectively improves the real-time performance compared with multi-objective optimization. |
