Motion Planning Method for Car-Like Autonomous Mobile Robots in Dynamic Obstacle Environments

Motion planning between dynamic obstacles is an essential capability to achieve real-world navigation. In this study, we investigated the problem of avoiding dynamic obstacles in complex environments for a car-like mobile robot with an incompletely constrained Ackerman front wheel steering. To address the problems of weak dynamic obstacle avoidance and poor path smoothing in motion planning with the traditional Timed Elastic Band (TEB) algorithm, We proposed a hybrid motion planning algorithm (TEB-CA,Timed Elastic Band-Collision Avoidance) that combines an improved traditional TEB algorithm and Optimal Reciprocal Collision Avoidance (ORCA) model to improve the ability of the robot to predict dynamic obstacles in advance and avoid collisions safely. Moreover, We also add new constraints to the traditional TEB algorithm, including: jerk constraints, smoothness constraints, and curvature constraints. The algorithm is implemented in $C++$ and evaluated experimentally in the Gazebo and Rviz simulation environments of the Robot Operating System (ROS), as well as in actual experimental tests on our car-like autonomous mobile robot “Little Ant” which proves the effectiveness of the method, and that the motion planning scheme is more effective in avoiding dynamic obstacles than the traditional TEB and DWA algorithms.