Fig. 2

MATLAB simulation results. The robot is set to run from the origin (50, 0) to the endpoint (250, 250), and the maximum speed does not exceed 2 m/s. There are multiple obstacles of different sizes in the environment. The size of the rectangular window is 40 × 34. In the scroll window, the small “。” indicates the position of the robot and forms the robot’s movement trajectory. a, b The same environment, in b robot can walk from small space to target point. This result show that Behavior Dynamics Rolling Window Path Planning adding Q learning algorithm is much suitable than Behavior Dynamics Rolling Window Path Planning