A Local Path Planning Algorithm Based on Pedestrian Prediction Information

With the continuous development and practice of robotics, a large number of service robots have appeared in the areas with large human flows, such as shopping malls, schools, hospitals, and restaurants. Many people make calls or operate mobile phones while walking, but fail to observe the surroundings carefully, thus they are prone to be collided with the moving robots. Therefore, the motion planning of the robot is subject to higher requirements by the highly dynamic environment. In order to improve the robot's mobility and intelligence, a local path planning algorithm based on pedestrian prediction is proposed in the paper. The position and speed of pedestrians relative to the robot is taken as the priori input of the algorithm, and the concept of an elliptical pedestrian zone which dynamically changes with the speed between the robot and the pedestrian is innovatively proposed. By extending the scoring function of the classic dynamic window approach, the robot's intelligent avoidance of pedestrians is achieved, and the interference caused by robot movements on pedestrians' moving intentions is minimized. The experimental results show that the algorithm can effectively reduce the risk of the collision between robot and pedestrian, and can make a pre-judgment and intelligent avoidance without the attention of the pedestrian.