Research on Human Body Detection and Trajectory Tracking Algorithm Based on Multi-sensor Fusion

Abstract

At present, most mobile robots regard the human body as a moving obstacle in the process of autonomous movement in a scene with pedestrians, and avoid the human body when planning a local path. This situation is not convenient for pedestrians, and it will also cause problems for robot navigation. In many cases, the path of the robot and the pedestrian conflicts, causing the robot to fall into a freezing (stand-by-place) posture until the pedestrian disappears from the local map of the robot. In order to solve this problem, the solution proposed in this paper adopts multi-sensor data fusion technology, uses laser scanners and depth cameras to detect pedestrians, and establishes a pedestrian motion model to realize human body recognition and motion prediction based on multisensor fusion.