Construction of a Mobile Robot Simulation Platform for Human-Following Tasks

The testing of human-following algorithms for mobile robots in real environments is frequently hindered by many challenges, including high experimental costs, inefficient training processes, and safety risks for personnel. To address these issues, this paper constructs a simulation platform based on Webots for testing and iteration of human-following algorithms. The platform was developed by constructing a human leg skeleton model and driving this model using motion capture data from publicly available datasets, thereby providing a target for the mobile robot to follow. Meanwhile, a two-wheeled mobile robot model equipped with a Kinect camera was created and validation experiments were conducted using a baseline human-following strategy to control the mobile robot.The experimental results demonstrated that the robot was capable of effectively following a target at a distance of 1.2 meters in the simulation environment, with a maximum error of 0.12 meters, and the visual tracking error was maintained within ±100 pixels. This result validates the effectiveness and feasibility of this simulation platform in human-following tasks. Future work will improve and test human following algorithms based on this platform to enhance the robot's recognition of human movement intentions and human following performance.