Tracking the trajectory of a swarm of mobile robots with a computer vision system

摘要

Swarm robotics research uses a range of tools for evaluating the behaviors and metrics of robot collectives. One crucial tool involves the capability to track each robot’s position and orientation at various intervals, enabling the reconstruction of individual robot poses and trajectories. Comprehensive analysis of swarm behavior hinges on the study of the collective trajectories of each robot within the group. This paper demonstrates the implementation of a computer vision system, utilizing a webcam and Python scripts, to effectively track a mobile robot group within a swarm. This shows the feasibility of developing such research tools using commonplace computing equipment. The design and development of the vision system, including a detailed calibration procedure, robot identification methods, and practical examples, are also shown. Furthermore, it offers an exhaustive explanation of the robot tracking process. Experimental trials with three robots validate the system’s ability to extract images from video feeds and accurately identify each robot. Subsequently, after image processing, the system generates a dataset encompassing image numbers, robot IDs, x and y positions, and orientations.