Design and Experimental Validation of a Minimal Adaptive Real-time Visual Motion Tracking System for Autonomous Robots

摘要

We present a real-time system for motion detection and tracking for autonomous robots with limited computational resources. The system uses an at-tentional mechanism to combine color and motion blob information, as well as prediction mechanisms to improve the overall system performance. We demonstrate the functionality of the system in several test scenarios and also discuss solutions to various problems that arise from the limitations of the robotic platform.