Vision-Based Control of a Mobile Platform Using Edge Artificial Intelligence

Mecanum platform Case Study

Abstract

This paper presents a vision-based motion generation system for mobile platforms. The robot is controlled throw human hand gestures. A lightweight camera captures video streams from real-world environments, while an artificial intelligence-based gesture recognition model identifies the operator's hand gestures and generates the corresponding motion commands. The proposed solution is implemented and validated on a real Mecanum mobile platform.  Two lightweight artificial intelligence models (Faster Objects, More Objects and MobileNetV1) for hand gesture recognition are embedded and evaluated on an ESP32-CAM module.