SuperBots: Context Inferencing using Computer Vision and WiFi based through wall sensing

SuperBots is a multi-robot autonomous system designed to perform context inference in unknown indoor environments through the integration of navigation, computer vision, and WiFi-based through-wall sensing. This proposal describes the navigation subsystem, which serves as the foundational layer upon which vision and wireless sensing capabilities are built. Deployed on the TurtleBot 4 platform with onboard NVIDIA Jetson processors and ROS 2, the system implements Simultaneous Localization and Mapping (SLAM) for real-time map generation and Adaptive Monte Carlo Localization (AMCL) for precise positional estimation. Global path planning via Smoothed A* and local obstacle avoidance via the Dynamic Window Approach (DWA) enable reliable autonomous traversal of dynamic, human-occupied spaces. Navigation performance is evaluated across metrics including localization error, mapping accuracy, navigation success rate, and obstacle avoidance behavior. Beyond standalone navigation, the system is designed to satisfy the precise spatial requirements of WiFi CSI-based sensing, which depends on repeatable robot trajectories, and to enable vision-guided behaviors such as person following. Additional capabilities under development include AI-powered object detection, facial recognition, activity classification using YOLO and FaceNet, and voice command interpretation via speech-to-text integration. Together, these subsystems advance toward a cohesive mobile robotic platform capable of deep semantic understanding of indoor environments without reliance on external infrastructure.