Smart Walker With Return-To-User Functionality via Kalman Filter-Based Tracking

This paper presents a smart walker designed to address mobility issues for individuals with various impairments. The prototype comes with autonomous and manual navigation options; it uses an NVIDIA Jetson Orin Nano with ROS 2 Humble, integrates sensors such as the Intel RealSense D455 camera for depth data and an MPU6050 IMU for orientation, and it is equipped with person-tracking algorithms for autonomous navigation, with face recognition to maintain focus on the target user. Commands are communicated between the Jetson Nano and Arduino Mega 2560, which controls motor functions and the mecanum wheels. The forward-looking aspect is that the prototype retrofits standard walkers for easier commercialisation and replication. The design addresses the need for dynamic obstacle detection, basic fall prevention, as well as return-to-user autonomous navigation – all of which are necessary for user independence and safety in various environments.