ABSTRACT
Robotic-walker-aided rehabilitation of patients with gait disorders has shown improvements over classical methods. However, it requires constant supervision from a medical expert so that collisions are avoided. In this paper, an autonomous controller capable of guiding the patient to a destination while avoiding collisions is proposed as a replacement for human supervisors. The association of a LiDAR sensor and Google’s Cartographer algorithm creates the map, registering obstacles. The global planner A* plans a path to meet the user’s selected destination. The local planner Dynamic Window Approach was also implemented, which considers the dynamic constraints. The autonomous control was validated with 3 healthy volunteers using a walker under 5 validation scenarios, assessing its suitability of use as a supervisor in a shared control. Results show a mapping and localization accuracy of 0.38 m. The walker never registered any collision and always reached the destination, highlighting its applicability as human supervisor replacement.
