Ergonomic Knee Exoskeleton System with Real-World Perception and Actuation Using Computer Vision, Image Processing and LiDAR Data Fusion

This paper introduces an innovative and accessible knee exoskeleton design to empower the visually impaired and differently abled individuals in their regular motor functions by integrating advanced perception and processing techniques. Our proposed design features a robust computational framework and dedicated computer system, acting as a surrogate for visual perception. Notable contributions encompass its affordability, ergonomic design, and the fusion of sensory data from a lightweight camera module and a LiDAR sensor, optimizing functionality to simulate a standard knee joint. Prioritizing user comfort and ease of implementation, this research significantly advances affordable assistive technology using commonly available actuating and sensing elements, offering enhanced mobility and a better quality of life for both visually impaired individuals and those with limited leg motor control. We have also developed and tested the physical prototype to demonstrate implementation feasibility of this concept, rendering acceptable detection and actuation performance. The proposed system demonstrates a sequential development strategy based on camera–LiDAR data fusion method, promising improved societal integration and independence for its users. This study exhibits how our proposed system achieves the end goals of a contemporary knee exoskeletal system with twice the functionality but at a fraction of its cost, given that our system also features an active sensing and data fusion paradigm as well. In addition, this study serves as a precursor to a general developmental roadmap for an economical albeit highly functional exoskeletal device which can facilitate the further development and implementation of affordable exoskeletal and bionic solutions at large.