Algorithmic Optimization and Validation of a Multi-Sensor Wearable System for Exploring Gait Under Extreme Terrain Conditions.

Through the examination of human gait in challenging environments, valuable knowledge can be acquired regarding the biomechanics of walking on unstable surfaces. This understanding is crucial for advancing the development of resilient and steady robotic systems. Human gait serves as a benchmark for performance and a significant source of inspiration for legged machines, improving the stability and functionality of robotic technology. Wearable systems offer distinct advantages over traditional laboratory-based setups when studying human gait in challenging conditions. They enable data collection in real-world environments that closely resemble the conditions encountered by legged robots. This study aims to validate the use of the INDIP system, a wearable multi-sensor system comprising inertial modules, pressure insoles, and distance sensors, for accurately characterizing gait in diverse and unstructured terrains. In this work, a comprehensive study was conducted to extract spatio-temporal parameters and events from human gait on irregular surfaces. The INDIP algorithm was improved to ensure its robustness across various types of terrains. The obtained results were validated using a stereophotogrammetry system called Vicon, known for its accuracy in gait analysis. The validation of the INDIP system in regular terrains for both healthy and pathological subjects has already been established. However, this research takes a significant step forward by validating its effectiveness in analyzing human gait within challenging terrains. Through the optimization of the algorithm, remarkable improvements have been achieved. The maximum time difference in event identification between the INDIP system and a Stereophotogrammetry system has been reduced to less than 0.1 seconds for different terrains. Additionally, the maximum mean error for stride length achieved is approximately 4%, for stride time it is around 1%, for cadence it is 1.3%, and for walking speed it is 3%. These advancements highlight the enhanced accuracy and reliability of the INDIP system in capturing and analyzing gait parameters in demanding terrains. The findings can aid in improving the stability and performance of legged robotic systems by incorporating insights gained from the biomechanics of human gait.