Improved Humanoid Gait Using Learning-Based Analysis of a New Wearable 3D Force System: Work Programme

摘要

The aim of this work is to develop a new framework for exploiting humanoid adaptive gait on different grounds, especially slippery floors with a low friction coefficient. To achieve this, the idea is to (1) enhance and extend a system for real-time measurement of the Ground Reaction Forces (GRFs) of a walking humanoid robot, (2) to develop a robot controller, and (3) to use machine learning techniques to interpret the GRFs and thus adapt the robot controller in order to achieve a robot able to walk on different grounds. The overall system contribution is the improvement and diversity of humanoids locomotion capacities, leading to a more easily integration of these robots into different areas where they can collaborate with humans.