3D Mapping and Stability Prediction for Autonomous Wheelchairs

Abstract: Autonomous wheelchairs can address a very large need in many populations by serving as the gateway to a much higher degree of independence and mobility capability. This is due to the fact that the big picture idea for autonomous wheelchairs integration into the transportation chain is to allow for individuals to be able to utilize the Intelligent wheelchair to reach the vehicle (regardless of terrain), mount into autonomous wheelchair that navigates to desired destination, and finally autonomous wheelchair dismounts. This will enable a higher degree of mobility for a handicapped population that experiences a large quantity of restrictions as a result of their circumstances. In order for this potential to be achieved numerous precautions must be integrated into the control system, such as stability maintenance. This paper focuses on mapping the environment through the use of a LiDAR sensor and predicting the stability of the given wheelchair. We utilize RTAB Mapping in combination with LiDAR odometry to construct a 3D map of the environment. Then Poisson reconstruction is deployed to convert the built 3D pointcloud into triangular mesh that allows for the norms to the surface to be calculated, which allows for stability prediction. This paper, not only outlines a novel pipeline but also deployed the pipeline on the recently released Intel RealSense L515 sensor and leverages its unique capabilities.