Hand Posture Control of a Robotic Wheelchair Using a Leap Motion Sensor and Block Sparse Representation based Classification

Abstract

In this study, a gesture and posture recognition method which is based on the Block Sparse, Sparse Representa- tive Classification, and its use for a robotic wheel-chair control are explained. A Leap Motion sensor is used to capture the postures of the left hand. There are five postures mapped to the control commands of the power wheel-chair. These commands can be expanded as the posture recognition commands can deal with high number of classes. The MATLAB functions used in the computations are compiled into .NET programing environment. We tested the hand posture control in a hall where are occupied by tables and chairs. The navigation experiments were successful.