RoboGait: a robotic system for non-invasive gait analysis

Abstract

The most common methods used in gait analysis laboratories are systems based on the use of markers and/or sensors positioned all over the patient’s body while performing a walking test. These approaches usually require individual calibration, a long time to set up the patient, and, therefore, discomfort of the users. Besides, some of the methods can only be performed in specific small scenarios that need to be previously set-up with external sensors. The presented system, RoboGait, is designed to overcome these problems while maintaining a good performance in terms of quality of the measurements provided. RoboGait is a mobile robotic platform that moves in front of a patient that is walking. The system measures the configuration of the patient’s body using an RGBD camera mounted on the top. Initial measurements provided by the camera are processed using an Artificial Neural Network that improves the estimated kinematic and spatio-temporal signals of the patient’s movement. This paper shows the effectiveness of the system by comparing with a validated method that uses a Vicon® system. Then, the work shows the usefulness of RoboGait in a clinical environment by using it to classify healthy and pathological gaits. In this case, the results have been compared to a reference system based on inertial sensors called Xsens®. The results show a great potential for the use of RoboGait for clinical patient assessment and monitoring, and for pathology identification.