Development and Validation of a Smart Hallway for Human Stride Analysis Using Marker-Less 3D Depth Sensors

Abstract

Human stride analysis is a viable tool for identifying mobility changes, classifying abnormal gait, estimating fall risk, monitoring progression of rehabilitation programs, and indicating progression of nervous system related disorders. Marker based motion analysis systems are accurate for stride analysis but are not feasible for monitoring larger populations since external markers or sensors must be attached on the body. These external markers or sensors may be awkward for the person and may affect their natural movements. Marker-less depth sensors have shown potential to capture human movement without external markers and, with automation, could be applied to larger populations with less human resource requirements. However, existing marker-less depth sensor research has reported poor foot tracking accuracy. Hence, the goal of this thesis was to determine an appropriate depth sensor, define a multiple depth sensor system for foot tracking that could work in an institutional hallway (Smart Hallway), develop a viable foot tracking algorithm with depth sensors data, and extract accurate stride parameters. Six non-interfering Intel RealSense D415 were set up in an institutional hallway scenario and captured the lower leg of a walking person. The developed foot tracking algorithm was applied to the combined point cloud data from these sensors and the obtained stride parameters were compared with a reference gold standard Vicon system. Results showed that this high frame-rate marker-less system (approximately 60fps) was in good agreement with the Vicon system for temporal stride parameters, with mean errors less than 10ms. For spatial related stride parameters, errors were observed for step width and foot angle due to random noise generated around the foot in the depth sensors data. This research supported the depth camera approach for stride timing analysis. Further research is required to improve noise reduction for better step width and foot angle measurement.