Human motion analysis is an important instrument for assessing gait in neurological diseases or for exercise recognition. Established gold standard solutions for human motion analysis such as OptoGait or the Vicon camera system are very expensive and some require additional tracking markers attached on the body. In addition, these high-quality measuring systems, especially the multi-camera setup, are not suitable for measurements outside the laboratory, e.g. in everyday life. This becomes a problem if, for example, the rehabilitation progress of a stroke or Parkinson's patient is to be measured continuously even after discharge from hospital. Usually, patients are isolated at home and have no supervision or therapist around them. This increases the need for unobtrusive gait monitoring systems that can work at home without an operator and that provide the same metrics and parameters from rehabilitation.
Therefore, our goal is to develop low-cost, unobtrusive (markerless) methods for capturing human motion using commercially available sensor devices such as Inertial Measurement Units (IMUs) that measure acceleration and low-cost cameras such as the Microsoft Kinect camera or smartphones that can work outside the lab. For gait analysis, by using these technologies, common spatiotemporal parameters such as step length, cadence, cycle/stand time/swing time, speed, play and rotation speed can be estimated [1, 2]. These gait parameters, along with other information derived from the raw signals, help us to quantify gait and general movement changes that are significant in daily life on an individual level. For example, detecting when a person (or rehabilitation patient) is affected by physical fatigue or excessive cognitive load, thus is susceptible for falling or injury. In addition, we are also developing algorithms to quantify rehabilitation progress in real-life scenarios, such as during stroke rehabilitation.
[1] Tunca, C., Pehlivan, N., Ak, N., Arnrich, B., Salur, G., & Ersoy, C. (2017). Inertial sensor-based robust gait analysis in non-hospital settings for neurological disorders. Sensors (Switzerland), 17(4), 1–29.
[2] Trautmann, J., Zhou, L., Brahms, C. M., Tunca, C., Ersoy, C., Granacher, U., & Arnrich, B. (2021). TRIPOD—A Treadmill Walking Dataset with IMU, Pressure-Distribution and Photoelectric Data for Gait Analysis. Data, 6(9), 95.
