Realtime Recognition of Complex Human Daily Activities Using Human Motion and Location Data

• Published in: IEEE transactions on biomedical engineering Vol. 59; no. 9; pp. 2422 - 2430
• Main Authors: Zhu, Chun, Sheng, Weihua
• Format: Journal Article
• Language: English
• Published: New York, NY IEEE 01.09.2012; Institute of Electrical and Electronics Engineers
• Subjects: Activities of Daily Living; Activity recognition; Algorithms; Applied sciences; Bayes Theorem; body sensor network; Detection, estimation, filtering, equalization, prediction; dynamic Bayesian network (DBN); Exact sciences and technology; Feature extraction; Humans; Information, signal and communications theory; Movement - physiology; Pattern recognition; Pattern Recognition, Automated - methods; Signal and communications theory; Signal processing; Signal, noise; Telecommunications and information theory; Telemetry - instrumentation; Telemetry - methods; Thigh; Three dimensional displays; wearable computing; Wearable sensors; Wireless communication; Wireless sensor networks; Human; Motion; Body area network; body sensor network; Bayes network; Activity recognition; dynamic Bayesian network (DBN); Localization; Sensor array; wearable computing; Signal detection; Biomedical engineering
• ISSN: 0018-9294; 1558-2531; 1558-2531
• DOI: 10.1109/TBME.2012.2190602

Daily activity recognition is very useful in robot-assisted living systems. In this paper, we proposed a method to recognize complex human daily activities which consist of simultaneous body activities and hand gestures in an indoor environment. A wireless power-aware motion sensor node is developed which consists of a commercial orientation sensor, a wireless communication module, and a power management unit. To recognize complex daily activities, three motion sensor nodes are attached to the right thigh, the waist, and the right hand of a human subject, while an optical motion capture system is used to obtain his/her location information. A three-level dynamic Bayesian network (DBN) is implemented to model the intratemporal and intertemporal constraints among the location, body activity, and hand gesture. The body activity and hand gesture are estimated using a Bayesian filter and a short-time Viterbi algorithm, which reduces the computational complexity and memory usage. We conducted experiments in a mock apartment environment and the obtained results showed the effectiveness and accuracy of our method.