Sensor System for Real-Time Classification of Manual Construction Tasks with Power Tools for Exoskeleton Control

Work-related musculoskeletal disorders (WMSD) continue to be a significant cause of work incapacity. Exoskeletons have the potential to prevent these disorders, with passive exoskeletons already proving their usefulness but also displaying limitations. However, currently there is a lack of suitable methods to control active exoskeletons that offer additional advantages like supporting the user just when needed. This paper proposes a novel sensor-based system and a classification methodology for various work-related activities, based on the kinematic and power tool activation signals. First, requirements and thresholds for the sensor system and its signals are derived from representative activities in the construction environment. Then, the paper introduces a sensor system that collects the tool's kinematic signals and activation. The sensor system consists of an inertial measurement unit (IMU), pressure sensor, and WiFi-capable micro-controller for data transmission. Based on the signals, a threshold-based algorithm, capable of identifying six predefined activities is designed. The validity of the sensor system and algorithm is assessed through a custom-designed test course, reflective of real-world construction scenarios. With the developed test course, a study is conducted, and the activities are classified using the signals of the sensor system. The results demonstrate that the defined activities can be distinguished based on the kinematic and activation signals of the power tools.