Safety analysis using a Smart Safety Helmet embedded with IMU and EEG sensors applied in industrial facility.

摘要

Some mental states, such as fatigue, or sleepiness, are known to increase the potential of accidents in industry, and thus could decrease productivity, even increase cost for healthcare. The highest rate of industrial accidents is usually found among shift workers due to fatigue or extended work hours. When using machine tool or interacting with robotic system, the risk of injury increases due to disturbance, lapse in concentration, vigilance decline, and neglect of the risk during prolonged use.;Usually, to guarantee safety of worker, the conventional means is to stop the machine when human presence is detected in the safeguarding area of machine tool or robot workspace. The popular human detection technologies exploit laser scanner, camera (or motion tracker), infrared sensor, open-door sensor, static pressure sensitive floor as described in CSA Z434 standard. Of course, in the field of robotic, human and robot are not allowed to work together in the same workspace. However, new industrial needs lead research to develop flexible and reactive chain production for enabling small quantity production or fast modification in product characteristics. Consequently, more efficient human-machine or human-robot collaboration under a safety condition could improve this flexibility.;Our research project aims at detecting and analyzing human safety in industry in order to protect workers. Comparing to the conventional human protection methods, our research exploits Artificial Intelligence approach to track and monitor human head motion and mental state using an instrumented safety helmet, labelled as Smart Safety Helmet (SSH) in the following. The contribution of this thesis consists in the design of data fusion algorithm for the recognition of head motion and mental state, which can be used to analyze the potential risky states of workers. A Smart Safety Helmet embedded with Inertial Measurement Unit (IMU) and EEG sensors will be used to detect and decode the human's mental state and intention. The acquired information will be used to estimate the accident risk level in order to stop machine and then prevent accident or injury. In human-robot interaction (HRI) paradigm, the human's intention could be used to predict the worker trajectory in order to control the robot moving trajectory and then to avoid fatal collision.