Fuzzy logic-based detection scheme for pilot fatigue

Purpose - The paper aims to present the development of a detection scheme for pilot fatigue using fuzzy logic. Evaluation parameters based on the dynamic response of the pilot/aircraft system are to be defined and criteria for online fatigue detection to be formulated. Design/methodology/approach - The approach is based on the idea that, while performing the same task, under otherwise identical conditions, the dynamic signatures of the pilot/aircraft system are different depending on the pilot condition, "rested" or "tired." Tests performed on a 6 degrees-of-freedom (DOF) flight simulator with pilots at two extreme levels of alertness are used to define parameters based on aircraft states and pilot input measurements that can serve as pilot fatigue detectors at steady state flight conditions. These parameters are computed using the statistics of the tracking errors (TE), state and control time histories, and the Fourier transforms of the TE. Fuzzy logic is used to evaluate the pilot condition based on composite detection parameters. Findings - Validation tests on a 6 DOF flight simulator showed that the proposed detection scheme has promising capabilities for safety monitoring purposes and design of control laws that can accommodate for pilot abnormal conditions. Research limitations/implications - The pilot fatigue detection algorithm presented in this paper can be used as a starting point for future research in the following directions: development of safety monitoring systems for warning and/or triggering of automatic control compensation; development of pilot fault-tolerant control laws; development of human pilot models for simulation, handling qualities assessment, and control laws design. Practical implications The approach for pilot fatigue detection proposed in this paper is a viable alternative to existing methods based on physiological measurements such as electrical activity of the brain, pulse, body temperature, etc. which imply direct and permanent connection of the pilot to the measurement system and interfere adversely with pilot comfort and his/her ability to perform the task. The proposed approach eliminates this drawback and does not require on-board additional heavy equipment. Originality/value - Pilot fatigue assessment from measurements of pilot/aircraft dynamic parameters has not yet been investigated as an alternative to the physiological approach.