Research On Mental Workload Evaluation And Prediction In Civil UAV Operation

The accident rate of drones is usually higher than that of manned aircraft according to investigations,and the accident rate is rising.A major cause of drone accidents is due to human operation.The separation of drones from human senses puts forward higher requirements on the operator’s control and attention.In complex tasks,operators are prone to misoperation and slow response in complex tasks.Evaluation of the mental workload is the key to improving the safety of the human-drone system.Existing research is mostly carried out in a simulator,this paper conducts a drone flight test to study the mental workload of the operator,which has important theoretical and practical significance for ensuring the flight safety of the drone and the early warning of high mental workload.Firstly,the UAV simulator test was designed and implemented to explore the rationality of the mental workload in the UAV operation.Data of subjective rating,physiology index and task performance were measured during the experiment.The differences of different scenes were tested after the normal test of data,and then the main factors that affects the mental workload was studied.The results show that the mental workload of dual tasks is higher than that of single task under the same wind speed,and the mental workload of high wind speed is higher than that of low wind speed under the same task at the same time.In other word,the more complex the operation and environment,the higher the mental workload.All measures can differentiate most task conditions.Then,the subjective rating scales and heart rate monitor were used as measurements in the UAV operation experiment.To explore the performance of measurements,fourteen participants without UAV operation experience participated in this study.They joined UAV flight training and then operated the UAV under four mission difficulty scenarios.A Polar V800 heart rate monitor was used to measure the heart rate and heart rate variability data during the experiment.The mental workloads of the participants were measured using the NASA-TLX scale,modified Cooper-Harper(MCH)scale and SWAT scale after the missions were completed.The subjective rating and heart rate data are compared between difficulty conditions.The results show that the subjective ratings and MEANHR increase,while MEANRR decreases with the increase of mission difficulty.All measures can differentiate most task conditions and there is a strong,positive correlation and high convergent validity between subjective rating scales,an explicit link between subjective ratings and performance is demonstrated.Lastly,a mental workload classification model is constructed to provide a reference for the early warning of high UAV mental workload.Principal component analysis(PCA)is used to reduce the dimensions of some indicators to obtain the mental workload index set,including the heart rate index,the subjective composite index,the number of errors,operation time,and the model base on gradient boosting decision tree(GBDT).The results show that the model accuracy is 92.54%,and the precision and recall is 0.943 and 0.936,respectively.The F1-score is 0.933,which also performs well.Operation time and the subjective composite index occupy a relatively important position in the model,accounting for 53.001% and 38.035%respectively.The GBDT algorithm has excellent practicability and fitting performance in construction of UAV mental workload classification model.This paper has 26 figures,44 tables,and 107 references.