Mental Workload Modeling of Time-Critical Tasks in Autonomous Driving Based on a Multi-source Information Fusion Approach

The supervision tasks of complex dynamic systems are usually characterized by dynamics, suddenness, timeliness and uncertainty, so it is necessary to keep human drivers at an appropriate mental workload (MWL) level to ensure that they make the best decisions, judgments and actions in the dynamic environment of the real world. This paper proposes and validates a quantitative assessment model for studying MWL in time-critical multitasking scenarios. The Multi-Attribute Task Battery II (MATB-II) is used as a multi-task platform in the experiment. The results show that the multiple linear regression model, which comprehensively considers human performance data and eye movement data, has better prediction performance compared with single data alone, and can predict the MWL level of different task scenarios, which provides a reference for switching control authority of human-machine system and alarm design of the system, and has goods application prospects.