The Application of Human Mental Models for Engineering to Improve Acceptance and Performance of Driving Automation

The vision of advanced information technologies to provide intelligent driving assistance and automation is currently being reconciled with humans operating these technologies in complex, real-time environments where sometimes unpredicted situations need to be mastered under time pressure. Could automation technologies be designed such that humans can collaborate with them more quickly and effectively to solve the Unpredicted? We investigate the utility of computational Human Mental Models for Engineering (HMMEs) toward developing automation systems that are more similar to human behavior. We validate and compare an HMME with a control model for a basic steering task and compare them both with driving data from 16 human drivers in a driving simulator. We report on the observed characteristics of the HMME to support multi-tasking, graceful degradation, and multi-sensory driver state integration.