Identifying real-world problems with automated vehicles by detecting behavioral differences in steering movements between the human driver and machine

In recent years, the validation of automated features in vehicles has become more extensive and complex than initially anticipated. Unfortunately, reality has caught up with the original optimism and enthusiasm that envisaged fully automated vehicles on the roads by 2020. The development of vehicles that can operate fully autonomously, without human intervention, worldwide in all environmental conditions remains an open problem and still requires some development effort. The preparation of an approach for certification and validation of autonomous car systems completely independent from the vehicles control system themselves remains a significant problem. To validate the safety of the systems, this work introduces the Shadow System (SS), an original and novel system that allows the detection and handling of problems in automated vehicles under real conditions. The method uses the human driver as a reference to verify the safe operation of a vehicle. A way to measure behavioral differences for steering movements is presented using a specially developed component called the Shadow Steering Wheel (SSW). This enables the steering movements of the human driver and automated vehicle to be measured independently. The proposed system is based on non-intrusive techniques. With the presented method, safety-critical and problematic traffic scenarios can be objectively identified in real time under real conditions during the validation phase and safely corrected by the human driver.