Human-Machine Interface Development For Modifying Driver Lane Change Behavior In Manual, Automated, And Shared Control Automated Driving

Rear-end crashes are common on U.S. roads. Driver assistance and automated driving technologies can reduce rear-end crashes (among other crash types as well). Braking is assumed for forward collision warning (FCW) and automatic emergency braking (AEB) systems. Braking is also used for adaptive cruise control (ACC) and in automated driving systems more generally. However, steering may be advised in an emergency if the adjacent lane is clear and braking is unlikely to avoid a collision. Steering around an obstacle when feasible also eliminates the risk of becoming the new forward collision hazard. Driver assist technology like emergency steer assist (ESA) and Level 2 or Level 3 automated driving systems might facilitate manual emergency lane changes but may require the driver to manually initiate the maneuver, something which drivers are often reluctant to do. An Human-Machine Interface (HMI) might advise the driver of a steerable path when feasible in forward collision hazard situations. Such an HMI might also advise a driver of normal lane change opportunities that can reduce travel time, increase fuel efficiency, or simply enhance the driving experience by promoting `flow.\u27 This dissertation investigated the propensity of drivers to brake only versus steer in both manual and automated driving situations that end in a high-intensity forward collision hazard. A audio-visual Field of Safe Travel (FOST) cluster display and haptic steering wheel HMI were developed to advise drivers in both discretionary and emergency situations of a lane change opportunity. The HMI was tested in a moving base simulator in manual driving, in fully autonomous driving, and in shared-control autonomous driving during a simulated highway commute that ended in an high-intensity forward collision hazard situation. Results indicated that a) driver response was affected by the nature of the automated driving (faster response in hands-on shared control versus hands-off fully autonomous driving); b) exposure to the HMI in normal lane changes both familiarized the driver with the HMI and introduced a mental set that steering was also a possibility rather than braking only; c) but that drivers used their direct vision to determine their response in the emergency event. A methodological issue related to mental set was also uncovered and resolved through screening studies. The final study brought the dissertation full-circle, comparing hands-off fully automated driving to hands-on shared control automated driving in the context of either providing some or no exposure to the developed LCA system concept. Results of the final study indicated that shared control lies somewhere between that of manual driving and hands-off fully automate driving. Benefits were also shown to exist for the LCA system concept irrespective of whether the discrete haptic profiles are included or not. The discrete haptic profiles did not statistically reliably increase response times to the FC hazard event, although they do show a trend toward decreasing response variability. This finding solidified the fact that by implementing a system for benign driving that aids in establishing a mental set to steer around an obstacle may actually be beneficial for rear-end crash scenarios. This dissertation’s contributions include a) audio-visual FOST display concepts; b) discrete haptic steering display concepts; c) a paired-comparisons scaling for urgency for haptic displays applied while driving; d) a new ``mirage scenario\u27\u27 methodology for eliciting subjective assessments in the context of a forward collision hazard, briefly presented then removed, without risk of simulator sickness, and e) a methodological lesson for others who wish to investigate semi-automated and automated driving interventions and must manage driver mental set carefully