Communicating the intention of an automated vehicle to pedestrians: The contributions of eHMI and vehicle behavior
External human-machine interfaces (eHMIs) are solutions that may be used to convey the intent of an Automated Vehicle (AV) based on its anticipated path and actions. They may help to resolve ambiguity, for example in situations such as crosswalks where pedestrians and AVs interact, as well as increasing pedestrians' sense of trust in AVs generally. However, some research has suggested that eHMIs do not significantly impact road crossing decisions, and that the motion dynamics of a vehicle, such as visibly slowing and stopping as it approaches a crosswalk, are more important to convey intention.
A study by a team of researchers based in the Netherlands sought to understand how an AV's yielding behavior impacted the efficacy of its eHMI system. The experiment used a prototype AV, with a human driver in a "seat suit" to visually preserve the illusion of a self-driving car. Participants were asked to use a sliding scale to rate their willingness to cross the street as the car approached.
Four behaviors were tested with an initial speed of 50 km/h:
- gentle braking, with a total braking distance of 40 meters;
- early braking, in which the car braked early but continued slowly forward to the intersection;
- aggressive braking, with a sudden braking distance of 19 meters starting at 24 meters from the intersection; and
- constant speed, in which the vehicle did not slow nor did it indicate a willingness to yield.
- In nearly all instances, the researchers found that the eHMI provided increased confidence in pedestrians' willingness to cross in front of a yielding vehicle. This effect was most pronounced for the "gentle braking" and "early braking" behavior scenarios.
- The impact of the eHMI varied over time, relative to the point at which a yielding vehicle stopped. Participants in scenarios involving the eHMI generally indicated earlier and higher levels of confidence to cross upon observing a vehicle's slowing behavior.
- For example, in the "gentle braking" scenario, participants observing the eHMI reported increased willingness to cross the road 6 seconds before the vehicle completely stopped, while those without the eHMI only became more confident 3 seconds before the vehicle completely stopped.
- Similarly, in the "aggressive braking" scenario, participants observing the eHMI reported increased confidence 3 seconds before the vehicle completely stopped, compared to 1.5 seconds before the vehicle's stop for participants without the eHMI.
- Additionally, participants seeing the eHMI reported increased confidence in their willingness to cross. in the "early braking" scenario, this difference reached around 25 percentage points; in the "gentle braking" scenario, the eHMI group was up to 50 percentage points more confident than the non-eHMI group at the same relative time.