04-06-2025
Toyota And MIT Research Reveals Surprises In Seeking Pedestrian Safety
Demonstration of Toyota automatic emergency braking system aimed avoiding collisions with ... More pedestrians.
Some past assumptions of how drivers interact with automation and pedestrians were found to be far off the mark. That revelation came as results of the 100th research project conducted between Toyota Motor Co.'s Collaborative Safety Research Center and the Massachusetts Institute of Technology AgeLab were announced Tuesday.
The study sought to discover how to reduce collisions between vehicles and pedestrians, in part, by looking at how they interact and communicate with each other, along with how automated systems and technology affect driver behavior.
After analyzing hundreds of hours and thousands of miles of driving data over a decade's time, it became evident at least one past assumption related to driver distraction is wrong—voice recognition.
'We kind of flipped how people were originally thinking about this,' said Josh Domeyer, principal scientist at Toyota CSRC, during a presentation at the American Center for Mobility near Ypsilanti, Michigan. 'We found that not only did voice recognition performance affect the driver, but we found that the visual features of these voice interfaces were influencing how demanding they were, so people would look down to confirm whether the voice system was doing something. This internally, at least, led to some changes in our own system, where we reduced the visual feedback in order to make it more related to driver-pedestrian interaction.'
Indeed, the term 'driver distraction' is distasteful to MIT AgeLab research scientist and co-director of the Advanced Vehicle Technology Consortium Bryan Reimer who pointed out there's almost never a time a driver's attention is not divided between paying attention to the road and other actions.
He prefers the use of the phrase 'non-driving related tasks,' or NDRT, which doesn't include listening to the radio.
'We are 100% of the time distracted, whether you're listening to somebody, whether you're fidgeting and writing,' said Reimer, during the presentation. 'We as humans, if our attention isn't devoted some places are continually, either externally, visibly or internally, thinking about something else, distracted by something.'
The researchers examined 450 hours of driving data to identify 154 hours of engagement in NDRT to evaluate how driver assist technology such as lane centering, influences the decision to partake in those tasks such as mobile phone use, texting, fiddling with the infotainment system, eating or drinking.
The results were not cut and dry, and a little bit sobering.
'We're human. We find ways to do what we're going to do anyway. It doesn't matter what car I'm in,' said Reimer. 'So lots of folks out there like saying, this is better than this entity. We find a way. If there's a will, there's a way, which means that we have to think about new approaches.'
Then there's the dynamic involving interaction between driver and pedestrian. Using a dataset of 348 'naturalistic interactions,' researchers looked at the importance non-verbal communications such as motion cues and direct eye contact to help avoid collisions.
Again, there were some surprising results.
'The first thing that we saw is that glance behaviors and hand gestures were not as important as we initially predicted, which this was very surprising to us, because we expected that eye contact would be very important,' noted Domeyer.
He explained there's a spot the vehicle is approaching at a certain rate, where people don't want to cross the street and if the rate is perceived as far enough away they will cross.
But there's also what Domeyer termed a 'gray area' where it's about a fifty-fifty chance whether someone decides to cross.
'We actually created the conditions where somebody might look in a vehicle, so depending on the time of day, the angle and other things, and we found that only about 40 percent of the pedestrians didn't actually know where the driver was looking, just because of the occlusion that happened with the windscreen,' said Domeyer. 'So it was very interesting, because our original assumptions about how this communication happened were kind of challenged by this early work.'
Test figures at Toyota's dedicated garage at the American Center for Mobility near Ypsilanti, ... More Michigan.
At its facility at the American Center for Mobility, Toyota is working on technology to help prevent vehicle-pedestrian collisions.
In two demonstrations of such technology, a vehicle equipped with automatic emergency braking stopped short of colliding with a mockup of a pedestrian. In another, newer AEB capabilities were shown on a pickup truck towing a trailer.
Having completed the 100th project with MIT AgeLab, Toyota's CSRC announced its five-year research phase is continuing as it continues to seek better understanding of driver behavior, crash avoidance and minimizing crash-related injuries.
