Putting humanlike ‘eyes’ on self-driving cars helps minimise collisions

Moving eye cues on self-driving cars may enable pedestrians to predict the intentions of approaching vehicles.

A new study conducted at the University of Tokyo suggests that the use of robotic “eyes” on self-driving vehicles could increase pedestrian safety. The study used virtual reality (VR) games, in which participants had to choose whether or not to cross a road in front of a moving vehicle based on how the vehicles eyes looked. The study found that when the vehicle was equipped with robotic eyes that either gazed at the pedestrian (registering their presence) or away (not registering them), the participants were able to make safer, more effective decisions.

Autonomy of vehicles or not?

Self-driving vehicles may become one of the mainstays of the ride-hailing industry in 2023, such as what has been seen in Las Vegas. As a result, many studies are being carried out on the suitability of the technology across a range of applications, such as package delivery, preparing fields for planting, or transporting children to school. Some studies are also focusing on the safety of autonomous cars for passengers and other road users.

Interestingly, the results of a survey published in September 2022 showed that 3 in 4 Americans, representing 76% of the population, said they would feel less safe in self-driving cars.

A study conducted by Finnish researchers in 2018 examined the experiences of people who travelled in self-driving vehicles of various types in different conditions. The outcome of that study showed that “trust, safety and security” were the major indices that influenced “people’s positive attitudes towards using autonomous vehicles.” The study indicated that the attitudes of participants towards self-driving vehicles were not significantly influenced by weather conditions, such as heavy winter rain or snow.

One significant distinction with self-driving cars is that either no one is behind the wheel at all or the driver may become more of a passenger and not pay full attention to the road. This makes it challenging for pedestrians to determine whether a car has noticed them or not because there may not be any eye contact or other cues from those that are inside.

One research project, which was published in 2021, warned that autonomous vehicles not the objects in their immediate environment may result in accidents, although they may considerably increase road safety. 

The National Highway Traffic Safety Administration documented 400 distinct accidents in the United States between July 2021 and May 2022 that involved or were brought on by cars that at least had partially automated control systems. Anyone involved in a car accident may need to contract an experienced personal injury attorney in San Francisco or other states to handle the case.

It noted that while it is possible for automated vehicles to cut down the number of road accidents, it may not be so for the severity of accidents.

The experiment

Researchers at the University of Tokyo tested out if the robotic eyes could help self-driving cars “see” their environment better and help reduce collisions. “There is not enough investigation into the interaction between self-driving cars and the people around them, such as pedestrians,” Professor Takeo Igarashi of the Graduate School of Information Science and Technology said.

So how might pedestrians be alerted when a self-driving car sees them and is going to stop? To answer the question, a self-driving golf cart was fitted with two sizeable, remote-controlled robotic eyes, just like a character from the Pixar film Cars (Figure 1). The researchers tested whether individuals would still cross the road in front of a moving vehicle when pressed for time if moving eyes were placed on the cart. 

Figure 1: Lightning McQueen from the Pixar animation Cars, in the Pixar Play Parade
Credit: Flickr

The group created four situations (Figure 2): two scenarios with the cart having eyes and two without; whether the cart had seen the pedestrian and was going to halt, or it had not seen them and was going to keep driving; when the cart had eyes, they would either be staring at the pedestrian and preparing to halt or they would be looking somewhere else and not going to stop.

Figure 2: The four scenarios
Credit: Chang et al. 2022.

Since it would be obviously dangerous to ask participants to decide whether or not to walk in front of a moving vehicle in real life (though there was a hidden driver for this experiment), the team recorded the scenarios using 360-degree video cameras. That way, the 18 participants (nine women and nine men, aged 18 to 49, all Japanese) experienced the experiment in VR. They were given three seconds each time to select whether or not they would cross the road in front of the cart after going through the situations many times in a random order, and the researchers recorded their responses.

Project lecturer Chia-Ming Chang, a part of the research team, said that the findings indicated a clear disparity between genders, “which was very surprising and unexpected.”

“While other factors like age and background might have also influenced the participants’ reactions, we believe this is an important point, as it shows that different road users may have different behaviours and needs, that require different communication ways in our future self-driving world,” Chang said.

“In this study, the male participants made many dangerous road-crossing decisions (i.e., choosing to cross when the car was not stopping), but these errors were reduced by the cart’s eye gaze. However, there was not much difference in safe situations for them (i.e., choosing to cross when the car was going to stop),” explained Chang. “On the other hand, the female participants made more inefficient decisions (i.e., choosing not to cross when the car was intending to stop) and these errors were reduced by the cart’s eye gaze. However, there was not much difference in unsafe situations for them.” 

In the end, the trial demonstrated that the eyeballs led to a safer or smoother crossing for everyone. 

In this study, the male participants made many dangerous road-crossing decisions (i.e., choosing to cross when the car was not stopping), but these errors were reduced by the cart’s eye gaze.

Chia-Ming Chang, University of Tokyo

But how did the participants feel after seeing the eyes? 

While some participants found them to be adorable, others found them to be spooky or frightening. Many of the male participants said they felt the situation was more dangerous when the eyes were turned away. Many female participants reported feeling safer when the eyes were fixed on them.  

“We focused on the movement of the eyes but did not pay too much attention to their visual design in this particular study. We just built the simplest one to minimise the cost of design and construction because of budget constraints,” explained Igarashi. “In the future, it would be better to have a professional product designer find the best design, but it would probably still be difficult to satisfy everybody. I personally like it. It is kind of cute.”

He believed additional research and effort are required to deliver safety and assurance to society in regard to self-driving cars.

The researchers said the study had limitations owing to the small number of participants acting out just one scenario. They also noted that it is probable that decisions made in VR differ from those made in the real world. 

Chidirim Ndeche

Chidirim Ndeche is a reporter at Breakthrough.

Latest from Blog