What pedestrians want from autonomous vehicles

When we think about the mobility of the future, automated driving is the topic that immediately comes to mind. Technology is often the main aspect we consider. But what kinds of information and signals do pedestrians actually want to receive from automated vehicles?

This is the issue I am currently researching as a doctoral candidate working at Daimler. As a former psychology student, I consider myself a sort of intermediary between human beings and machines — a technology coach, so to speak. In this situation, the most important element is communication.

In the road transportation of the future, the central safety features will include not only sophisticated technology such as cameras, radar, and lidar but also communication between automated vehicles and pedestrians. This is an overarching topic that affects all automakers across national boundaries.

For many people it’s absolutely inconceivable that they will no longer be able to communicate with drivers by means of eye contact. After all, a friendly nod from the driver behind the wheel that tells me, “Go ahead and cross!” will no longer exist in fully automated vehicles. This change gives rise to many questions.

What do pedestrians want as an alternative to the friendly nod?

For us — an interdisciplinary team consisting of designers, futurologists, engineers, electronics specialists, and me — that was a compelling reason to take a closer look at this relationship. Instead of looking at this situation on a computer display, we did our research in a place where the focus is on the future.

A group of 60 test subjects visited us at our new Testing and Technology Center in Immendingen in order to participate in our field study of cooperative light systems. For us, the best part of the new test center was definitely the city neighborhood, which has traffic lights, intersections, a parking lot, and everything else you need for a study of urban traffic at a site that is as realistic as possible.

Here everything has been designed so realistically that you could sometimes forget that our test vehicle — which is known in technical jargon as a Cooperative Autonomous Vehicle — wasn’t really rolling up to the intersection or pulling out of the parking lot automatically. The reason it wasn’t driving autonomously is that we only wanted to investigate pedestrian behavior as it related to the cooperative light system.

The vehicle was equipped with a homemade roof attachment that looked like a sensor, and it was driven by our trained driver Selina, who was wearing a “seat disguise.” Yes, there really is such a thing! It was specially made for us in the seat workshop. To our test subjects, it looked as though the vehicle was driving toward us without a driver at the steering wheel. By means of this scenario we made sure we would get the test subjects’ genuine reactions to automated vehicles.

It’s not unusual to have this kind of setup in a research project. This makes it possible to run through certain test scenarios easily and frequently at an early stage of development.

Apropos the setup, everyone’s familiar with the light signals in road traffic — turn-signal lights, brake lights, and the lights that are probably most important for pedestrians: green for go and red for stop. Studies conducted by the Society of Automotive Engineers (SAE) have shown that turquoise is the color best suited to represent automated driving.

There are several reasons for that. 1. Turquoise has not been used before as a color in road traffic. 2. Test subjects perceive turquoise as an innovative and pleasant color. 3. By contrast to white, turquoise can be used on every side of a vehicle. The neurobiologist Annette Werner from the University of Tübingen, has confirmed that turquoise is the best color for indicating the automated driving mode. In order to find out what kinds of light signals pedestrians want, we boldly simulated the future in two different test scenarios.

Meeting place: an intersection. The first test situation

In the first test group, we investigated several different light signals at an intersection that had no traffic lights for pedestrians.

During the first run-through, the vehicle made no signals whatsoever as it met the test subjects, who were just then crossing the street. As a result, the pedestrians didn’t know if the automated driving mode had been activated or if the vehicle was being driven manually. There’s one thing I can tell you right away: The pedestrians were not enthusiastic about this variant.

In round two, the setup was very different. A stationary turquoise light on the vehicle told the pedestrians that the automated driving mode was active.

The pedestrians reacted even better to a combination of the stationary indication of the automated driving mode with a dynamic light signal. The latter signal showed what the vehicle was planning to do next. By means of this signal, which was a pulsating brake light at the front, the vehicle clearly showed that it stops for pedestrians.

In the next-to-last round, we tested a kind of “eye contact.” In addition to the stationary light indicating the automated driving mode, the vehicle also had movable “eyes” in the sensors on the roof, which showed that the vehicle saw the pedestrians. When the pedestrians crossed the street, the light signals followed the pedestrians — just as the eyes of a careful human driver follow pedestrians in real-life road traffic. In the last round, the test vehicle turned the corner with all of the signals working in combination.

Meeting place: a parking lot. The second test situation

The second group of test subjects met our test vehicle on a parking lot. Here too, we tested the various signals, ranging from a baseline (without any kind of signal) to a flashing light above the windshield and movable “eyes” on the vehicle. The key difference between this situation and the first one was that the vehicle was communicating not that it was braking but that it wanted to drive off.

After the first day it was already clear that whatever the situation may be, pedestrians want to know if they are dealing with an automated vehicle. This is also indicated by our initial evaluations of the data. The final results are being directly passed on to the SAE, which is currently working on a recommendation for the design of light concepts for the automated vehicles of the future. When I imagine autonomous vehicles driving along the streets in a few years with a light concept that I helped to work on…that’s really cool!

Stefanie Faas, a doctoral candidate at the Innovation Workshop, likes to do in-depth research on fascinating future-oriented issues.