Driverless cars in the ‘80s: the future takes root in Japan

With Patrick Stelmaszyk,
Group IT, Head of Enterprise Architecture at Capgemini

Read article

Driverless cars in the ‘80s: the future takes root in Japan

The autonomous driverless car is a major breakthrough on the verge of revolutionizing the automotive industry, but its roots go back to 1990, when Mazda and Capgemini first started working on this idea. Let’s take a look at this human and technological journey.

Welcome to the robotics lab

When Patrick Stelmaszyk arrived in Japan back in 1990, he found a small team working on a massive project. Recently recruited by ITMI (IT and Technology in the Industrial Sphere), a high-tech company specialized in artificial vision technologies and acquired by Capgemini in 1987, the young engineer was sent to the Land of the Rising Sun as part of an official exchange program. “ITMI and the Capgemini group financed the project. The goal was to see what was happening there in terms of new robotics and artificial intelligence technologies, to make contacts and, if possible, sign contracts,” says Patrick Stelmaszyk. What he found exceeded all his expectations…

There, in a tiny laboratory of roughly 50 people from Osaka University where he began his “apprenticeship”, Patrick Stelmaszyk discovered a world with its eyes squared on the future. “Japan was (and still is) the global hub for mobile robotics. Around the year 2000, a student from that same laboratory whom I worked with went on to create the first anthropomorphic assistant robots with life-like human faces, a rapidly growing trend in robotics today. Solid research was also conducted on artificial intelligence, automated driving and computer-aided vision. The Osaka lab where I worked was among the most advanced in the world for self-driving cars. The researchers were in constant communication with car manufacturers and there was an explosion of advancements happening. That’s how I got in touch with Mazda,” he continues. And thus began a long and fruitful relationship.

2 million
the number of pixels per image that the onboard computers analyzed in the late 1980s.

Unidentified driving object

In the final days of 1990, with Patrick Stelmaszyk acting as an intermediary, a formal contract was signed between ITMI and the Mazda group. “The goal was to work alongside Mazda’s local teams to study and design a computer-aided artificial vision system leading to the development of a self-driving car prototype,” explains Patrick Stelmaszyk. The project may have seemed a bit crazy at the time, yet back then, Mazda manufactured 10 million cars per year and exported them all over the world. The group even opened a plant in the United States. “Mazda was driven by a genuine desire to innovate and set itself apart from the competition. With a clear vision for the future, it turned to onboard computer systems. The idea to create a driverless car was developed in this context,” continues Patrick Stelmaszyk. The French engineer was now working at the Mazda plant in Yokohama with a small team of fifteen people. “It was still a traditional world with very different cultural norms and customs from our own. I had to learn everything from scratch, starting with the Japanese language. Almost none of the engineers spoke English,” recalls Patrick Stelmaszyck.

The driverless car emerged in California in the early 2010s with the Google Car. But the technology was developed by ITMI and Mazda back in the 1980s. That’s where the concept was born.Patrick Stelmaszyck

A promising outlook

He worked with his Japanese colleagues for nearly a year to develop an effective artificial vision system. “This system is an essential prerequisite for creating a driverless car. Its purpose is to automatically identify obstacles on the road and adjust the car’s behavior accordingly. The cameras and software we developed at the Yokohama plant converted images into pixels that were then analyzed by the onboard computer system. The pixel density made it possible to define the type of obstacle. It was an extremely innovative technology for its time.”

Innovative, but impossible to use. Due to the limited processing speed of computers at the time, it took a few long minutes to analyze the pixels. That wait time equated to insurmountable safety concerns and prevented any mass development. “It’s a wonderful adventure but we were too early. The technology wasn’t ready yet, and concrete developments came much later,” concludes Patrick Stelmaszyck. He returned to France in 1990. But the collaboration between ITMI and Mazda continued for almost a year.

Fichier 1

Thanks to rotate your device.