Can We Trust AI? by Rama Chellappa

Author:Rama Chellappa
Language: eng
Format: epub
Publisher: Johns Hopkins University Press
Published: 2022-02-15T00:00:00+00:00

CHAPTER 4

The Promise of Autonomous Vehicles

ERNST DICKMANNS REMEMBERED sitting inside a van as it accelerated through empty stretches of the autobahn in his native Bavaria. Dickmanns, a former German spacecraft engineer turned professor, had been recruited by German carmaker Daimler-Benz in 1986 to build a self-driving vision-guided vehicle that could eventually be sold to the public. “It was a strange feeling the first time I got into [it] . . . it was a five-ton van,” Dickmanns told an interviewer for an oral history project in 2010.1 “And when this accelerates like a human driver, really, with all its power it has, for the first 10 minutes, you are very anxious [about] what’s going to happen, and then you see, well, [it] behaves like a human behaves, and you get accustomed to it.”

The first phase of Dickmanns’s project was focused on transforming the boxy Mercedes van into a rolling scientific workshop. The professor and his team of engineers crammed it with video cameras, range-finding sensors, and state-of-the-art computers that would control the van’s steering wheel, gas pedal, and brakes. Software translated the video image data into driving commands using an Intel 8086 16-bit computer chip that was unable to capture the entire image of the road in front of the van. This forced the engineers to devise a method in which they divided the video screen into 12 boxes, and were able to load enough pixels to get several of the boxes filled with the image. As each frame of the video moved along, the team members were able to switch to a different set of windows, giving a spotty, but ultimately useful view of the road to the computer driving the vehicle. Despite these obstacles, Dickmanns’s engineering achievement was remarkable: creating the equivalent of the Model T for autonomous vehicles.

Instead of relying on electronic signals embedded in the highway that would communicate with the vehicle as some previous efforts had done, his team of engineers from the Bundeswehr University Munich designed a computer vision program that could rapidly identify obstacles using video cameras and translate the data into driving commands (the US Army was also developing a computer vision program for a robotic truck at around this time). Dickmanns’s team continued to progress on its little-known Prometheus project, which culminated in a stunning public demonstration in October 1994 when they took the self-driving technology they’d developed from the van and installed it in a pair of S-Class Mercedes SEL 500 sedans. On a sunny, clear morning, the engineers picked up a group of VIPs from Charles de Gaulle Airport in Paris, drove them to Autoroute 1, and switched the two cars into self-driving mode on the highway. Over the next few days, the robot cars drove in traffic at speeds up to 130 kilometers (81 miles) per hour, changing lanes and settling into a convoy of sorts through the outskirts of Paris. At least one engineer kept his hands on the steering wheel in case something went wrong—but the cars were doing the driving.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.