Advanced Computing in the Age of AI | Friday, March 29, 2024

GM Production Car Gets Robotic Vision 

General Motors vision system could be key to creating the robotic car of the future.

Fans of the science fiction genre have been patiently awaiting the advent of robotic vehicles for over a century. In the 50s, it was easy to imagine that by the dawn of the 21st century, driverless cars would be as commonplace as black-and-white televisions. Who knows? In 50-plus years, we might even have hover-crafts and vertical freeways, the technophiles of the day theorized.

While the timing may be a bit later than anticipated, there are signs that this long-awaited dream technology is starting to come to fruition. Last week, General Motors installed the first camera-based robotic vision system into a production vehicle and invited reporters to check it out. This is an important first step on the road to what may one day become the robotic car of the future.

GM Terrain graphicAn article at Popular Mechanics explains how the robotic vision system uses a single forward-looking camera, with a 37-degree range of vision, to monitor road conditions and surrounding traffic in order to safely navigate obstacles and execute lane changes.

The camera software performs its checks 14 times per second, noting the position of nearby vehicles to determine whether they pose a threat. If an object is becoming larger in the camera's field of view, the system concludes that it is getting closer. The rate at which an object grows indicates how quickly it and the car are moving towards each other. The software also looks out for broken lines, curbs and other objects.

There are other vision systems in use by various car manufacturers. Radar warns the driver of upcoming dangers, and downward-facing cameras can track the lines on the road. But the GM camera system is modeled after the way a human driver would see and respond to changes in the environment. That's what sets it apart.

EnterpriseAI