Siemens worked out a way for autonomous car to suck even more fun out of driving by developing a self-driving system that could be fitted to classic cars, such as the 1965 Ford Mustang it ran at the Goodwood Hillclimb this week. It’s too bad for them that the system sucked.
A couple of days ago, Siemens and Cranfield University revealed its supposed self-driving Mustang, and claimed that it had a suite of sensors and algorithms that would allow it to navigate the hillclimb without human input.
Put to practice, this is what actually happened. The hot mess starts about 24 minutes in:
As you can see, the car couldn’t stop yoyo-ing the throttle, and frequently made drastic steering dives towards the hay bales and barriers as it wove back and forth along the route. There were multiple times the driver had to take over and try to keep the car on course, all while one of the developers had to keep up a cheery attitude for the voiceover.
That video is one of two times the car ran up the hill, but it had just as many issues both times it ran.
According to the developer, the car relies on GPS for its navigation, and isn’t actually using what the sensors see of the road ahead to navigate. In theory, it should just be following a map of Goodwood based on a GPS signal tracking its current position, but clearly that wasn’t working out.
Luckily, the Mustang is still fully capable of being driven as a normal car, and the intention behind the development of the system is to figure out a way to help people relax during the “boring” aspects of driving, or when they need to be busy doing something else.
But clearly they still have a lot they need to work out. That run was just as terrifying as it was hilarious.