Tesla CEO Elon Musk reportedly said the company’s new autopilot software could have prevented the auto crash that killed a man in northern Florida earlier this year.
Advertisement
Moving forwards, Elon Musk says the cars will make greater use of the on-board radar for detecting and avoiding obstacles on the road.
In May 2016 a Model S in Autopilot hit a tractor trailer after the camera sensor didn’t see the white truck against the bright sky. Musk said the parking feature is a “baby step” toward his eventual goal: Letting drivers summon their self-driving, self-charging cars from anywhere using their phones. “It’s really about improving the probability of safety”. “There won’t ever be zero fatalities, there won’t ever be zero injuries”, he said.
More importantly, all Teslas that enable Autopilot will now start using radar to build a three-dimensional map of the world. As the update is done the autopilot will use images from cameras to add on to the radar system. “It will be a dramatic improvement in the safety of the system done entirely through software”.
This comes following months of coverage focused on the safety of Tesla’s Autopilot in the wake of several incidents involving it, including a fatal crash. The Florida accident resulted in the death of Tesla Model S owner Josh Brown in May of this year.
The update could affect the federal investigations surrounding Autopilot.
One of the main challenges of using cameras and radars for a braking system is how to prevent so-called false positives, in which a auto might think an overhead highway sign, for example, is an obstacle to be avoided.
Musk said that, with the latest updates, the system would have identified a “large metal object across the road”.
Advertisement
In another key development, the automatic-steering software has been created to disengage if a driver ignores repeated warnings to keep hands safely on the steering wheel, and to stay disengaged until the driver parks the auto, the blog stated. The blog says, “By comparing several contiguous frames against vehicle velocity and expected path, the auto can tell if something is real and assess the probability of collision”.
