Firmware

Upcoming Tesla Autopilot 1.01 Improved through Fleet Learning

Following a week of overblown Autopilot ‘fail’ videos making its way through media, Tesla CEO Elon Musk takes to Twitter to announce that an upcoming Autopilot “1.01” will improve upon the vehicle’s adaptation to curve speed and lane holding.

Tesla has always explicitly stated that the initial launch of Autopilot was still in a beta-testing period and that responsibility for maintaining control of the vehicle remains on the driver. But despite the warnings, overzealous Model S owners took to YouTube to show how their 1.0 versions of Autopilot can sometimes misjudge lane markings.

Tesla has been capturing real-time driving and road data from its Autopilot drivers through the vehicle’s array of onboard cameras, radars and sonar equipment. Tesla refers to this ability to crowd-source data as their “fleet learning network” where Model S drivers implicitly share their learnings with one another. The end result is an Autopilot system that’s able to learn and constantly improve itself especially as more data continues to be collected.

Musk’s latest tweet gives us insight on how powerful of a technique this can be. It hasn’t been long since Autopilot was released to Model S and X owners in North America, but the ability for Autopilot to better control speed during curves, and hold lanes with higher precision, seems to have already worked its way into an upcoming update.

Comments
To Top