Lyft is developing its own self-driving technology

As Lyft's partners begin picking up passengers and accounting for a portion of the 100 million miles covered monthly by human Lyft drivers, the cars' sensors will be collecting information on their surroundings that will gradually contribute to a centralized repository of data controlled by Lyft, Vincent said.

Lyft is mirroring - whether it wants to admit it or not - Uber's autonomous strategy, though years behind its larger rival and with a shallower bench of talent. "Lyft is also uniquely positioned to build technology in collaboration with partners in a way that makes it possible to roll out self-driving cars at scale in the fastest, safest, most efficient way", Lyft Vice President of Engineering Luc Vincent wrote in a Medium post announcing Lyft's plans.

This is the biggest move Lyft itself has made into the space, and the company will be developing its own technology.

The ride-sharing service announced Friday, July 21, 2017 it will develop the hardware and software to power its own self-driving vehicles, a dramatic departure in strategy. Individual manufacturers may retain the data collected by their own vehicles, but Lyft declined to say whether, for example, Waymo will be able to study data collected by Nutonomy's vehicles and vice versa.

Lyft is accelerating its push into robot taxis, opening a new Palo Alto division where hundreds of engineers will work on autonomous technology. As the Chron notes, recently deposed CEO Travis Kalanick once infamously bragged that the company's self-driving vehicles would allow him to stop having to pay "the dude" in the driver's seat.

More news: Chinese dissident Liu Xiaobo dies, Chinese government says

Automakers without self-driving cars will find an eager partner in Lyft. It's an "open platform", Lyft says, apparently created to help the company win over more partners in the autonomous race. But Lyft is also beginning to pursue a more proprietary approach.

The company is promising always to use a "hybrid network" in which drivers are on hand to assist the robot vehicles - noting that this is going to remain necessary indefinitely in numerous 350 cities worldwide where Lyft operates and where detailed maps are not available.

Building autonomous driving systems is a complicated and expensive endeavor, and brings a new layer of complexity to Lyft.

By harnessing deep learning - a powerful type of AI, which wasn't a proven technology when competitors first pursued autonomous driving - Lyft thinks it can catch up.

"When a passenger requests a ride that a self-driving auto can complete, we may send one to complete the trip".

  • Rita Burton