Honda Wants Level 4 Self-Driving Cars by 2025

Honda is the latest automaker to commit to an ambitious self-driving car goal. It wants cars with SAE Level 4 autonomy on the road by 2025, CEO Takahiro Hachigo announced at a media event in Japan.

The SAE scale includes six levels of autonomy, from zero to five. While Level 5 is defined as a car that is designed to operate entirely without a human driver under all circumstances, Level 4 applies to cars that can fully drive themselves, but may only be able to operate autonomously in certain routes or areas. 

Honda is currently testing self-driving cars both at its own R&D center in Japan and at GoMentum Station, a dedicated autonomous-car test facility built on the site of a decommissioned Navy base in the San Francisco Bay Area. However, the company's efforts have been somewhat lower profile, and smaller scale, than some of its competitors' endeavors.

Late last year, Honda and Waymo announced that they were discussing a possible partnership. Waymo is already aligned with Fiat Chrysler Automobiles (FCA), which supplied the Chrysler Pacifica Hybrid minivans that replaced Google-era electric pod cars and assorted Toyota and Lexus hybrids in Waymo's test fleet. Waymo also recently struck a deal with Lyft to collaborate on self-driving cars. 

Honda previously said it would achieve Level 3 autonomy by 2020, so it expects to go from semi-autonomous to basically fully-autonomous cars in five years. While the technology to make a car fully autonomous isn't vastly different from even the tech that backs driver-assist systems like adaptive cruise control, that's still a pretty ambitious goal. While significant resources across multiple companies are being devoted to self-driving cars, replacing human drivers is a complex problem. Automakers and tech companies have proven that cars can drive themselves in principle—but that's different from saying that self-driving cars can handle anything the real world can throw at them, day in and day out, indefinitely.

Honda and its competitors may be capable of putting self-driving cars into production within the next decade. The question is: How good will those cars actually be when they arrive?