Baidu's Apollo program, named after the American space program that took us to the moon, was conceived of as an open source framework that developers could use to help push self-driving car tech forward. The announcement of a second version so soon after the Apollo 1.0, which debuted in July 2017, would lead us to believe that Baidu's plan is working.
The announcement of Apollo 2.0 comes hot on the heels of Baidu's partnership with Canadian firm BlackBerry and the adoption of its QNX system as a critical part of Apollo's architecture. The open nature of Apollo is compelling primarily because it takes a page from Google's handbook, something that Google hasn't done with its self-driving car bid, Waymo. Apollo currently boasts over 165,000 lines of code, with new developers adding more every day.
Some of the highlights of version 2.0 include more HD mapping services, new cheaper sensor package requirements, new reference vehicles, and support for Intel, NXP, Nvidia and Renesas. Another huge update for Apollo 2.0 is that it will allow Apollo 1.0 users to drive autonomously on some simple city roads. One of Baidu's US-based partners, AutonomouStuff, has already updated from 1.0 to 2.0 and has said that the update allowed their test vehicles to operate during both day and night on urban streets.
Part of the reason for Apollo's rapid success is Baidu's 90 partners in the project including Ford, Nvidia, Bosch, TomTom and Hyundai. The access to technology and data that these partners provide is invaluable when it comes to being able to quickly adjust and adapt Apollo to changing technology.
We're looking forward to getting more hands-on experience with Apollo-equipped vehicles as updates continue to roll out and we'll be sure to document our experiences when we do.