Tesla Motors announced plans to produce an autopilot feature within three years for its Model S electric car, which currently offers a Linux IVI and telematics system. Tesla’s system falls short of being a fully autonomous car, a la Google’s Linux-based Prius systems, and will only handle control of the car for “90 percent of the miles,” according to CEO Elon Musk.
Just as we were trying to figure out how Elon Musk’s Hyperloop transport works, the Tesla Motors CEO has jumped in with plans for a self-driving car. After a tweet in which Musk said that “Intense effort underway at Tesla to develop a practical autopilot system for Model S,” he explained to reporters that a Tesla self-driving prototype would be completed within three years using Tesla’s own technology. Tesla also posted a job opening an Advanced Driver Assistance Systems Controls Engineer for autonomous cars.
Although Musk did not detail which OS the car’s control computer would run, Tesla’s Model S is already equipped with a Linux-based computer (pictured at right). The system provides access to “just about every aspect of the car’s performance and entertainment system,” according to a July interview in PC World with Tesla CTO JB Straubel. Since Musk said Tesla would use its own in-house technology, this would suggest Linux will form at least part of the autopilot solution.
Tesla’s autopilot will handle driving duty on “90 percent of the miles” a car drives, said Musk, who went on to explain that a fully autonomous car would take longer to develop. The initial autopilot would appear to go farther than most semi-autonomous projects underway at car manufacturers around the world, but not as far as the fully self-driving Prius that Google has been test driving the last few years.
Google’s Ubuntu-based system (shown at right) has already racked up over 500,000 miles of autonomous driving with no accidents caused by the computer, says Google. This impressive record has encouraged California, Florida, and Nevada to approve testing of autonomous vehicles, with certain restrictions.
After signing a deal with Uber this summer that could lead to self-driving taxis based on a future Google design, it was speculated that Google would announce an autonomous car partnership with Continental AG and IBM. However, the German auto components firm, which has been testing its own self-driving car prototype, instead announced plans to do autonomous car research with IBM alone.
Other self-driving projects
Several car manufacturers are also working on self-driving cars. These include Linux-based models under development at GM and Volkswagen that were based on cars that competed in the DARPA autonomous car challenges of the previous decade. Nissan recently set the industry buzzing with announced plans to introduce its first autonomous vehicles by 2020, the same general timetable Google has set for a fully integrated commercial autonomous car. Nissan will offer autonomous functionality across the model range within two vehicle generations.
It’s unclear what technology Nissan will use. It has tapped Windows Embedded Automotive in its Nissan Leaf IVI system, but it’s also a member of the Linux IVI group, the GENIVI Foundation. Nissan is also collaborating on autonomous research with Renault, which has invested in Android-based R-Link IVI systems.
Toyota’s TRINA project for autonomous cars may also be experimenting with Linux. Although the carmaker has collaborated with Microsoft on telematics and IVI in the past, it recently released a Linux-based IVI system in its Toyota Lexus IS, and is a founding member of the Linux Foundation’s Automotive Grade Linux working group. Other carmakers working on self-driving systems, such as Daimler AG, Ford, and Volvo appear to be using Windows, QNX, or other RTOSes, although like Toyota and Nissan, most are keeping their technology under wraps.
Autonomous vs. semi-autonomous
The industry is now actively debating whether cars should be fully autonomous or semi-autonomous, such as more sophisticated adaptive cruise control and self-parking designs. Although there are arguments on both sides for better safety, either should offer the possibility of greatly reducing the over 34,000 automotive deaths per year in the U.S. A fully self-driving car would open up even more litigation and liability issues, and could draw opposition from unions. On the other hand, it would have the advantage of freeing up drivers’ commute time, and conceivably helping mobilize the elderly and handicapped. Full autonomy could also enable more flexible car sharing programs.
On the technology side, prices need to drop, especially for the $50,000-and-up Lidar range-finding system used in fully autonomous cars like Google’s. Also, the industry needs to agree on whether the cars will be independent like Google’s current prototype, or dependent on wireless communications, with so called vehicle-to-vehicle (V2V) technology. One of the leader’s in V2V technology is Australia’s Cohda Wireless, which sells a Linux-based MK2 WAVE-DSRC Radio for V2V.
The car industry appears to prefer semi-autonomous cars like the Tesla model, as opposed to fully self-driving cars like Google’s. Carmakers are concerned about the popularity of Uber, and the already growing trend of shared vehicles displacing car ownership.