Self-driving cars are veering closer to the roads.
Torc Technologies, which collaborated with Virginia Tech on an autonomous Ford Escape Hybrid that placed third in DARPA's Urban Grand Challenge, started selling a version of its self-driving creation Tuesday.
Called the ByWire XGV, Torc's drive-by-wire converted Ford Escape Hybrid will be available as a research platform for academics and car industry types who are interested in developing new applications in the field of autonomous autos.
While the world may not be ready yet for self-driving cars on the road, someone could have a lot of fun on a closed-course race track. Torc's car will sell for about $60,000, excluding the cost of the Ford Escape Hybrid with four-wheel drive. The starting sticker price on the SUV with 4WD is about $29,000. (As a platform, the car is ideal for U.S. research because it's American-made, has a powerful battery onboard, and is sturdy over rough terrain.)
"It's an experimental ground vehicle that provides researchers and developers with a car that's already converted," said Anu Saha, robotics product engineer with National Instruments, whose technology is used in the car. "There is a base level of technology, and researchers can come up with the next level, like what's stopping us from having autonomous cars on the road? We need more intelligence."
Torc is among the few contestants in DARPA's autonomous road races from 2004 to 2007 that are turning their hard work into a product. Team Dad commercialized an advanced laser sensor that it developed for the 2004 race, and then sold it to as many as 12 semifinalists in the 2007 event. The so-called lidar, which costs about $80,000, helps create a highly detailed map of the surrounding terrain so an autonomous car can more easily detect and avoid obstacles. Other teams have tried to sell their cars for research purposes, too, but on a more informal basis.
The auto industry is moving closer all the time to adopting these technologies. Infiniti has a new car that features "lane departure control," which is capable of keeping itself between lane lines. It will use the brakes to kick itself back on course if it veers from a lane, according to Car and Driver.
Torc announced other new autonomous vehicle technologies that it will sell to the military.
The Blacksburg, Va.-based company has developed components that would convert a vehicle into one that could act autonomously. It developed hardware that contains the software algorithms for obstacle detection and avoidance, and route planning. The box is Joint Architecture for Unmanned Systems (JAUS) interoperable, meaning that it's compliant with a U.S. Department of Defense standard for syncing manned and unmanned systems. Also, Torc built the SafeStop, an emergency stop system for unmanned vehicles to get out of trouble. That's also in a hardware box that can be plugged into a vehicle.
So far, Torc has demonstrated how the technology works with the iRobot PackBot, an unmanned ground system used in Iraq for defusing explosive devices. Soldiers typically control the PackBot with a remote. But with Torc's system, called Autononav, they can use a monocular to mark a target for the robot to reach 100 meters away, for example; and the robot will drive itself there.
"The soldier doesn't have to worry about driving around ditches. It figures it out. We call that site and click autonomy," said Torc CEO Michael Fleming.
Fleming said that by mid-2009, a small number of the systems will be evaluated in Iraq. Although the system is newly available for purchase, he would not say how much it costs. It's sold only under military contract, he said.
But the technology can easily be ported to a variety of markets such as mining, industrial automation, and farming, Fleming said. Carnegie Mellon University's autonomous vehicle team has worked with Caterpillar on mining applications, for example.
"As an industry, we need to focus on taking these to commercial markets, such as mining automation, automated highways, farming," Saha said. "There's a cultural gap that needs to be filled, where users understand that in order to fill that void we need to continue to conduct experimentation."