Emerging from stealth mode, self-driving car start-up Drive.ai not only knows how to program cars to recognize people and everything else in the driving environment -- its cars will also signal their intentions to humans through lights, sounds and movement.
Drive.ai announced today its first commercial foray, a retrofit kit to make fleet vehicles, from delivery trucks to car services, self-driving. The kit includes a sensor array, computer and an LED sign to communicate with pedestrians and other drivers.
Self-driving car technology is being developed by most major automakers, automotive equipment suppliers and start-up companies such as Drive.ai. The intent is to eliminate the 95 percent of fatal car accidents every year that are attributable to human error.
Carol Reiley, President and co-founder of Drive.ai, said the 20 or so people on staff came from Stanford's Artificial Intelligence Lab, and decided to put their PhDs on hold to solve the problems self-driving cars will face. The company focuses on deep learning, sometimes referred to as machine learning, a relatively new area of computer science. With deep learning, rather than responding to a set of specific instructions, computers can extrapolate from prior experiences.
Using deep learning programming, a car's computer can accurately identify what its sensors perceive, then decide how to react to each situation. However, pedestrians in an urban environment can by very unpredictable in their movements, or, as Reiley puts it "people are very dynamic in their decision-making."
To help self-driving cars co-exist with humans, Reiley says that these cars "need to emote what they are trying to do." As a simple example, Reiley points out that a car could include an external signal to show pedestrians and other drivers that it is in self-driving mode.
Further, Drive.ai's cars could use movement or sounds to indicate what they intend to do. Reiley suggested a self-driving car could rock backwards to indicate that it was about to go forward, similar to a runner getting ready to sprint forward. Pedestrians would recognize the movement and, hopefully, not walk out in front of it.
On the road, now
Beyond theoretical, Drive.ai holds a license to test self-driving cars in California, and has already been testing its systems on the roads. Working with delivery companies and ride-sharing services, it will offer a kit to retrofit business fleets with its self-driving technology for its initial commercial venture.
Maintaining a limited scope for initial deployment, Reiley says the cars and delivery trucks will be confined to urban environments with set routes defined by Drive.ai's partners. As the vehicles run their routes, their computers will learn more about how to cope with an urban environment, and share that data with Drive.ai to further improve their systems.
The retrofit kit itself largely relies on camera-based sensors, although it includes LIDAR sensors as well. The actual computer in each car, running a neural network, uses an automotive grade chipset, and is relatively compact. Without offering a specific timeline, Reiley suggests Drive.ai's technology is ready for the road, pointing out that deployment will come as soon as possible.
No baby steps
When it comes to self-driving cars, the Society of Automotive Engineers defines six levels, with zero signifying no automation and five meaning full automation. Current production cars are solidly level two, with some approaching level three. Think of Tesla's Autopilot and pedestrian recognition systems being deployed in some newer cars.
Drive.ai is solidly in the level four and level five camp, according to Reiley. Rather than execute limited automation in the car, and requiring a driver to take over under some circumstances, Drive.ai seeks to eliminate the need for a human driver entirely.
And depending on how well Drive.ai's pilot programs go with its partners, that day may come much sooner than anyone expects.