Google Self-Driving Cars Cross the Two-Million Mile Mark
By Michael Cheng
When Google announced it wanted to build a self-driving car seven and a half years ago, everyone thought the company was crazy. Now, with other big brands, like Uber, Tesla and Ford, revealing their plans to pioneer the nascent sector, the concept of self-driving vessels overtaking large cities may come to fruition a lot sooner than analysts have initially predicted.
Despite the sudden upswing in competition, Google still has a very large lead in the development of autonomous technology. To show its dominance over the space, the company revealed that its self-driving car program has completed over two million miles of autonomous driving on public roads. This roughly equates to 300 years of human driving, which greatly supersedes the number of driving hours racked up by today's professional drivers.
A Closer Look at the Milestone
Does this mean that Google's iconic "koala cars" are ready for mainstream use? Not quite. According to Dmitri Dolgov, the company's self-driving software lead, the first 90 percent of driving is not that difficult to master. This includes navigating through light traffic, crossing intersections and driving on the freeway. The other 10 percent consists of critical skills that tests the vehicle's sensors and decision-making algorithms. Situations wherein the car must choose between saving the driver and a pedestrian in the event of a potentially fatal collision are included in this very complicated aspect of autonomous driving.
To iron out such kinks in the program, Google has started to incorporate social signals and insights that drivers typically encounter on the open road. For example, when a pedestrian standing at a crossing looks both ways, this could tell the car that the individual is likely preparing to walk to the other side of the street.
"Every day in simulation we re-drive about 3 million miles," explained Dolgov. "Every time we make a change to the system, we can pipe it through all the data we've collected, and evaluate it against that data set. That's kind of a big deal for us."
Quality Experience and Interactions
Driving around in a closed, controlled environment offers very little "real-world" experience for the vehicles. That's why in the past few years, Google has been pushing their self-driving cars on public roads (highways and city streets). In cities participating in self-driving pilot tests, the vessels behave like other cars driven by humans. The company has also rolled out new challenges for its autonomous fleet to tackle. With sensors capable of seeing as far as 200 meters in all angles, the prototypes impressively navigated through an obstacle course filled with clumsy cyclists and pedestrians. During a significant encounter, the car was able to pick up on a cyclist's hand signals (gesturing a left turn) and reacted in a safe, reassuring manner.
"There are miles and then there are miles," said Dolgov. "An even better way to think about it is not just in terms of miles or time, rather it's the number of interactions that you have with the world and richness and complexity of those interactions."
- Applied Intuition Develops Virtual Simulator for Driverless Cars, Completes Series A Round
- US Department of Energy Issues $2M Grant to Develop Autonomous Fueling Stations
- Scientists Develop Hazardous Gas-detecting Drones, Receives Grant from the NSF
- France Partners with Hitachi and Bosch to Launch Autonomous Trains
- Can More Home-based Charging Options Boost EV Adoption?
- Volvo Showcases Autonomous, All-electric 360c Driverless Concept Car
- Fiat Chrysler to Open New Driverless Test Site, Invests $30 Million
- Auto Startup HEVO Wants to Make EV Charging Wireless