Posts

Google-driverless-car

VTTI Report: Self-Driving Cars Have Lower Crash Rates Than Conventional Cars

/0 Comments/in , ,

Jennifer van der Kleut

A new study by the Virginia Tech Transportation Institute (VTTI) declares what many have long suspected would be true-that self-driving cars have lower crash rates than conventional, human-driven cars.

Interestingly (but not surprisingly), the VTTI study was commissioned by Google, though the study’s authors say the findings are solely those of the institute.

As to how the study was conducted: “The report examines national crash data and data from naturalistic driving studies that closely monitors the on-road experience of 3,300 vehicles driving more than 34 million vehicle miles, to better estimate existing crash rates, and then compares the results to data from Google’s Self-Driving Car program.”

According to the report, self-driving cars have a rate of 3.2 crashes per million of miles, where traditional human-driven cars have a rate of 4.2 crashes per million miles. The study reportedly adjusted the data for unreported crashes, and takes into account the severity of the accidents.

Industry followers and news outlets alike have been closely following Google’s reporting of around 17 accidents involving its self-driving test cars in Silicon Valley, and just in the past week, the company’s report of the few hundred times a human passenger in one of their test cars had to suddenly take over control of the vehicle for safety reasons.

Early news outlets picking up on the report are already debating the study’s ultimate conclusions.

As Insurance Journal says, “The [study’s] authors note that the data also suggest that conventional vehicles may have higher rates of more severe crashes than self-driving cars, but there is insufficient data to draw this conclusion with strong confidence, given the small overall number of crashes for the self-driving cars.”

In truth, only one of Google’s roughly 17 accidents involved any reported injuries. That crash reportedly involved minor whiplash for the Google employees in the vehicle, and after being evaluated at a hospital, they were cleared to return to work. The driver of the other car also reported minor back and neck pain.

The study’s authors assert that there is “statistically-significant data that suggest less severe events may happen at significantly lower rates for self-driving cars” than conventional vehicles.

Insurance Journal also points out a January 2015 report by University of Michigan and the Sustainable Worldwide Transportation consortium of researchers, which said that “It is not clear that a self-driving vehicle would ever perform more safely than an experienced, middle-aged driver, and during the transition period when conventional and self-driving vehicles would share the road, safety might actually worsen.”

Another point many parties have brought up is that early data is not truly representative of the situation because self-driving cars have not yet been tested in varying weather conditions.

It has been less than two months since Ford began testing its self-driving prototypes in Michigan’s fierce winter weather, including snow and ice, at the Mcity testing grounds. Early findings from those tests have been positive, and Ford executives say the cars have been performing well in such conditions.

 

 

Ford-driverless-test-car-in-snow

Testing Driverless Cars in Snowy Winter Weather - Check!

/0 Comments/in , , ,

Jennifer van der Kleut

It’s an announcement industry followers have been waiting for.

Finally, a company that is heavily invested in autonomous and connected-car technology is putting it to the test in extreme weather.

Ford Motor Co., together with the University of Michigan, announced this week that the partners have been testing the technology in snowy, icy winter weather over the past month, and will continue.

As Forbes points out, one of the factors that makes Michigan an ideal location for testing autonomous cars is the widely varying weather from season to season-that and, of course, the fact that the University has Mcity, its 32-acre testing ground with a fake cityscape, built specifically for testing autonomous and connected-car technology.

One burning question that industry professionals have long been asking is how well autonomous car technology would fare in extreme weather when rain, snow or ice might obstruct cameras and sensors. As WIRED puts it, “Radar and LIDAR do most of the work looking for other cars, pedestrians, and other obstacles, while cameras typically read street signs and lane markers.”

If those systems are obstructed, one could find himself in a dangerous situation. This is why many are eager to hear how Ford’s tests are going.

Jim McBride, Ford’s head of autonomous research, told WIRED that Ford creates a high-fidelity, 3D map of the area its test car is going to travel before a test drive. This form of “self-locating” helps its cars compensate in inclement weather conditions.

According to McBride, “Those maps include details like the exact position of the curbs and lane lines, trees and signs, along with local speed limits and other relevant rules. The more a car knows about an area, the more it can focus its sensors and computing power on detecting temporary obstacles—like people and other vehicles—in real time.”

News like this sheds light on why high-profile deals and partnerships with mapping and navigation companies like TomTom and Nokia’s HERE are such big business right now, and why industry analysts think Google’s acquisition of traffic tracking app Waze a few years ago will prove to be a big boon in the driverless race.

All in all, McBride told WIRED he is very confident Ford’s tests in snow and ice will go well.

“We’re able to drive perfectly well in snow,” he said.

An illustration of Localizing Ground-Penetrating Radar (LGPR) from MIT.

MIT’s Localizing Ground-Penetrating Radar Could Be the Solution to Mapping and Sensing Challenges Facing Autonomous Vehicle Technology

/0 Comments/in ,

Jennifer van der Kleut

Researchers at the Massachusetts Institute of Technology think they have a viable solution to one of the biggest obstacles facing autonomous driving technology.

Mapping and GPS technology is a crucial element for driverless transportation technology. There are millions of miles of roads across the globe - how do we map them all, so that our driverless car always knows how to get us to our destination?

With construction and development constantly changing the face of our cities and roads, updates will also constantly be needed. Plus, adverse weather conditions such as heavy rain, snow and fog can blur lines separating lanes and block sensors and cameras.

“Most [Autonomous Ground Vehicle] sensors cannot determine the vehicle’s location [in] adverse conditions,” MIT researchers wrote in a report released in November of last year that is making the rounds on the Internet again this week.

MIT’s Lincoln Laboratory has developed localizing, ground-penetrating radar (LGPR) in answer to those challenges.

LGPR utilizes a sensor that provides real-time estimates of a vehicle’s location even in challenging weather and road conditions.

“LGPR uses very high-frequency radar reflections of underground features to generate baseline maps and then matches current [ground-penetrating radar] to the baseline maps to predict a vehicle’s location,” MIT’s report explains. “The LGPR uses relatively deep subsurface features as points of reference because they are inherently stable and less susceptible to erosion or damage over time.”

The very high-frequency radio waves can penetrate all kinds of weather, from snow to heavy rain to fog, they added.

Read more about this technology from MIT here.