Published on April 24th, 2019 | by Steve Hanley0
Apple & Cornell Offer Alternatives To Lidar For Autonomous Cars
April 24th, 2019 by Steve Hanley
Lidar is a technology similar to radar that uses light instead of radio waves to gather information about the surrounding environment. Every company in the world that is working on autonomous cars — except Tesla — uses lidar as part of the sensor package that feeds data to the car’s self-driving computer.
Lidar sensors are big, bulky, expensive, and unsightly. Not only that, they do a poor job in snow, sleet, hail, smoke, and smog. If you can’t see the road ahead, neither can lidar. That last part is one of the reasons Elon Musk refuses to incorporate lidar sensors into the self-driving hardware package for Tesla automobiles.
Musk is “full of crap,” says Scott Miller, head of autonomous driving technology at General Motors. That may be so, but Tesla scheduled a full blown Autonomy Day this week. General Motors did not. It’s possible that Scott Miller is the one who is really “full of crap.”
Stereo Cameras Almost As Effective As Lidar
As the battle over lidar rages, researchers at Cornell say they have created a system of cameras that is almost as good as lidar at creating three dimensional images. Many camera systems rely on a single camera mounted in the center of the windshield in front of the rear view mirror. A single camera cannot accurately provide depth perception. But add a second camera and put each one in the top outside corner of the windshield and you have a system that does just that.
The researchers say their new system has triple the accuracy of a single camera system, making it almost as good as lidar. Cameras today cost a few dollars. A single lidar sensor costs hundreds or even thousands of dollars and every self-driving car needs several of them. One of the factors that led to the death of a pedestrian by an Uber test vehicle last year was the company was trying to decrease the number of lidar sensors in its self-driving sensor package to reduce costs.
“One of the essential problems in self-driving cars is to identify objects around them — obviously that’s crucial for a car to navigate its environment,” Kilian Weinberger, associate professor of computer science at Cornell and senior author of the paper, tells Science Daily. “The common belief is that you couldn’t make self-driving cars without LiDARs,” Weinberger said. “We’ve shown, at least in principle, that it’s possible.”
He believes stereo cameras could be used as the primary means of identifying objects in lower cost cars. It could also be employed as a backup method in higher end cars equipped with lidar.
[A brief aside: there is debate about how to abbreviate the phrase Light Detection and Ranging. Some call it LIDAR. Some call it LiDAR. Some call it Lidar. 50 years ago, the same debate applied to radar but eventually the all lower case version become commonly accepted. In the future, lidar will become the preferred choice. Why not just leap ahead and start using the acronym as it will be commonly known a few years from now? Those who insist LIDAR, LiDAR, or Lidar is correct are fighting a losing battle over semantics that has no relevance to the value of the technology.]
For more on the Cornell stereo camera research, watch this YouTube video.
Apple Files Patent For New Sensor Technology
Apple is still messing around with its Project Titan program. Is the company intent on building its own electric cars or is it creating technology it can market to other companies to install in their own vehicles? Nobody seems to know. From all the backing and forthing involving Project Titan over the past several years, even Tim Cook himself may not know for sure.
What is known is that in 2018 Apple filed a patent request centered on adding a variety of sensors to the head and tail light assemblies of automobiles. The sensors could be cameras, lidar units, temperature or humidity monitors, infrared receptors, wind speed and direction detectors, or GPS units.
Extending the Cornell research a bit, why couldn’t the stereo cameras be mounted in the headlight units instead of the windshield? One possible answer is that mounting the cameras higher off the ground yields better 3D imaging. Another is that defrosters and wipers can clear a path for cameras mounted on the windshield easier than adding heating and wiping equipment to headlights.
Autonomy & Society
Autonomous cars are everywhere in the news these days. Elon Musk has said that within a few years, any Tesla owner who cares to can make $30,000 a year by using the car as part of his company’s robotaxi fleet. Two questions arise. What Tesla owner of sound mind is going to allow strangers to vomit, fornicate, or do other antisocial things in the car? If everyone is renting out the family bus to strangers, won’t the market quickly reach a saturation point that will diminish the financial rewards?
While the CleanTechnica community and most of our writers are gaga over autonomous vehicles, I take a contrarian view. If the ultimate objective is to drastically reduce global climate emissions, what good does it do to build more and more gigantic factories to manufacture more and more vehicles? Shouldn’t we be focusing instead on creating a circular economy that is locally based and reduces the need for vehicles?
From what I have read, the biggest impetus for getting self-driving cars on the road is so the companies that own them can make enormous profits in the transportation as a service industry. Uber and Lyft drivers could make a decent income some years ago, but today their compensation has dropped to the point where many struggle to meet their expenses.
It’s fine and dandy that better self-driving technology is coming, but if it encourages more travel rather than less, what is the point? Just my opinion and worth precisely what you paid for it.