PHOTO: MIKKEL FLY KRAGH
PHOTO: MIKKEL FLY KRAGH
PHOTO: MIKKEL FLY KRAGH PHOTO: MASKINBLADET
AUTONOMOUS DRIVING ▶▶▶
The future: Unmanned machines, but not for some time
BY KATE FRANK POMMER S
elf-driving cars and agricultural ma- chinery are the future of precision farming. This is what Mikkel Fly Kragh of Aarhus University discovered when
writing his recently defended PhD thesis on autonomous tractors. The focus in his research was at greater safety through 3D detection. “It was a fairly random decision to choose an ag- riculture-related subject, but I ended up feel- ing really passionate about making it possible to optimise farming with the help of robotics,” says Mikkel Fly Kragh. The research he worked on for almost four years, is about how to improve the safety of autonomous tractors that use LiDAR technolo- gy. This equipment measures the distance be- tween points using laser light. The system in question consists of several lasers that con- stantly measure the distance to surrounding objects by scanning 360 degrees horizontally, ten times a second. Each rotation creates a 3D point cloud containing 70,000 distance measurements. Kragh has developed analysis methods for this system.
Many small streams Deep-learning algorithm technology needs to
Self-driving tractors are almost here, but legislation is needed first that allows them to work without direct supervision. Researcher Mikkel Fly Kragh shares some of his ideas on the future of autonomous driving.
be good enough for the robot to be able to tell the difference between a football goal and a small child. So it knows that it can drive close to the goal without any problems, but it must be careful close to a child. “It’s simply not good enough for the robots to keep stopping all the time,” he says, and explains: “We need to be able to prove that the robots can make smarter decisions.” The system is more than just lasers. There are lots of equipment located on the tractor – vari- ous sensors, comprising a normal camera, Li- DAR lasers, a thermal camera, a colour camera and a stereo camera, itself made up of two depth-vision cameras. “The conclusion was partly that it’s always best to use lots of sen- sors, as it’s more accurate, and especially useful if the sun is shining into one of the cameras or there’s an extreme amount of dust in the air,” he explains. Snow and extreme rain can also be a problem because the lasers hit the water parti- cles and this produces some incorrect data.
Smaller machines Existing legislation is an obstacle since it pro- hibits driverless vehicles from using public roads. The machinery used on a farm would also benefit from a change in the law. “If there are going to be self-driving agricultural ma- chines, then they will need to be smaller and thus more of them will be required. That’s also an advantage, given that large and heavy ma- chines compact the soil in an undesirable way,” says the researcher. And although the ques- tion, “when will self-driving tractors and com- bine harvesters appear on the fields?” has been answered with “soon” for quite a few years now, Mikkel Fly Kragh is more sceptical. “I be- lieve that self-driving cars will come first. The car industry is much bigger and leagues ahead. But after that, things will progress very quickly with agricultural machinery. So perhaps in 10 years,” he says. From a technological perspective, it would be possible to start tomorrow but legislation and
The Lidar sensor can detect whether an ob- stacle is a human being to keep a good dis- tance to, or whether it is an object the ma- chine can approach much closer.
42
Here all the sensors have been placed on an ordinary tractor during the testing of the sys- tem.
▶ FUTURE FARMING | 22 February 2019
The collection of different sensors. Using sev- eral parallel inputs is always better, in case one system experiences trouble in certain conditions.
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52
