At this year’s Consumer Electronics Show, John Deere introduced a 40,000-pound self-driving tractor that will be commercially available by the end of 2022, adding additional muscle to the self-driving vehicle sector.
The system steers a Deere 8R tractor with a chisel plough and the capacity to tow additional equipment using six pairs of stereo cameras and GPS direction. A farmer may start the tractor with a swipe of a smartphone app and then walk away to spend time with family or take care of other business, using the app to track the tractor’s progress ploughing a field or completing other tasks — and get warnings if the programme detects irregularities. The tractor can collect data on the health of the crops in the field, the health and moisture content of the soil, and other parameters while it’s operating.
Deere & Company’s chief technology officer, Jahmy J. Hindman, hailed the news as a watershed moment for agricultural production. “Until recently, agriculture was all about doing more with more — more horsepower, more inputs, more acres — but the new digital age is altering everything.” “Over the past decade, we’ve focused on doing more with less and giving farmers more tools to make better choices,” he added.
For the last few years, self-driving vehicles have been a mainstay at CES, and this year’s event included exhibitors selling LiDAR (light detection and ranging) sensors for autonomous driving and driver support systems. The Deere tractor, on the other hand, does not employ LiDAR, and the company’s researchers were unable to adapt hardware or software from the self-driving automobile sector to their use in general. In terms of vibration, temperature, dust, and other problems, a tractor is fundamentally different from a vehicle as a computer environment. According to him, Deere designed the stereo cameras, for example.
Deere, according to Hindman, may start with certain off-the-shelf components but must always tailor them for its purposes. The tractor, for example, employs Nvidia Jetson Xavier GPUs with a bespoke assembly for passive cooling in a dusty environment where traditional computer fans would be ineffective.
“As far as software goes, that’s all us and all-new,” he said.
Deere paid $305 million acquiring Blue River Technology in 2017 to put itself in a position to investigate AI applications in agriculture.
Deere’s See and Spray herbicide application technology, which Deere claims may cut the quantity of chemicals sprayed on a field by up to 80%, saving farmers money and supporting more sustainable agriculture, was developed using Blue River’s technology. See and Spray, which uses computer vision and stereo cameras to discern between plants and weeds when a sprayer passes over them, ensures that pesticides are only sprayed on the weeds.
While Deere’s application is unique, it does take use of more general breakthroughs in AI, such as common topologies for deep neural networks, according to Willy Pell, Blue River’s vice president of autonomy and new ventures. He estimates that the Deere subsidiary employs roughly 30 workers.
Deere’s engineers loaded prototype tractors with a variety of sensors and drove them over fields, recording data to determine which ones offered the most relevant information, according to Pell. He said, “We settled on stereo cameras, which was counterintuitive since we believed we’d be doing LiDAR.” Lidar is great for identifying other cars on the road ahead, as well as their distance, direction, and velocity, but it’s less useful for a tractor driving across acres of open farmland.
Stereo vision cameras, on the other hand, provide good depth perception, and Deere’s software can split the image into multiple views — the raw image, a depth map, and per-pixel classification of each part of the image that distinguishes between ground and sky, as well as a crop and a foreign object that may have fallen onto the field.
The tractor is pre-programmed to deal with typical scenarios, such as navigating around animals crossing a field, but it also has an anomaly detecting system to assist it deal with the unexpected. If it identifies an item in its route that isn’t in its training data, for example, it simply pauses and tells the farmer to take action.
Deere stressed the attention required to ensure that the equipment can work independently while being safe.
This is one instance when a self-driving tractor outperforms a self-driving vehicle. “We stop when we see an unusual thing because we don’t have to worry about being rear-ended by another motorist,” Deanna Kovar, Deere’s vice president of production and precision farm systems, said.
Subtly charming pop culture geek. Amateur analyst. Freelance tv buff. Coffee lover