Autonomous technology well established in mining, agriculture
The auto industry’s biggest current focus is self-driving cars. Established automakers as well as technology companies such as Apple Inc. and Alphabet Inc.’s Google are working to develop cars and light trucks that either assist human drivers or guide themselves.
Heavy vehicle manufacturers have already accomplished that. Caterpillar Inc. (Peoria, IL), with the assistance of Carnegie Mellon’s National Robotics Engineering Center (Pittsburgh), markets autonomous trucks for mining and other applications. Deere & Co.’s John Deere (Moline, IL) was an early developer of autonomous vehicles, in this case tractors and other farm vehicles. Over the years, it has picked up help from NASA.
The catalyst for such products was the need to reduce costs and increase productivity. Self-driving mining trucks, for example, work longer because they don’t have drivers. Autonomous tractors can plant seed and spread fertilizer more precisely, saving money by reducing or eliminating overlap.
Such vehicles also travel designated off-highway routes at mines and fields. That’s simpler than cars self-navigating on the open road.
“The application [agricultural and mining vehicles] is definitely more constrained, even though it is still very complex,” said Herman Herman, director of the National Robotics Engineering Center, part of Carnegie Mellon’s Robotics Institute. The center has worked with both Caterpillar and Deere.
Recent advancements in technology have hastened the continued development of self-driving off-highway vehicles, Herman said.
“From the hardware side, there are many new sensors that are not only more capable, but are also lower cost, enabling a whole new set of applications,” he said in an e-mail interview. “From the software side, there have been many advances in AI [artificial intelligence] algorithms that also enable much better capabilities.”
What’s more, Herman said, “As we design and build more robots, there has been much better understanding on the system engineering perspective on what it takes to build autonomous systems, such as the self-driving truck.”
Automating Mining Operations
Self-driving heavy vehicles go back more than a quarter century. Caterpillar and Carnegie Mellon began working together in the 1980s on ways to automate surface mining equipment.
In the late 1990s, the two partners worked together on a robotic excavator for automatic truck loading. That system used scanning range finders to recognize the truck to be loaded and to sense obstacles in the work area.
“We have demonstrated an autonomous loading system for excavators that is capable of loading trucks with soft material at the speed of expert human operators,” the robotics institute said in a technical paper published in 1998. “The system modifies both its digging and dumping plans based on the settlement of soil as detected by its sensors.” Herman was part of the team that worked on the project.
In the 2000s, Carnegie Mellon and Caterpillar worked together on Defense Advanced Research Projects Agency (DARPA) autonomous vehicle competitions. After that, the duo collaborated on Autonomous Haulage Solutions (AHS), or self-driving mining trucks such as the Cat 793F.
Caterpillar started with three autonomous mining trucks for testing and validation in 2011, according to background material supplied by the heavy equipment maker. Two years later, Caterpillar commercially deployed six autonomous trucks. By the end of this year, the heavy equipment maker expects 100 such trucks to be operating on three continents.
The trucks have no human operators. They have designed routes but can select alternate routes to reach their assigned destinations. They’re deployed at remote mining sites and overseen by a central control system manned by human supervisors. Sensors and related technology enable trucks to avoid other vehicles and mining personnel. The trucks stop if an object or vehicle gets in the way. They also provide diagnostic data and alert the control center if repairs are needed.
Caterpillar continues to be “actively working” with the Carnegie Mellon robotics center, Joe Forcash, engineering supervisor at the company’s Pittsburgh Automation Center, said in an e-mail.
“There is a lot of complexity and technology that go into this solution,” Herman said.
Down on the Farm
Farm fields are also the site of self-driving vehicles, including tractors and combines. John Deere’s vehicles aren’t fully autonomous. Operators still are onboard. They drive the first curved row, in effect teaching it to the tractor’s guidance system. After that, the system does most of the driving and the operator can take his hands off the wheel.
Deere looked at self-driving technology as a way for customers to reduce costs and run equipment longer.
“Imagine driving on an interstate and having to keep your tire on the white line,” said Terry Pickett, manager of advanced engineering for John Deere Integrated Solutions. “Every time you moved off the white line, it cost you $3. You’d be almost bug eyed.
“We really started down a focused path in 1996,” Pickett continued. “We worked with Stanford University on GPS-based guidance.” In the late 1990s, the company worked closely with Stanford on an advanced GPS-based guidance system as well as a Deere system engineered for robustness and cost effectiveness, Pickett said: “We were trying to decide at that point how accurate we needed to be to do operations in the field.”
In the early 2000s, Deere introduced its first AutoTrac system. The goal at the time was to have accuracy within 40 centimeters “and we came in at 25,” Pickett said.
Also during this time, Deere established an agreement with NASA’s Jet Propulsion Laboratory on GPS receivers.
In 2004, Deere modified its StarFire GPS receivers, enabling them to tap into NASA’s ground station network and use NASA software for correcting GPS signals. The company’s license with NASA ended in 2015 after Deere developed its own system.
The NASA pact “got us to market three or four years earlier,” Pickett said. “It allowed us to do a more focused and deeper thought process” while developing its own system, he said.
“It got us from a 25 centimeter [accuracy] system down to a 7 or 8 centimeter system,” he said. “Now it’s 3 centimeter accuracy.”
Besides assisting operators, Deere systems help guide implements for tilling, planting and fertilizing. Deere systems help a tractor and implement follow the same precise path. A tractor pulling a grain cart can get into the right spot for loading grain from a combine, for example.
Deere has built prototypes for fully self-driving farm vehicles but doesn’t expect such vehicles to be deployed soon, Pickett said. “It’s not so much the technology itself,” he said. “It’s how you do the whole job.”
Other companies are looking into totally self-driving tractors. CNH Industrial (London), which sells agricultural equipment under the Case and New Holland brands, and Autonomous Solutions Inc. (Logan, UT), last year unveiled two prototypes at the 2016 Farm Progress Show in Boone, IA.
“Farm equipment right now can’t go driverless,” said Matt Nielsen, marketing director for Autonomous Solutions, which is CNH’s technology partner. “There are some technologies that help with turns, but nothing truly driverless.”
At the same time, “That’s the direction it’s going,” he said. “It’s real easy to get excited about this technology. It’s a magnificent leap. But there’s still a lot more to do.”
The prototypes are equipped with cameras, radar and LIDAR, which uses lasers to sense objects. “That’s what gives the vehicle its eyes,” Nielsen said. Operators can program routes and monitor the vehicle with a laptop or tablet computer. “We don’t have smartphone capability yet; that will be in the future,” he said.
Technology on the self-driving tractor prototypes has been helped by development work for autonomous cars, which has brought down the cost of sensors, Nielsen said.
“We are very grateful to the GMs, Fords and Googles of the world,” he said. “The amounts of money being poured into the technology [have] certainly helped us.”
It’s not known when completely self-driving tractors will be deployed commercially. As with autonomous cars, there likely will be legal liability issues to resolve. And there is the question of cost.
“Technology in its young state is very expensive,” he said. “Early adopters will be your large-scale farms.”
Another area for self-driving heavy vehicles is just off the highway—at loading docks and other facilities.
Eaton Corp. (Dublin, Ireland), a power management company, is developing self-parking devices for heavy-duty commercial trucks. The company refers to it as Advanced Driver Assist System (ADAS).
There are two forms of ADAS. With Dock Assist, a driver backs a truck as close to a loading dock as he can. The driver then applies the service brake, stopping the vehicle in drive. A switch on the device is flipped and the truck slowly begins to move in reverse. Dock Assist engages transmission controls.
The transmission detects the dock based on torque being applied while pushing against the dock. The truck stops and is parked. Dock Assist is known as a “Level 2” system and can’t be operated without someone in the driver’s seat.
Autonomous Self Docking is more elaborate. It includes GPS and finds the dock where the truck is scheduled to park. The system sends commands to the transmission, which adjusts speed and applies the clutch. The system controls the steering wheel without the driver touching it.
“We don’t have a firm launch date,” said Chris Nielsen, global product strategy manager for Eaton’s commercial powertrain group. “It’s safe to say that within a couple of years you’ll see some technology from us to help dock a truck.”
Driving the development is cost. Trucks often get damaged in docking mishaps. “You can do damage to clutches, to infrastructures, to trailers,” he said. Also, “One of the biggest challenges we have industry-wide is drivers. There’s a big shortage out there.
“We’ve been playing around with it for three to four years now to develop what we have,” Nielsen continued. “It’s a little bit of waiting for the industry to catch up. With the surge of interest, we’re starting to ramp up our efforts.”
It’s not just heavy vehicles going autonomous in industrial settings. Seegrid (Coraopolis, PA) is introducing its GP8 Series 6 self-driving pallet truck for loading and unloading pallets. It’s the size of a fork-lift truck and uses five sets of video cameras.
Herman Herman of Carnegie Mellon expects autonomous driving systems to pick up speed in off-highway settings.
“I expect the adoption of self-driving technology to ramp up in many industrial applications before it is widespread in consumer/public use,” he said. “We would love to develop similar solutions for other industries.”