Slow Speed Object Detection for Haul Trucks
Caterpillar integrates radar technology with its current camera-based system
Caterpillar has developed the Integrated Object Detections System, a slow speed object detection system for mining haul trucks. Object detection is a system that aids the truck operator’s awareness of their surroundings. The system consists of a color touch screen display along with medium- and short-range radar as well as cameras, harnesses and mounting hardware. It’s integrated into the truck’s Work Area Vision System (WAVS).
Recently Caterpillar conducted some research that revealed some interesting information about their customers and collisions on site. “We surveyed customer and, when looking at safety on mine sites, found that slow speed object detection was high on their priority list,” said Mark Richards, marketing supervisor-large mining trucks, Caterpillar. “Research also showed that 79% of collisions at mine sites involve large trucks. That’s not surprising, considering that the truck fleet represents the majority of the vehicles on site and they tend to move about more than loaders or dozers.” Of the haul truck collisions, 59% involve another piece of mobile equipment, 38% are immobile objects, and 3% people, Richards explained. Most people on mine sites are usually in vehicles, so it’s rare that a person would be standing in the path of a haul truck.
Oftentimes, large truck to large truck collisions do not result in severe damage unless high speeds are involved. “It’s important to note that [most collisions] 69% are at slow speeds around the shovels,” Richards said. “Our slow speed object detection system is designed to cover the high majority of incidents--collisions with objects other than people and at slow speeds.”
The Caterpillar WAVS system, a three-camera system mounted on the truck, has been available for several years. The Integrated Object Detections System is an evolution of that system. The radar alerts operators to objects in their path and the visibility around the truck has been improved by adding a fourth camera. The display is integrated with the truck’s controls. “Previous systems, or after-market systems, might not work with the truck as well as one would like,” Richards said. “It takes time to integrate technology such as this into a machine’s system, but we feel we will have a better product because we have taken the time to do it properly.”
An Added Level of Confidence
Anyone who has spent any time around large haul trucks can easily appreciate that they have blind spots. No matter the make and model, none of them have great visibility to all corners of the truck--that is the nature of large mining equipment. In the pit, the trucks are routinely encountering many different vehicles sizes. Mines have experimented with all sorts of methods (buggy whips, flashing beacons, etc.) to add to the visibility of small vehicles operating among the haul trucks. One could, however, see how easily a haul truck driver could miss smaller vehicles in their immediate area of influence.
The new Cat Integrated Object Detections System allows the operator to be alerted in those circumstances. “When we say it’s a slow speed system, what we are really saying in a lot of ways is it’s a short distance detection system,” Richards said. “It only detects objects up to about 20 meters and then it switches back to the camera system.”
When the operators look at the controls, they can see a camera view (left) as well as the radar projection (right). A green light in the lower lefthand corner indicates that the system is fully functional. A red light or a flashing yellow light would indicate otherwise. The system automatically activates when the operators start the truck, operate at slow speeds and switch directions suddenly. It’s designed to be an intuitive system focused on minimum involvement from the operator.
The system is fairly simple and, according to Richards, does not require a lot of training for the operators to determine what they are seeing. It employs a graduated warning system; as the threat increases, the warning level increases. As soon as the engine starts, it goes through a complete system diagnostic check to make sure everything is in order. If the radar is obscured, it will give an error message. “It’s a pretty smart system,” Richards said. “It will activate when they need it and deactivate at high speeds going. At that point it shifts to a full camera view.”
Medium- and short-range units are mounted on the front and back f the truck. Two short range units are mounted along the side. The WAVS cameras are mounted near them. From where they are mounted, the operators should be able to reach out and clean them. The radar units on the front and back can see 18 m ahead and behind in a cone-shaped area of vision. The side units can see 1.5 meters out on either side.
“We wanted to eliminate the scenarios where a pickup truck pulls behind a haul truck undetected,” Richards said. “If the operator doesn’t know it, it can be a very dangerous situation. This system is optimized for pickup truck size objects or larger. It’s not set up to detect smaller objects on a consistent basis.” Smaller objects are detected with decreased range and coverage area. The system might detect a person, Richards explained, but it’s not designed for it. The geometry and size of the object are very important. Angles and sharp edges are easier to detect with radar.
There are many technologies available for object detection and Cat researchers studied a lot of them. “All of them have advantages and disadvantages,” Richards said. “The performance of ultrasonic technologies is affected by rain, ice, snow, and vibration, and it has a limited range. RFID works quite well, but it requires off-board infrastructure and it does not detect untagged objects. If a new service truck comes on site without an RFID tag, the system would not be able to see it. Infra-red was an immature technology for mining equipment at the time we were evaluating it and it required greater computing resources than the approach we have taken. GPS is making leaps and bounds. Again, it requires off-board infrastructure and it can’t be fully integrated into the truck. A lot of mines sites also have limited GPS coverage and, similar to RFID, other systems without GPS would not be detected.”
Cat elected to go with a radar-based system because of the advantages it offers. “It’s a self-contained system,” Richards said. “That’s very important for our trucks. We can offer it as an option to the dealers and customer and they do not have to go out and buy off-board infrastructure. It’s outfitted correctly from the factory.”
Radar is a robust and mature technology that most people know and understand. The range fits the application needs, Richards explained. “It’s a good value for the capability,” Richards said. “It’s fairly resistant to the effect of rain, snow, and fog. One challenge is that, if the radar gets caked with wet material, it interferes with the system. If the materials dry, it can see through it.” The system has been well-proven at mine sites.
Cat wanted a system custom designed for hauling applications. It had to be able to endure the harsh conditions encountered on the mine sites. “It had to meet customer expectations and what customers expected from Caterpillar,” Richards said. “We wanted to reinforce safe operator behavior. There is no substitute for training. We still need operators that are alert and looking out for danger. This is simply an enhancement.”
The new Cat Integrated Object Detections System is an enhancement that alerts the haul truck operator to immediate threats with minimum intrusiveness. “Minimum intrusiveness mean eliminate constant false alarms,” Richards said. That was one of the biggest issue with the radar-based technology was not detecting things, but detecting the right things. If truck operators think it’s a nuisance, they will find some way to bypass it.
Cat conducted field research on the Integrated Object Detections System throughout 2007. Two systems were installed at mine sites. A total of 20 operators used the system. “Overall, they liked the system,” Richards said. “It incorporates a lot of the features from the WAVS camera system with which they were familiar. The only issue, and the reason we have no commercialized it until now, is that we had to work through this issue of false detection. We now feel like its ready for commercialization.”
System commercialization began in 2008. Cat updated software, components, brackets, and harnesses. “We tested system on four trucks at Caterpillar test facility,” Richards said. “We spent nine weeks testing system. We put prototype systems in the field on nine trucks. We have them operating in Australia, Utah, and Arizona—three trucks at each site. We are trying to get a wide range of experience and we feel like we have done enough to bring it to market and have a commercially viable product.”
Caterpillar has developed a robust system that will meet mining companies expectations—one that would not require a lot maintenance and be easy to use. “We did not want to spend a lot of time training operators,” Richards said. “It’s self-contained to the truck and non-intrusive. Anything that requires a lot of operator input probably is not going to get used. The only issue is false alarms and we have tried to minimize those. Plus the cameras are there. In the case of a false alarm, the operators can look at it and determine that it is in fact a false alarm, switch it off and move on.”
The Integrated Object Detections System will be available in the fourth quarter of 2009. It will be available on current production trucks: 785C, 789C, 793D, and the 797B.
This article was adapted from a presentation Mark Richards, marketing supervisor-large mining trucks, Caterpillar, made at the Haulage & Loading 2009 conference, which was held during May in Phoenix, Ariz.