Autonomous Trucks Self-Driving Trucks Technology

01 The Origin of Autonomous Self-Driving Trucks

Research on autonomous trucks dates back earlier than most people imagine. In 1939, General Motors showcased the world's first autonomous vehicle at the New York World's Fair. From the 1970s, developed countries like the United States, the United Kingdom, and Germany started researching autonomous vehicles, making significant breakthroughs in feasibility and practicality. However, there have been no fully commercialized autonomous vehicles operating on city roads or highways to date.

When discussing autonomous vehicles, the primary focus is often on the private passenger vehicle market or public transportation. However, autonomous vehicles have long been used in other scenarios, such as in industries like open-pit mining, ports, among others, mostly within fenced-off confined areas.

In 2014, SAE International (Society of Automotive Engineers) formulated a set of standards for autonomous vehicles, categorizing them into levels ranging from L0 to L5. Some experts in the industry believe that apart from the SAE levels, there's a need for an additional dimension based on the concept of Geofence, categorizing them into Geo1 to Geo5 levels. As data accumulates and technology improves, the feasible area for autonomous vehicles has expanded from Geo1, limited to mining areas, to Geo5, virtually without any fences, gradually enlarging the area and eventually achieving full autonomy. This classification seems more scientifically grounded.

Companies entering the autonomous driving industry have observed that achieving autonomy in Geo1, a single scenario, might be the first step, gradually transitioning to autonomy in more complex scenarios. Additionally, autonomous driving in low-speed vehicles and limited scenarios might obtain demonstration operation permits from the government or might not require them at all. Therefore, in mining, especially in open-pit mines, not only have single trucks achieved autonomous operation, but fleets of dozens of such vehicles have been operating continuously for months and years.

Why do we need autonomous driving? Firstly, humans need rest, but vehicles can operate continuously. Secondly, working conditions can be extremely harsh, even dangerous, making it unsuitable for humans to work in such environments. Having devices to replace human labor in such conditions is desirable. Thirdly, humans can't be completely focused at all times, leading to the possibility of errors or accidents due to lack of concentration. A device that can maintain 100% focus at all times is desirable.

Trucks, as productivity tools, have one of the most important driving forces for automation - "money". Mining operations are extremely arduous and monotonous, making it challenging to recruit truck drivers and manage transportation costs. In the United States, the annual salary of a driver for an open-pit mine truck is $200,000. If operating in three shifts with breaks, one truck requires a minimum of four drivers, resulting in a wage cost of nearly $800,000 per year for a single truck. If trucks could operate without drivers, the savings in wages alone would be substantial.

02 Working Principle of Autonomous Trucks

Taking Komatsu from Japan as an example, their autonomous trucks mainly rely on the Autonomous Haulage System (AHS). This system uses control devices, GPS satellites, wireless communication technology, and software to replace drivers who used to sit in the driver's seat.

Autonomous trucks in open-pit mines are part of a smart truck dispatch system. Under the overall dispatch of this system, trucks perform the following steps:

• Managed by the vehicle fleet monitoring center's control device equipped with high-precision GPS, each vehicle is assigned the position of a loading machine and a transportation route. The vehicle receives wireless commands and runs at an appropriate speed along the designated route.

• The truck uses GPS, the monitoring center's control device wireless commands, and other guidance devices to determine the precise coordinates of the vehicle in the mine and understand its surroundings. This enables the truck to perform complex loading, transportation, and unloading cycles automatically without human operation.

• During loading, excavators or loading machines equipped with GPS guide the truck to the correct position for automatic loading.

• During unloading, the monitoring center's control device sends the unloading point's location and route information. The truck, guided by the corresponding equipment, reaches the unloading point for precise unloading.

• For safety, under the AHS operation, if the obstacle detection system detects other vehicles or people on the road, the truck immediately reduces speed or stops.

Autonomous trucks are typically equipped with sensor fusion systems primarily based on vision, including cameras, LiDAR, and millimeter-wave radar sensors. Among these, cameras are used for environmental perception, while LiDAR and millimeter-wave radar enhance system redundancy. Using deep learning perception algorithms, these systems can provide real-time environmental perception similar to human eyes, detect and track various objects in the field of view, and interpret visual scenes at the pixel level.

03 Application of Open-Pit Mine Autonomous Trucks

Research on autonomous trucks in mining started in the 1970s, progressing slowly at that time. However, progress accelerated over the past twenty years. Below are the trials and applications of autonomous trucks in open-pit mining.

Komatsu, Japan:

Komatsu's first 77t autonomous truck was initially tested at a quarry of a Japanese cement company, using radar to detect obstacles. It achieved a maximum operating speed of 36km/h and was used in an Australian Queensland mine in 1996. In 1996, Komatsu deployed five autonomous trucks in Western Australia, powered by overhead lines with guide markers every 150m along roads. They employed laser calibration guidance at a rate of 10 times per second and GPS positioning to operate with centimeter-level accuracy within the mine. In 2005, Komatsu began testing the autonomous transport system at a copper mine in Chile. At that time, they didn't develop new trucks but equipped existing models with various sensors, controllers, communication components, etc., to ensure automatic operation under instructions. The entire system required ground control centers, communication stations, and operated with the support of navigation satellites.

By late 2008, Rio Tinto, situated in several iron mines in Australia, also began operating Komatsu's autonomous trucks. Located in the Pilbara region in Northwest Australia, famous for its iron ore deposits, trucks transported iron ore 24/7 across vast red land. These towering trucks, without drivers or accompanying personnel, were remotely controlled by a computer control center 1500 kilometers away in Perth (see Figure 4). To date, the Rio Tinto Group has deployed 73 autonomous trucks across four mines in Australia.

By the end of 2017, Komatsu had over 100 autonomous trucks operating in six mines across Australia, North and South America, including copper, iron, and oil sands mines. They transported over 1.5 billion tons of material. These vehicles and systems have been tested in terms of safety, productivity, environmental resilience, and system flexibility.

Additionally, Komatsu plans to enhance the application capability of the autonomous transport system in mixed vehicle fleets, where both human-driven and autonomous trucks operate together. This gradual transition aims to enable existing mines to evolve into fully automated mines.

In 2016, Komatsu released an autonomous mining truck that completely eliminated the driver's cabin. The new design distributed the weight of the truck evenly across its four wheels and equipped it with four-wheel drive and steering, improving control and maneuverability. This thorough execution of the autonomous driving concept resulted in a 15m-long truck with a payload of 230t, a maximum power of 2700 horsepower, and a top speed of 64km/h.

Volvo:

As early as 2011, Volvo developed technology for unmanned truck fleets and, five years later, completed the world's first cross-border test. The autonomous trucks identified road conditions using the vehicle's LiDAR and GPS technologies, automatically navigating in mining areas without human supervision, following predetermined routes for completely autonomous driving. In June 2016, Volvo Trucks unveiled the world's first underground test of an autonomous mining truck in a Swedish underground mine.

Caterpillar:

In the late 1980s, Caterpillar, a giant in engineering machinery, was actively involved in the field of autonomous mining trucks, researching autonomous dump trucks based on the 135t 785 model. Between 1994 and 1995, Caterpillar equipped two 777 autonomous trucks with scanning radar systems on the front, rear, and sides, detecting personnel and obstacles within 100 meters on the road. This allowed the trucks enough time to decelerate or stop and automatically park when reaching the loading position. Two 777 autonomous trucks were successfully tested in a limestone mine in Texas and showcased at the 1996 MinExpo mining exhibition.

In late 2008, Caterpillar engineers collaborated with well-known university robotics institutes to improve the core technology for integrated mining operations, developing large autonomous trucks. The new control module could drive 20 injectors and switches, monitor over 30 critical engine functions, record data for 100 engine parameters representing engine health, and assist in diagnosing problems.

The Fortescue Metals Group (FMG) in Australia operates in the Pilbara region, a sparsely populated area with extremely high labor costs. In 2013, Fortescue Metals Group signed an agreement with Caterpillar for autonomous trucks at the Solomon iron mine. The initial batch of eight fully automatic 793F mining trucks has grown to 54, with a cumulative transport volume of 240 million tons. Recently, the fleet of autonomous trucks will increase to 59, becoming the world's largest single-mine autonomous fleet. Data monitoring shows that the productivity of the autonomous truck fleet is 20% higher than that of ordinary similar fleets. Additionally, the autonomous truck system reduces human errors, enhancing production safety.