Meeting Truck Autonomy Hurdles Head-on

Combining sensors, inter-vehicle communications and controllers poses major challenges in the effort to bring greater levels of automation to commercial trucking.

December 1, 2017

Terry Costlow

Daimler’s autonomous road testing of the Freightliner Inspiration proved some of the automated technologies that will help improve the efficiency of trucks. (Credit: DAIMLER TRUCKS NORTH AMERICA)

Myriad technical challenges and a few non-technical issues face those tasked with creating the systems that let trucks operate without humans to improve efficiency and address the driver shortage. Design teams will have to add vehicle-to-vehicle (V2V) communications and integrate sensors and controllers while keeping an eye on regulatory and insurance hurdles.

Some observers feel that long-haul trucking may be among the early adopters of autonomous technologies, since much commercial driving is on highways, which can be simpler to navigate than urban environments that have more challenges such as bicycles and pedestrians. The efficiency improvements that come when drivers don’t have to constantly control trucks will be attractive to many owners.

TARDEC is using a range of technologies to go from two soldiers per truck to zero. (Credit: TARDEC)

“Some statistics say that with things like traffic jams and inefficiencies, only 35% of the driver’s time is spent driving,” said Derek Rotz, Director of Advanced Engineering at Daimler Trucks North America. “When 29% of the costs go to fuel and 26% to the driver, there’s a lot of potential for improvement. Autonomy is an enabler for increased revenue and increased use of vehicles.”

Trucking industry design teams are leveraging the huge investments being made by the automotive industry, which is also racing to come up with cost-effective sensor suites, vehicle to vehicle/infrastructure (V2X) communications and processing technologies that can swiftly analyze the mass of data coming from multiple sensors.

There’s also substantial government research in driverless trucks. The U.S. Postal Service announced a research program in October, and military branches are also working on driverless vehicles. While the threats are different for military and commercial vehicles, the primary issues of manpower and safety remain the same. The U.S. Army Tank Automotive Research, Development and Engineering Center (TARDEC) recently completed a 50,000-mi (80,500-km) test, driving unmanned vehicles on and off highways.

“One reason we focus on logistics is that we want to let robots drive the fuel and ammunition vehicles and let people drive the water trucks,” said TARDEC’s Robert Sadowski. “Now we have two soldiers in each cab; we want to get that down to none. We will add autonomy slowly, possibly starting with four to seven unmanned units following a manned vehicle.”

Integration fixation

Artificial intelligence can help suppliers like ZF reduce software integration challenges, but there are pitfalls. (Credit: ZF)

Regardless of where safety systems are created, integrating all the inputs and actuators is the most critical aspect of efforts to let vehicles drive themselves. Inputs have to be collected and analyzed, then controllers must make life-and-death decisions in milliseconds. All the elements have to work together flawlessly.

“Integration is the key for doing all aspects of driving,” Rotz said. “In the past, this has been an a la carte industry; we put on the systems customers asked for. Now, with active braking and active lane keeping, the margin for error if you mis-detect something is very slim. The question is, can we support three, four or five brake partners or work with one or two and get a high level of integration?”

Software will be a mainstay of integration efforts. But given the countless things that can happen on roadways, it’s nearly impossible to write code that properly responds to all the possibilities. That’s prompting a lot of research in artificial intelligence (AI) systems that can learn as they encounter different scenarios. While AI can solve a number of problems, its inherent ability to alter activities and analysis make it difficult to test software and validate results.

“One problem for [truck makers] is that they need to know in all circumstances how algorithms will behave,” said Andy Whydell, Vice President, Systems Planning and Strategy at ZF. “In some environments, AI will totally fail. The problem is you can’t predict where that might be.”

That’s not just a technical challenge. AI is among the many factors that will come up in lawsuits likely to follow an accident involving an autonomous truck. Proving that AI and safety systems responded in ways that minimize accidents will impact both litigation and insurance rates for driverless vehicles.

“AI is a big challenge; we’re not always sure how the system will respond,” said Jack Martens, Project Manager of Advanced Technology at DAF Trucks. “I see potential issues when we have to start dealing with insurance companies. Legal issues will also impact the use of platooning — each country has its own legislation. Some cities require at least 50 meters between vehicles.”

Talking trucks

DAF Trucks is using Wi-Fi to link two trucks in a platoon. (Credit: DAF TRUCKS)

Onboard sensors can collect a lot of data, but their range is typically only 200 to 300 m (660 to 980 ft). A lot more data can be collected by using V2X. Its distance, several hundred meters, isn’t the only factor. V2X communications can provide information in blind spots, such as hidden intersections or when several trucks are traveling together.

This capability is being tested to let vehicles platoon together, letting them save fuel by drafting and moving efficiently through stoplights. Some of the Smart City programs now being developed around the globe include ways that let two or more trucks move almost as one, reducing the distance between them without impacting safety.

In the U.S., NHTSA (National Highway Traffic Safety Administration) has focused its efforts on dedicated short-range communications (DSRC), which uses 5.9- GHz communications to let vehicles “talk” to each other. Peloton Technology is using DSRC to let two trucks platoon, reducing a driver’s stress. Currently, there’s a strong focus on quantifying the benefits of platooning.

“We’ve developed a system for certain driving conditions; drivers can have their hands off the wheel and feet off the pedals,” said Joshua Switkes, Peloton Technology’s founder. “Both drivers have to push the platoon button, and the driver of the rear truck can cancel out by stepping on the brakes. The system tracks benefits, logging J1939 messages to show relative improvements between platooning and non-platooning runs.”

DSRC isn’t the only option. NHTSA hasn’t mandated a timetable for implementing DSRC, so alternatives are being examined. Some think that 5G cellular communications can provide many of the benefits of V2X. But 5G is still a technology of the future, so some companies are using a well-proven wireless standard, Wi-Fi, in two-vehicle test programs.

“We use Wi-Fi for platooning,” Martens said. “It only takes 50 milliseconds to send a message from vehicle 1 to vehicle 2. That’s quicker than brake lights.”