To most of the world, truck platooning arose in the last few years as part of the larger automated driving wave. But there is a small and dedicated cadre of players who have pursued it for over 20 years! What insights does this history hold for the pending commercial deployment of platooning?
The first truck platooning research started in Europe in 1996 as the CHAUFFEUR project, funded by the European Commission and cost-shared with industry. The first drive-by-wire Class 8 truck tractor was developed under the project which involved truck-makers Daimler, Renault and IVECO, among others. Rather than “platooning” the term of art at the time was “Electronic Towbar” – a term clearly created by engineers, not marketers!
My first experience of truck platooning was in a Mercedes truck in 1998 on a public road in Southern Germany near Konstanz. As I recall, three fully automated CHAUFFEUR trucks were traveling quite close together, around 10m, supported by a circular pattern of infrared emitters on the rear doors of the preceding trailer; this pattern was detected by an IR camera in the follower truck cab which allowed a precise calculation of range.
This approach made sense technically but even then it was well understood that this would not be feasible in the real world, because trailers are frequently switched between tractors and therefore a platooning system cannot depend on specially equipped trailers.
This work was preceded by (mainly) passenger-car-focused automated driving projects in the EU, US, and Japan. The US Department of Transportation (USDoT) Automated Highway Systems programme spanned 1993-98, in partnership with the National Automated Highway System Consortium led by General Motors. I was the guy who had the privilege of leading this exciting programme from the government side. NAHSC activities culminated with “Demo 97” on a closed HOV section of I-15 in San Diego; the use cases demonstrated included platooning operations but focused on cars due to the extensive traffic flow benefits offered by close-following cars.
This would have required dedicated lanes and extensive market penetration but is still seen as viable for the long term when AV cars are common.
Subsequent work by the California PATH programmeme at UC Berkeley, funded by Caltrans, entered the realm of truck platooning, motivated by fuel economy. This work occurred from 2000-2003 with fully automated trucks operated on test tracks. Using only V2V communications and radar, inter-vehicle gaps at highway speeds were quite small, something like 3 meters. Then the German government funded the KONVOI project (2005-2009), in which fully automated platooned trucks were tested on public roads to evaluate traffic interactions. Traveling with a police escort, they ran at 10m inter-vehicle spacing for nine days and accumulated over 3,000km of testing.
The project included the heavy vehicle industry and major trucking firms for stakeholder input. A bit later Japan’s Ministry of Economy, Trade, and Industry got into the truck platooning act with the Energy ITS project (2008-2013), motivated by reductions in emissions as well as fuel economy. Again, good technical performance at highway speeds with small gaps was implemented, with testing occurring on closed courses.
In 2009, a unique concept was pursued by the European SARTRE project (2009-2012): why not have human-driven trucks followed by fully automated platooned cars and trucks? The lead truck driver would handle the main driving tasks while everyone behind could relax. SARTRE was led by Ricardo PLC and included Volvo Cars and Volvo Trucks. Here fuel economy was evaluated on the open road and for the first time cars were platooning behind trucks! The work included public road evaluation plus examination of business case issues.
The US military has pursued automated vehicles since the 1980’s. In 2010, the US Department of Defense (US DoD) got into the platooning game with the Autonomous Mobility Applique System (AMAS) developed by Lockheed Martin. AMAS logged more than 55,000 testing miles during “Extended Warfighter Experiments” at US Army bases. The platooning trucks were operated by soldiers, who appreciated being relieved of the driving task so they could concentrate better on their military mission. These military trucks were platooned at full automation, but for tactical reasons were following at large gaps (i.e. 100+ meters).
In the more recent past, the USDOT Federal Highway Administration (FHWA) launched two Exploratory Advanced Research (EAR) projects. The first, starting in 2013, was led by Auburn University with partners Peloton Technology, Peterbilt Trucks, Meritor-WABCO, and ATA’s American Transportation Research Institute; this work consisted of two-truck Level 1 platoons with emphasis on assessing business case and fuel economy testing. The second, starting in 2014, was led by Caltrans with partners UC-Berkeley, Volvo Group, Cambridge Systematics, LA Metro, Gateway Cities COG, and Transport Canada, with Peloton Technology serving in an advisory role. This work consisted of three-truck Level 1 platoons with emphasis on fuel economy testing, traffic interactions, and driver preferences.
In the last few years, the US Army Tank Automotive Research and Development Engineering Center (TARDEC) has demonstrated cross-border (US-Canada) inter-brand platooning on public highways with 4- to 5-truck platoons, using a combination of vehicles operating at SAE Level 1 with plans to expand to Level 2 and higher automation. These and related activities have served to prove out L1 platooning technically and operationally. I’ve heard that the US Army recently issued a purchase order for platoon capable trucks. In addition, the high profile and very successful European Truck Platooning Challenge in 2016 saw trucks from DAF, Daimler, IVECO, MAN Truck & Bus, Scania and Volvo travel from their company research centres to Rotterdam.
Over the last two decades, the fuel economy benefits of platooning were proven and effective control systems were developed. This research provided a technical baseline but oddly enough did not provide the spark for commercial development.
Getting from research to production systems is always a winding path. During the 1990s and into the 2000s, the USDoT programme managers (myself and my successors) were in touch with the US DoD Unmanned Ground Vehicles programme managers. Building on this earlier work, the DARPA Challenges were launched (attracting those who had worked on previous automated driving research projects from the USA and around the world) and automated driving leapt into the public consciousness. DARPA’s initiatives inspired Google’s self-driving car programme.
This was a vitally important turning point: whilst heretofore a startup company trying to accomplish anything major in the vehicle technology space (i.e. not software) would get little traction with venture capitalists, the Google move created the necessary appetite among investors to launch the self-driving development boom.
Truck OEMs could have initiated product programmes at some early point along this timeline, especially the ones who had led projects in truck platooning. But incumbents don’t feel the urgency that startups do. The business of a truck OEM is in building and selling trucks, which is relatively secure as long as they keep up with their competitors. There was no “forcing function” to bring out new capabilities and no evidence at the time that the highly conservative trucking market was ready for it.
Enter the startups
So far, the sole truck platooning focused startup that I’m aware of is Peloton Technology, which launched in 2013. Although focused on L1 driver-assistive truck platooning, Peloton could have had significant challenges attracting funding if not for the self-driving-car activity of Google and a few others at that time. Now, in the same way that Google tested the waters for and spurred action by passenger car OEMs, Peloton has served to initially test the waters for commercial truck platooning, a topic now of intense interest across the trucking manufacturers and fleets.
After extensive testing of fully automated truck prototypes over the years, why are Peloton and other platooning providers bringing to market “lowly” SAE Level 1 automation systems? In first generation platooning systems, drivers are fully engaged in steering and road monitoring while the system handles only longitudinal control and provides linked safety between the trucks. Most systems coming to market are just for pairs of trucks. Each driver remains in command of their vehicle at all times and the driving experience is similar to using Adaptive Cruise Control— which is on the market today on millions of vehicles.
How uncool compared to the highly automated research era! Where are the robot drivers? The answer lies in the benefits and the risks, plus understanding the timelines of product development as contrasted with research.
Engineering researchers define key research questions and implement technology to answer these questions. It’s true that the previous research prototypes boasted full automation (steering, brakes, throttle) with up to four trucks running at very short inter-vehicle gaps. They pushed the performance envelope immensely, working in the protected environment of testing grounds. This served as a great starting point for product developers.
From a market perspective, the good news is that platooning does NOT require full vehicle control. The benefits of platooning come from linking collision avoidance systems and providing automated longitudinal control, i.e. controlling the accelerator/brakes so as to achieve and safely maintain close headway platooning. Adding steering (which would be more satisfying to the self-driving tech geeks) does very little to increase return-on-investment.
Plus, in bringing product to market, it’s ideal if you can build upon what’s already out there and familiar to the end user. The ability to control accelerator/brakes based on awareness of in-lane road traffic ahead, generally called Adaptive Cruise Control, was first brought to market in the early 2000’s for passenger cars.
It didn’t catch on too quickly, as car customers in dealer showrooms had trouble understanding the benefit (while those who did purchase it quickly learned to love it for lowering the stress of freeway driving.) These systems built on radar-based forward collision warning, which then evolved to safety-focused forward collision avoidance and mitigation (FCAM) systems. Since the mid-2000’s, FCAM systems (also known as Automatic Emergency Braking) have been steadily penetrating both the passenger car and the heavy truck markets. FCAM, developed independently of the platooning use case, provided a superb technology platform upon which to add platooning capability, as the end-users were already familiar with some degree of automatic throttle/brake control.
Thus, for entering the trucking market, where fleet decision-makers were just starting to wrap their heads around the idea of automated driving, it was a godsend that such high benefits could be provided at such a low (and familiar) level of automation. Classic low-hanging fruit.
The tough part, however, was getting from a basic working unit to a robust, high availability, fail-safe and fail-operational, cost effective product that the market would accept. This presented a tall order indeed. Referring to the Gartner Hype Curve, the ‘trough of disillusionment’ that follows peak hype has, in the case of platooning, been focused on detailed engineering and testing within a process permeated by careful functional safety analysis (ISO 26262) and implementation of best safety practices into the final product.
I’m not an expert on functional safety but I have great admiration for those who are. This is a key component of the IP behind safety-critical vehicle technology and it is highly proprietary. While you may not see it spelled out too much in the media, completing a comprehensive safety validation process is a key part of achieving industry acceptance of new vehicle technology and bringing solutions to market.
Back in 2014, the American Trucking Associations’ Technology and Maintenance Council’s Future Truck programme was prescient enough to establish a Task Force on Automated Driving and Platooning as a way to engage the tech developers with the User Community. I was pleased to be asked to lead this Task Force and we’ve had some fantastic discussions leading to two Information Reports. Truckers are rightly skeptical and risk-averse regarding new technologies. At TMC meetings you can hear endless stories of “tech that disappoints.”
But they’re also savvy business people and the fuel economy numbers offered by platooning (on the order of 7% average across each pair of trucks) are highly intriguing. In the truck industry, quote numbers like this and you will turn some heads and raise some eyebrows. Is this for real? Trucks traveling closer than is “safe” based on the human driver paradigm? A significant education process has been underway by all the tech developers regarding the basic tenets of platooning and why it can operate safely. A large portion of industry effort has been to assess the validity of platooning and its practical applications.
Right now they’re in the final stage prior to market introduction: fleet customer trials with near market-ready platooning systems.
If casually browsing the web to keep track of this space, one might conclude that we’re poised to see “Level 4” platooning near term (generally thought of as a human-driven lead truck, followed by one or more driverless trucks). But note in the chart that the higher automation activity is in the Public Research Stream, thus still quite far from market. The Commercial R&D work extends only as far as Level 2 to evaluate benefits of lateral control—mostly in the form of Lane Keeping Assistance for the drivers.
Most importantly, Commercial Product activities are focused solidly on Level 1 for first generation platooning to bring the significant fuel economy and linked safety benefits to market in the near term.
Most truck OEMs are active in bringing Level 1 truck platooning to market. In the US, this includes Navistar, PACCAR, and Volvo. Freightliner has been active but recently announced plans to place greater emphasis on Level 4 driverless standalone trucks or possibly platooning at higher level of automation which will come in later years.
We can see that the research stream starts, works through engineering tradeoffs, then proves/disproves basic technical viability. When commercial players get involved they boil it down to a ‘minimum viable product’ with a business focus and the necessary safety validation to bring the product to market. Meanwhile research continues for more ambitious targets. Across the truck technology industry, is there more going on at high levels of automation than you see in the chart? Very likely. As is true across the entire automated driving space, most of the “real action” is happening behind proprietary walls. As one moves from Research to Commercial to Product, the publicity becomes quite limited until market launch is imminent. If you really want to know what’s actually happening, go work for a tech developer.
Richard Bishop provides intelligent vehicles strategy and perspective to the industry. Seek him out at www.richardbishopconsulting.com