Much of the push for autonomous trucks comes from the huge potential cost savings. Drivers account for roughly a third of total costs – an issue compounded by the industry’s struggle to recruit and then retain sufficient drivers, the sector’s ageing workforce and the need for high hiring standards. This is illustrated by the American Trucking Associations’ prediction that by 2024, the US driver shortage will increase from 48,000 available driving jobs to 175,000, while here in the UK last year, haulage associations estimated that there was a shortfall of 45,000-60,000 drivers. And while a report commissioned by the Freight Transport Association says there are signs that the shortage has diminished, it also highlights the fact that 10 per cent of the driver workforce in the UK are EU nationals – a potential issue should a hard Brexit trigger an exodus from the UK.
In addition to direct savings from automation, there is the potential to save roughly 20-35 per cent in fuel costs. There is also safety to consider, given that roughly 94 per cent of accidents on the road are the result of human error.
Some of the benefits are expected to materialise long before completely driverless trucks hit the roads, given that partial automation might make trucking more attractive as a career, enable drivers to get more done while they are on the road and help reduce crashes caused by driver error. One important threshold is the point where autonomous systems can pilot a truck long enough for a driver to sleep without having to stop and find a berth to park, especially given regulations that limit how long drivers can stay on duty.
The fact that some benefits are expected to occur prior to full autonomy is a good thing given that getting trucks to drive themselves is only one component of a system that must be overhauled if road freight is to take place without regular human intervention. For example, who handles refuelling, deals with tyre blow-outs, assists with unloading and handles relationships with the customer in the absence of a driver? It would be surprising if these issues couldn’t be addressed in time, but the need to change entire support networks and distributed infrastructure suggests resolving them could take a while, to say the least.
Aaron Dale, senior analyst, automotive at IHS Markit, says completely driverless trucks are expected to be a late-stage development. “A lot of [the] initial concepts being shown by Tesla and Mercedes are still centred around the driver being present and playing an active role in monitoring the system when on the road, and certainly the other side of their duties they’d be present for. When fully autonomous trucks happen, the whole logistics of the operation and automating some of the other driver responsibilities as well will absolutely have to be looked at.
“For the time being, we’re seeing a middle stage whereby the driver will be present, and that will help with the public acceptance of the technology as well; almost as you would have on planes, where you might have an autopilot system that is fully capable but the pilot is [present to reassure] passengers – it’s a similar trend.”
With so much at stake, there’s plenty of companies seeking to be first to market with an autonomous truck. Back in May 2015, Daimler made headlines with its Freightliner Inspiration Truck, which was the first licensed autonomous commercial truck to be demonstrated on an open highway in the United States. However, during the journey in Nevada, a driver was in attendance, and the company has been quiet on this front since the announcement – though it was one of the founding members of the 5G Automotive Association, which was formed in September 2016.
Uber-owned Otto (now subsumed into the Uber brand), together with Anheuser-Busch, completed a 120-mile shipment of Budweiser beer without human intervention. A professional truck driver was in the vehicle for the entire route, monitoring the delivery from the sleeper berth.
It’s worth noting that this technology could disrupt more than just the road haulage business. The Volvo Group is currently investigating the use of autonomous vehicles for three different use cases: refuse collection, mining, and sugar cane harvesting.
In the case of the former, the group is working with Swedish waste and recycling specialist Renova. The autonomous refuse truck currently being tested is fitted with a sensor system for identification, navigation and monitoring of the vehicle’s vicinity. The truck follows a pre-programmed route, allowing the driver, who walks ahead of the reversing vehicle, to focus on refuse collection. It also removes the need to climb into and out of the cab every time the truck moves to a new bin. The system also constantly optimises gearchanging, steering and the vehicle’s speed to deliver greater fuel economy and lower emissions. The project is scheduled for completion at the end of the year.
The mining truck use case makes use of much of the same technology. Unveiled last year, the modified and self-driving Volvo FMX is being trialled at the Kristineberg Mine, 100km from Arvidsjaur in northern Sweden, at depths up to 1,320 metres below ground in narrow mining tunnels. The sugar cane harvesting use case is self-steering rather than fully autonomous and is aimed at reducing damage to crops and soil during the harvesting process.
IHS Markit’s Dale says that in addition: “Google through Waymo has been pushing into this technology for a while as part of its general autonomous driving research. Tesla is another one; it’s due to unveil its technology on Semis [trucks] around September time. It is expected to be able to leverage a lot of its development [for] the Model S and the upcoming Model 3 in terms of autonomous driving tech, sensor development and sensor fusion.”
He adds that the level of autonomy being pursued by these companies varies, with Daimler focusing on “platooning and development focused ultimately on saving costs on fuel”, while “the likes of Uber and Google are looking more to replace the driver as a whole in a shorter timeframe”.
The technologies underpinning autonomous trucks are much the same in the passenger sector: radar, LIDAR (Light Detection and Ranging – a remote sensing method that uses a pulsed laser to measure ranges), video cameras supported by analytics, differential GPS (which uses ground-based correction systems to improve accuracy from metres to centimetres), together with other radio systems to enable vehicle to vehicle (V2V), vehicle to infrastructure (V2I), vehicle to pedestrian (V2P) and vehicle to cloud (V2C) communications, all of which fall under the umbrella term of V2X. A range of technologies are being discussed, including a combination of Wi-Fi, DSRC (Dedicated Short-Range Communications) and LTE/5G.
One of the benefits of V2X, could be reconciling the extra stream of input against that coming from the onboard sensors, alerting the driver or auto-pilot to discrepancies that might be caused by edge cases such as the one that led to the first fatality to take place in a Tesla running on autopilot – the car’s sensors failed to recognise a white tractor trailer crossing the highway against a bright sky.
Dale says another benefit of V2X will be the way it will enable vehicles to pre-empt upcoming traffic and weather conditions (particularly important for trucks given their need to brake earlier than cars) and that while 5G is the best solution and will be “a major enhancer of the technology”, intermediary solutions can achieve some of the same benefits. He adds that despite the benefits of V2X, autonomous vehicles need to be designed so that V2X acts as an enhancement “rather than being essential to the driving process”.
“A lot of the sensors have been developed at an increasing rate over the past five years. Technically the capabilities are already there to put such vehicles on the road. There are a few things that are holding it back, the first of which has been cost. While LIDAR systems have been around for a while, typical mechanical scanning LIDAR [costs] around $75,000 for the unit, so it wasn’t at a commercially viable stage.
“Camera radar sensors have been gradually falling in price, but there’s been a big step-change in production for LIDAR systems. We’re starting to see a lot of Flash LIDAR coming out; it’s significantly cheaper – it has about 70 per cent of the resolution seen with some of the early mechanical scanning LIDAR, but it costs about $300 rather than $75,000.”
The other barriers that Dale sees to the use of autonomous trucks today include the public’s awareness and perception of the technology, because while “in some ways [autonomous trucks] represent the best use of the technology because the majority of the driving is on highways and motorways, which is best suited for the technology, people generally are a little bit more fearful of having autonomous trucks on the roads”.
There is also regulation. “A lot of the existing transport legislation in the EU [was written when] there wasn’t any autonomous driving technology. A lot of that is being looked at and rewritten and discussed, but that will take time and is currently holding back deployment of some technologies.
“Concepts and testing [have proven] that the capability is there now. [There’s] a little bit of a way to go on reducing price, but mainly public acceptance and regulation are the key things that have to be addressed.”
He adds: “In the UK, government-wise there’s been a lot of money to start with and a lot of interest in government-initiated testing and establishment of special routes and autonomous driving lanes. That’s something I see as having the potential to open up this market, to get vehicles out there on the public road; it’s a big step in terms of regulation and will massively accelerate the development of these systems.”
Many issues remain to be resolved, but the considerable financial benefits and the industry’s struggle to find a solution to its current driver shortage suggest there is a real need for autonomous trucks, while the fact that the underlying technology is well established and falling in cost is encouraging.
However, with public acceptance being key, there is a clear need for a slow and measured path to deployment – all the more so given terrorist attacks like that in Nice, which have made us all aware of large vehicles’ potential for destruction.