The ‘idea’ of self-driving vehicles has been around since the 1960s. As a tractor-mad child I watched horrified a black and white film of a driverless Massey Ferguson 135 tractor steering its way between the trees of an apple-orchard while it sprayed the blossoms with DDT.
Leaving aside the non-benefit of spraying insecticide on a crop depending upon insects for pollination, it was pointless technology, as the tractor was following a wire buried in the ground and any attempt at using it for a cultivation task was doomed to a rather obvious failure.
Things have moved on since then, and the arrow-straight rows of maize we see in the fields are a result of tractors steering themselves with the aid of GPS, while the human driver just turns the thing around on the headlands and checks the flow of seed and fertilizer. The skill to sow a field with rows so straight you could fire a rifle down them is no longer a priority for farmers, just as handling an Eaton Twin-Splitter gearbox is no longer a sought-after ability for truck drivers.
With gear-changing tasks now handled by microprocessors and relays, are we approaching the time when we can dispense with the services of truck or indeed bus and coach, drivers altogether?
It’s certainly an idea that has no difficulty in attracting investment, although results have been mixed to say the least. As long ago as 2016, Uber subsidiary Otto announced that one of its self-driving trucks had successfully transported 2,000 cases of Budweiser beer 120 miles between Fort Collins and Colorado Springs, Colorado with the truck’s driver in the bunk for the ‘middle mile’ interstate portion of the trip. Uber was delighted: it had paid out over $600 million for the Otto business earlier that year, and it looked like the investment was going to pay off.
Two years later, Uber announced it had ‘shuttered’ the Otto business and decided to concentrate its self-driving efforts on the passenger car business for reasons that were never made clear.
Even in the tech sector, which is notorious for spending investors’ money like water, $600 million is quite a sum to write off in two years!
But it hasn’t stopped others trying: most recently Loadsmith, a North American freight-forwarding start-up, announced it was to buy 800 autonomous driving systems from Kodiac Robotics for delivery by the second half of 2025. It intends to offer capacity-as-a service using driverless trucks to move freight between interstate hubs in the southern states.
Dig a little deeper, though, and it emerges that this isn’t the first such announcement by Loadsmith. It had previously announced it had reserved 350 driverless trucks from another company, TuSimple, with the objective of driverless ‘middle-mile’ operations by the middle of next year. But things clearly haven’t worked out as planned with TuSimple, which has shed more than half its American workforce since the end of last year. Its auditor, KPMG, abandoned TuSimple last year, citing reputational risks, and at time of writing the company had been delisted by Nasdaq, although an appeal is pending.
UK EXPERIENCE
The UK freight industry has been rather less interested in driverless trucks: particularly since the government’s platooning trials unsurprisingly found that even Britain’s motorways were just too crowded and complex for anything approaching the predictable conditions required (see also www.is.gd/ikosici).
The bus industry has perhaps had more luck, even though it operates in more complex conditions than ‘middle-mile’ motorways.
Automation specialist Fusion Processing is a partner in a project to introduce self-driving buses on the CAVForth service. Fusion Processing is also involved with MultiCAV, where a 16-seat autonomous electric midibus is running around Milton Park in Oxfordshire with a safety driver present.
CEO Jim Hutchinson says that part of the exercise is to work out if other road users behave differently in the presence of autonomous vehicles. “There is clear guidance in the Highway Code about how drivers should treat vulnerable road-users, and this is programmed into our system,” he says. “Fusion Processing was involved with the Gateway Project in Greenwich, London, where small people-carrying pods operated on an unsegregated ‘shared path’. There was certainly an issue there of people deliberately walking in front of the pods to see what would happen!
“The system can anticipate behaviours…for instance if it sees pedestrians on the pavement it anticipates they may step into the road and adjusts speed accordingly. Some of the more advanced means of predicting behaviour are interesting, but you have to get it right all the time.”
We are already seeing partial automation of driving on trucks with driver aids such as AEBS, which alerts drivers to obstacles and will initiate emergency braking if required. “When you look at partial automation, the driver needs the sensors at moments when their performance is worst, such as poor visibility,” Hutchinson points out.
DESIGN CONSIDERATIONS
He adds: “So we need the right sensor set-ups, and it is crucial how the data is processed. The systems currently in use are too crude. In autonomy, you need redundancy; if you lose a vital part of the system you need enough left to do a minimum risk manoeuvre to safety and then terminate the journey. This requires a good level of redundancy in sensors and primary and secondary processing. Secondary redundancy is also needed in the brake and steering control.”
With electrification, the power requirements of the system are also a consideration. “We design our own components and the range of our electric vehicles is not significantly impacted by automation. Power is drawn by sensors and processors, but there is no power-sapping cooling system for the vehicle’s ‘brain’.”
Although currently more heavily involved in the passenger sector, Fusion Processing is also working with newcomer hydrogen power specialist HVS in the freight market, but this work is not nearly so far advanced.
Technology which once may have been developed with autonomy in mind is increasingly finding application as an aid to human drivers, and Emily Hardy, UK marketing manager at Brigade Electronics, points out that systems such as Sidescan Predict and Radar Predict are there to gather information for the human driver and not to control the vehicle.
“There’s a lot of fear and misinformation about AI because some of our systems use an element of machine vision, but the systems don’t connect to the vehicle’s CAN-Bus, so can’t control it.
“Sidescan Predict uses ultrasound with intelligence, so it can distinguish between roadside street furniture and a vulnerable road user. This stops it sounding false alerts to the driver. It takes the image and looks for patterns, which enable it to ‘know’ what a human pedestrian or cyclist looks like. Other systems such as driver drowsiness detection and side-swipe protection need AI, but they are driver aids, not driverless technology.”
BOX: AUTOMATED IVECO NOW ON PUBLIC ROADS IN GERMANY
An IVECO S-Way truck equipped with PlusDrive, an enhanced driver-assist solution based on the autonomy platform of tech company Plus, began operating on public roads in Germany in July.
It is equipped with AI-based autonomous driving software and advanced sensors including Lidar, radar and cameras.
Public road testing will continue in Germany over the next several months and expand to Austria, France, Italy, and Switzerland, so that real-world data can be collected from the diverse roadways and driving conditions across the region. The testing is also said to demonstrate that PlusDrive can safely and automatically handle driver-supervised manoeuvres, including lane centring, cut-ins, lane merges, driver- initiated or system-suggested lane changes, traffic jam assist and nudging.
A promotional video is available via www.is.gd/jinare.