The project is the world’s first use of Level 4 automation on a scheduled bus route, according to Jim Hutchinson, CEO of system supplier Fusion Processing, which is working on the project with bus operator Stagecoach East Scotland, bus maker ADL (Alexander Dennis Ltd), Transport Scotland, ESP Group, Edinburgh Napier University and Bristol Robotics Lab in the £4m CAVForth government-funded trial.
There are three fundamental differences between this bus and the common, series-produced single-decker bus, the ADL-built Enviro200. First, it is decked out with a sensor array. Although Hutchinson wouldn’t be drawn on the numbers or exact configuration, he did say that it includes multiple redundant imaging technologies to best detect obstacles and the environment around it in all conditions, including radar, optical cameras, LIDAR (light detecting and ranging), ultrasonic detection and a GPS tracker to monitor the bus’s location in real time.
Second, the bus was modified to fit electrical actuators in place of the standard mechanical, hydraulic and/or pneumatic linkages connecting the driver’s controls with the bus’s Cummins engine, Voith transmission, Knorr-Bremse pneumatic brakes and the steering mechanism. Converting the driveline components to provide torque and speed inputs to the transmission was relatively straightforward; braking was less so, but still possible thanks to R&D work done by ADL for a previous autonomous emergency braking project, according to Stuart Cottrell, ADL head of advanced engineering. But steering continues to be the most difficult, as no suitable steer-by-wire system currently exists in the heavy vehicle supply chain, he adds. The team continues to evaluate the capabilities of suppliers, and is also considering the installation of a steering safety back-up carried out by another system, such as the brakes.
Third, it is fitted with a Fusion Processing navigation computer, the brain of the system, called CAVstar. This receives the sensor data that provides a picture of what is happening around the bus, and knowing its pre-set destination and current location, it calculates a safe path and trajectory through the environment. Having done so, the system breaks down the plan into lower-level instructions sent to throttle and brake actuators, via the vehicle’s CAN-Bus network.
A COMPLEX ENVIRONMENT
Along the 14-mile journey, the buses will encounter many kinds of different road and traffic environments. Starting from the Ferrytoll Park and Ridepark in Inverkeithing, Fife, the bus travels south, joining a stretch of single-lane carriageway, then moves over the Forth Road Bridge. Over some parts, it will have access to a bus lane. Near its terminus at the Hermiston Gait Retail Park, it will pass through a more built-up urban area to reach the Edinburgh Park tram and rail station. The buses will operate at normal speeds, up to 50 mph in most cases (40mph in the bus lane). And, as the trial is scheduled to last at least a year, they will run night and day, rain or shine.
The relatively high speeds of operation impose a major signal-processing burden on the navigation computer, as well as requiring high-performance sensing to allow for sufficient system reaction time. That, plus the variety of the route and climatic conditions, make the job one of the most complex implementations faced by the company so far, Hutchinson says. He also points out that the 11.8m bus will be the longest non-test vehicle fitted with the system.
As the level of automation is defined as Level 4, drivers are required by law to remain in the cab, where they will have access to a full set of operational controls. In normal conditions, they won’t need them. That includes emergency in-lane stopping; the system will be able to set and maintain a safe distance from the vehicle in front to allow for braking. It also includes dealing with vulnerable road users; in addition to sensing cars, the system will also automatically identify and monitor vulnerable road users (bicycles and pedestrians), and perhaps change the vehicle’s line of travel to allow them extra room, or to slow down. That particular function comes out of existing technology developed by the Fusion Processing several years ago and incorporated into its CycleEye product for buses, installed on First Group’s Bristol Metrobus fleet.
In general, Hutchinson asserts that Fusion Processing will err on the side of caution when the bus encounters a road risk. However, he points out that there can be a trade-off between assuring maximum safety and the efficiency of the journey. For example, while the safest option when entering a roundabout might be to always yield the right of way to other vehicles, doing so might increase the journey time. He says that data gathered during the trial will help inform the company assess how large needs to be the vehicle’s operational safety margin.
But what is clear is that the automated bus control system cannot cope with extraordinary circumstances. If half of a lane of the bridge were closed due to some catastrophe, for example, and police were diverting traffic on to what would normally be the wrong side of the road, the bus driver would need to take over, because the bus could not cope with the change of circumstances. If for some reason the driver did not intervene, the system would pull the bus over to the side of the road and stop.
ROLLOUT
The buses for the project are currently being built by ADL in series production; as they come off the line, they will be moved to a specialised conversion centre to be retrofitted with Fusion Processing components, again by ADL.
Once the buses are fully-equipped, they will undergo testing on a private track and on the route. During these preparations, a new set of cameras will be installed along the route, so partner Transport Scotland can monitor the buses at every stage. Operations are scheduled to begin in summer 2020.
BOX OUT: Other work
Fusion Processing technology is involved in other autonomous vehicle trials. Earlier this year, it installed an automated manoeuvring system at a Stagecoach bus depot in Sharston, Manchester; after the driver left the vehicle, a depot operations manager could take control and send the bus on its way to the wash bay, the refuelling station, and then park it up for the night.
Meanwhile, the Helm UK project, run by TRL, is testing close-running convoys of heavy goods vehicles. Fusion is supplying equipment to monitor the effect of the so-called ‘platoons’ on other traffic.
2018’s Venturer Alliance project investigated the use of autonomous vehicles (cars) on public roads. Fusion Processing supplied technology and participated in research on the handover of control between driver and the autonomous system, as well as the interaction between the system and vulnerable road users. Another project finishing in 2018 saw the firm participate in an autonomous shuttle trial in Greenwich, London, near the O2 entertainment complex.