Driverless robotic shuttle pod Harry takes to the streets in Greenwich
Image credit: Louise Murray
A robotic shuttle is being trialled across Greenwich to gauge public reaction to its presence. So what do people think? Is there room on the streets for Harry?
Harry is attracting a lot of attention as he trundles along a two-kilometre riverside route on the Greenwich Peninsula, a finger of land surrounded on three sides by the River Thames in the Royal Borough of Greenwich. His circuit runs between a residential area and the InterContinental O2 hotel, via another conspicuous transport hub – the Emirates Airline cable car that crosses the river – and the O2 concert venue.
Trials of the robotic shuttle over shared pedestrian and cycle pathways in this corner of London are exploring public reaction to vehicles for ‘last mile’ transportation between transport hubs and residential areas. Results of the study will guide the wider rollout of automated vehicle technology in a complex urban environment.
Garbed in an eye-catching peppermint and white livery, the small pod is the focus of many selfies. Harry is a prototype autonomous vehicle, but the one-tonne, four-passenger shuttle won’t be mixing it with lorries on the motorway any time soon. The ultimate purpose of the pod is the so-called first mile/last mile transport concept, and trials are designed to test a future transport-on-demand link between the home and the nearest hub. The hope is to encourage people to use public transport more readily than their own private cars, which remain parked for over 90 per cent of the time.
Harry is part of the GATEway project, an £8m research project led by TRL, the UK transport laboratory, and includes the Royal Borough of Greenwich, Oxbotica, Royal Sun Alliance, Shell and designer of the pod itself – sports car maker Westfield. GATEway has the ambitious goal of demonstrating safe and efficient integration of automated transport systems into real-world city environments.
Harry’s route may appear unchallenging to the untutored eye, but the ever-changing construction hoardings bordering the route, impatient cyclists, dogs on and off lead, joggers, and the great British public strolling obliviously into his path don’t make for an easy journey for the robotic vehicle, which is controlled by Selenium AI software from Oxford Robotics Institute spinoff Oxbotica Ltd. Fed data by a barrage of sensors and cameras around the pod, Selenium must navigate safely through the crowds, some of whom appear to be intent on behaving badly. A mere 300 metres into my ride, Harry pauses because people are not giving him a comfortable safety margin. Small children reach out to touch the trundling robot. A feckless father urges his sub-10-year-old children to jump in front of the vehicle to see if it will stop.
These are precisely the type of public reactions you cannot predict ahead of real-world trials. You cannot legislate for stupidity, but you have to design for it. “We were aware that people would test the vehicle,” says Oxbotica CEO Graeme Smith. “It’s obviously not something that we would recommend. The laws of physics still apply. A one-tonne moving object will still take time to stop, no matter how smart it is.” Prior to trials, the company added an additional laser curtain sensor to the front of the vehicle as a ‘belt and braces’ precaution for just this kind of eventuality. The GATEway team expect the novelty factor to wear off in time.
Under Department for Transport guidelines for autonomous vehicle trials, Harry has a trained safety steward on board monitoring its every move, ready to intervene if necessary, and on each of these early journeys, Harry is also accompanied by stewards in high-vis on electric bicycles. The journey feels like a kind of royal procession complete with outriders. The AI is naturally cautious, watching pedestrians and cyclists as they approach the vehicle via its multitude of sensors, calculating their trajectory and speed. If there is movement towards the vehicle, it brakes and comes to a stop.
Although the wide pedestrian lane alongside the River Thames that Harry trundles along has been segregated with signage, and road-marked ‘pod lane’ with clear signs requesting cyclists and pedestrians to use the other lane, people have other ideas. Such is the interest from the public in Harry’s progress that project engineer Emily Williams, our on-board safety steward for the ride, frequently has to place the robot into a ‘crawl speed’ mode. This is done when Selenium encounters a significant change in environment and the safety steward feels the normal maximum speed setting is too high, e.g. if a group of people are playing football near the pod or a large crowd fails to notice Harry’s approach. This does make for pretty slow progress as dog-walkers and joggers continue along their normal route, oblivious to Harry’s considerate and halting response to their presence.
Speaking to several of the 100 passengers selected from 5,000 applicants for the three-week research period, E&T found that public expectation was consistently higher than what the state of technology can deliver in this real and complex environment.
Everyone spoke about the slow speed of the vehicle, often overtaken by joggers. Oxbotica’s Smith is aware of this, and comments: “The pod does not respond differently to a cyclist, a dog or a pedestrian. It will wait respectfully and work out how to move forward. If we are approaching a gaggle of pedestrians from behind, we need to work out how to politely alert them to the pod’s presence and allow it to move through them; we have to do that or the pod will be slower than walking.”
One objective of GATEway is to help the consortium to learn how to communicate intention of the autonomous vehicle to other road users. For example, a cyclist approaching a robotic vehicle on a shared route needs to understand whether the pod will pull over and stop, and in what direction. Human drivers and cyclists or pedestrians communicate intent just by looking at each other. One aspect of the trial is figuring out a common language of intent so other road users can respond accordingly.
GATEway has not spent its money on a slick and stylish vehicle finish, and the diminutive vehicle in its current form will not be winning any design awards, but there is something of the ‘cute factor’ about the pod. It certainly invited a lot of unwanted touching, which set off Harry’s proximity alarms. Trial participants and Greenwich residents Emma and Roger describe the robot as “like being in a very slow, very small Docklands Light Railway (DLR) train carriage”, or “in a closed-in golf cart”, or “a covered dodgem, and quite a rattly one at that”. Hardly flattering, but indicative of the gulf between public expectation and reality of the build quality of a prototype vehicle.
The fact there was no forward or backward view also seemed to bother passengers. Emma points out that surely one of the benefits that autonomy should offer the redundant driver is the opportunity to look around and enjoy a panoramic view. All agree the experience was positive overall and were keen to use vehicles like Harry in the future and could envisage an on-call service via a phone app for going to the shops or station in place of a private car, or to help people with impaired mobility get around.
Judging from responses of the public to Harry, TRL director Nick Reed’s objectives for the trial have been accomplished. “We wanted to reduce psychological distance between potential users of the system and vehicles themselves. It’s clear watching people interact with it that they see it as quite friendly,” he says.
Trevor Dorling, director of Digital Greenwich, explains the rationale for the council’s involvement in GATEway: “We want to be at the vanguard of innovation and the inevitable changes that autonomy will bring. There is no doubt there will be many positive benefits for city dwellers in reduced congestion and road casualties, and improved air quality. Yet we also want to learn about how cities will need to respond to these changes in terms of urban planning.”
Dorling wants to stress that Greenwich is not at all anti-car, but there are a number of trends already evident: falling levels of car ownership among city dwellers, less parking provision in urban residential areas, much greater use of shared car services like Zipcar, and fewer young people sitting driving tests. “In the future, if you book a hospital appointment or a ticket to a concert at the O2, I believe you will be offered a shuttle pod to take you from transport hubs like the DLR station, and onward to your destination. Autonomous shuttles can provide an efficient means of removing some of the disincentives to public transport use.”
Future tests this summer in Greenwich will involve multiple electric shuttles with fleet coordination software choreographing their passage, as well as adding cargo pods to the mix: carrying groceries and parcel deliveries and pickups.
As E&T reported recently, the driverless pod trials in Greenwich have met with a decidedly mixed reaction, particularly from pedestrian and cyclist advocacy groups.