CIP SIG Event Roundup: The vehicle is the easy part

I came expecting a day about autonomy and left thinking about coordination, supervision and connectivity.

Start with the off-road case, because that’s where this lands first. The opening port slot from Belfast Harbour didn’t run in the end, but the bounded-environment argument turned up everywhere else regardless. Autonomy works first where the operating domain is fenced, repeatable and owned by one party: a yard, a port, a logistics campus. No members of the public to design around, no regulator asking who’s liable when it goes wrong on the A14. Chris Britton from Voltempo did the serious commercial work on the autonomous HGV case, and it was a grounded read rather than a hype reel. On the open road there are still real liability and regulatory questions to answer before the case closes, and driver shortage alone doesn’t carry it. Where the numbers already work is the constrained stuff: hub-to-hub, yard shunting, the bounded moves. That’s where it starts.

Remco Derksen from Rebel Group made the point of the day, and he made it about buses, which only proved how portable it is. To run an autonomous vehicle in public service, the clever bit isn’t the vehicle. It’s the control room. Remote supervision, the ability to support an autonomous fleet from a desk, is the success factor. Reframing autonomy as a remote-operations problem rather than a robotics problem might be the most useful thing said all day, because it moves the question from “can the vehicle drive itself” to “can we run a fleet of them from somewhere else”, and that second question has a clearer path.

WPark picked up the same thread from a different angle. Jacky and Mabel Chu are building what they call an AI traffic controller, Spectre, and the framing is air traffic control for ground vehicles. Not a smarter car, a smarter coordinator: a layer that knows where the spaces are and routes vehicles into them. Their early numbers are off-road again, parking and logistics sites, a single-digit lift in utilisation and a cut in per-vehicle emissions in lab conditions. The pitch is that the next gain in mobility isn’t another sensor on another bonnet, it’s vehicles moving as one coordinated system instead of a few hundred clever machines each solving the same junction alone.

Dan Clarke from the Greater Cambridge Partnership gave the counterweight, and it was the talk that stuck. Taking the Cambridge Connector autonomous bus from trial towards commercial service is the on-road reality check the day needed, and Dan told it through the edge cases rather than the vision. A wheelie bin left out in the wrong spot confuses the buses. Low-hanging branches set them off. And there’s one particular corner where the safety driver has to take over every single time, because the bus obeys the highway code to the letter and there’s no way to make that turn without breaking it. That last one is the whole problem in miniature. The vehicle following the rules perfectly is sometimes the reason it can’t proceed. On-road public deployment is hard precisely because it’s none of the things that make the off-road case easy: the public, the regulation, the edge cases, the politics. Put Dan’s talk next to Chris’s and you understand why this starts in the yard and not on the motorway.

Pete Donaldson from BT closed the technical case with the part nobody puts on the brochure: the connectivity. None of the above works without it. The control room can’t supervise what it can’t reach, and the coordination layer can’t coordinate vehicles it can’t talk to. The models for connecting autonomous operations are still being worked out project by project. He framed the whole space as physical AI, land and air, which is a useful umbrella, and a reminder that the comms problem is the same whether you’re flying a drone or running a bus.

Dejan Bojic chaired the panel that pulled it together, and by then the through-line was hard to miss. Every speaker had quietly stopped talking about the vehicle and started talking about the system it sits inside. The control room, the coordinator, the connection, the bounded environment where it all gets to work first.

That’s the value of a room like this one. On road, off road, one future was the title, and the day’s real argument was that the “one future” part is the layer sitting across all of it: the supervision, the coordination, the connectivity. It’s the same problem whether you’re moving a bus through Cambridge or a container off a ship, and you only see that clearly when you get the port people, the bus people, the freight people and the network people in the same room. That’s what the SIG is for. The vehicle, it turns out, was the easy part.