Delivery robots are set to become a regular sight on UK pavements. Deliveroo has partnered with select restaurants to roll out delivery robots starting in London, with more towns and cities to follow, mirroring a path DoorDash has already taken in the United States, writes Nishith Rastogi, CEO and Founder of Locus. The mix will widen beyond food too, with Evri announcing its own UK trial of two-legged delivery robots.
While most orders will still go to human riders, the arrival of robots creates a mixed delivery network in which work has to be divided between people and machines. For warehouse and fulfilment operators, that division raises practical questions about which orders go to which vehicles, how robots will be tracked, and how that tracking feeds back into the fulfilment operation.
The answer lies in orchestration and how well dispatchers coordinate human teams and robot fleets. The first test within this coordination is deciding who gets which order, and why.
Robots or riders?
Every order now comes with a choice. With last-mile delivery already accounting for 41% of supply chain costs, getting that choice right matters: is a human rider or a robot better placed to complete each delivery? Robots and riders suit very different jobs. Robots move at walking pace and are best kept to short urban trips, which leaves longer distances and tight deadlines with riders and drivers. A good transport management system can take drop-off distance, order size, queue times at the pick bay and rider availability into account, then revise the choice as conditions shift. Fixed rules cannot do that.
Live decision-making
Revising those decisions requires a live view across warehouses, outlets, riders, drivers and robots. Dispatchers already need that full picture for their human fleet; robots add another layer. A robot can be held up at a kerb, or a picking line can fall behind, and either can delay the order.
Even when a robot can navigate around an obstacle, the extra journey time may put the delivery promise at risk. That delay needs to feed back into dispatch planning so the wider operation can respond.
This is where agentic orchestration comes in. It lets retailers use AI to weigh live conditions across the warehouse and the street, and adjust plans as those conditions change. But adjusting plans is only half the job; the network also needs a safety net.
Contingencies that protect the delivery promise
Delivery robots will only be as effective as the contingencies built in to protect the delivery promise. That means a workable fallback whenever a delivery falls behind.
At an evening peak, for example, an order still being packed when its robot arrives leaves the robot idle. An orchestration layer that sees the delay as it happens can assess whether an available rider could still meet the promised window and reassign the order where appropriate. Managing each vehicle type as a separate operation would only multiply the coordination work, so riders, drivers and robots need to be planned as one network.
Retailers can also prepare before robots arrive in their area. Measuring the time between pick completion and handover shows where a rider or robot would sit idle. Using live warehouse and delivery capacity to set delivery promises keeps checkout commitments in line with what the network can deliver.
Two checks are worth making early. Can the orchestration software add a new delivery mode without rebuilding the technology stack? And what handover data can robot partners share with dispatchers?
The coming months will show where robots earn their place in a delivery network and how shoppers respond to them. What makes this exciting is that retailers will be able to choose how each order is delivered. And when those choices are made on live data, the promises built on them have a much better chance of being kept.
Comments are closed.