The time is right for autonomy-enabling technologies to be integrated into the mass transport market for public transportation and last-mile delivery.

While several car manufacturers now offer Level 2 or Level 3 driving automation, there is a growing realisation that personal autonomous vehicles (AVs) are still some way off from becoming a mass transport solution. 

But a long-term trend towards sub-urban living and the growth of e-commerce is now inspiring a host of new business models in public transportation and last-mile delivery that take advantage of autonomy-enabling technologies. 

As these applications are also less demanding in terms of driving automation, they will provide a faster route by which these technologies can arrive to benefit the masses.

Robo-shuttles for public transportation

The case for robo-shuttles practically makes itself. Almost everywhere, we are seeing growing cities, growing suburbs (often without the benefit of historic transport links such railway lines) but also growing congestion [1,2]. 

City planners now work on the assumption that personal vehicles – whether they are autonomous or not – won’t be the solution. Nor will traditional bus services provide an answer, as many countries are facing driver shortages, which are only expected to worsen [3-5]. 

An increasing number of companies are now preparing to serve this market with autonomous vehicles that will make it possible to either increase the frequency of services on existing routes or to open new ones with a reduced cost model. 

The experience so far shows that cities won’t need to overhaul their public transport network to incorporate these shuttles. EasyMile’s 11-person vehicles, for instance, already integrate with and extend the public transport network of the historic town of Monheim am Rhein in Germany [6].

An additional benefit of new point-to-point services (e.g. Fusion Processing) is that they can make transit networks less “interdependent” by offering more independent and direct routes, thus avoiding congestion at transport hubs and timetabling constraints on operators. 

Autonomous services can also ease labour shortages, by splitting route networks between drivers and robo-shuttles, with bus drivers allocated to more demanding routes. Over time, the drivers’ responsibilities can expand to oversee autonomous buses remotely.

Recognising this potential, the UK Government is supporting the integration of autonomous technology with a £40 million competition to fund projects to bring autonomous commercial vehicles to market and create a predicted 38,000 new skilled jobs by 2035 [7].

Last mile delivery with “Autonomous Prime”

We are also seeing sustained growth in e-commerce, accelerated by our pandemic penchant for online orders. By 2025, e-commerce will account for an estimated 25% of global retail sales [8].

One popular company tapping into this market is Starship Technologies. Their six-wheeled robot has found its way onto the pavements of several towns close to TTP’s Cambridge home, with schemes popping up across the country for last-mile deliveries. 

The clever thing about this business model is that local councils can deem the rovers safe to roll alongside pedestrians on pavements where they only need to abide to “soft law” rather than highway codes [9]. This is creating opportunities to roll out new services rapidly.

To support the business models of large logistics companies and supermarket chains, however, multi-stop and road-based delivery systems with much greater capacity will be needed (e.g. UdelvCleveron).

These delivery systems will operate as mobile parcel lockers that serve a whole neighbourhood. Multi-stop and road-based capabilities open opportunities in urban settings, where a delivery driver would normally service several streets or buildings in a single delivery run before returning to a depot or retailer.

Easier driving

In both cases, the technical burden on driving automation is less than for personal AVs. Robo-shuttles benefit from more predictable routes than personal AVs, which need to be ready to take their passengers to practically any destination. 

From a software perspective, the environment along a fixed route mostly stays the same. The vehicle perception system can therefore more easily learn where to expect objects of concern, such as pedestrians or other vehicles, and what responses are warranted.

And moving parcel lockers have the advantage that they can prioritise the safety of other road users without inconveniencing any passengers. When delivering parcels in a neighbourhood, they may encounter obstacles that slow down the vehicle or cause it to stop.

Passengers in personal Avs, or even in robo-shuttles, would see this as a considerable inconvenience. But the recipients of e-commerce deliveries can cope with some delay. Or rather, much like we currently do when waiting for a delivery, we will be able to check on the progress of our moving parcel locker and when your order will reach your house.

Fewer sensors per passenger

Players in public transportation and last-mile delivery tend to use a similar number of sensors as personal AVs, a combination of camera, lidar, radar, and ultrasonic, while achieving SAE Level 4 autonomy.

But the “sensor-to-person” ratio is reduced for even 15-person capacity shuttles relative to personal AVs. This – in addition to the reduced automated driving demands – points to the fact that cost constraints on all the autonomy-enabling sensors will be less than in AVs for individual consumers. The same goes for form factor and vehicle aesthetics, as autonomous shuttles and moving parcel lockers tend to be larger or less aerodynamic than personal AVs.


Autonomous shuttles and delivery vehicles will need to work in the evolving regulatory environment for AVs. For example, U.K. proposals assume that there will be a person in the vehicle who can resume control when necessary. 

For the applications considered above, there will be no such “user-in-charge” in the vehicle. Alternative methods for the safe handling of edge cases would be required, and a remote operator would have to be responsible for overseeing operations.

Future autonomous vehicles will come in a variety of shapes and sizes, but promising applications for autonomy-enabling technologies are now emerging in public transport and last-mile delivery. The underlying technologies will need to be adapted to different use cases and markets, but the lower demands on driving automation suggest that this will be an easier route by which companies can break through and serve the masses. 

The sensors for autonomous vehicles are getting smarter, cheaper and more accurate. As the world gets closer to deploying autonomous vehicles at scale, TTP’s Autonomous Tech team are developing the enabling technologies and building the systems required to enable a safer future.

Gregory Dayao
Mechanical Engineering Intern