CamCross: An innovative concept for the Cambridge Autonomous Metro - Part 2 - Vehicles
Updated: Feb 21
Hello everyone, this week I will be sharing my thoughts about the vehicle design for CamCross.
Before that, I would like to thank Ben Hatton, the Local Democracy reporter, for writing a lovely article about my work and this blog in the Cambridge News. He's also written a great summary of all the current transport plans within Cambridge before 2030.
If you've not read part one I would start with that blog as it sets the scene for this. :)
In the first blog post, I wrote about the philosophy behind CamCross (the need to provide reliable, frequent, affordable, fast, and attractive transit) and gave a summary of the system.
Over the course of the following blogs, I am going to be exploring each of the system elements in more detail.
Firstly, is CamCross another example of GadgetBahn?
Gadgetbahn - (slang, transport, pejorative) - In public transport, transportation concepts which seem to be infeasible or unnecessary.
It is easy to brush the CamCross concept off simply as “GadgetBahn” and this is understandable, as there is a lot of noise in this field and I have seen all sort of proposals for pod-based systems aimed at various cities/ airports/ conference centres. I too was initially sceptical about using smaller vehicles when I first started looking into the topic but became more convinced as I found solutions to key issues.
There is a lot of nuance in CamCross’ design to overcome issues such as limited capacity, high technical risk, and safety that plague many proposals in this space. I kept to using commercially available technology and made design choices which allow high passenger capacities to be delivered.
Whilst I’m talking exclusively about the vehicle today, they cannot be treated in isolation, and a "systems thinking" approach to the whole system was required - it is the combination of the vehicle concept, network layout and novel station designs proposed which allow the system to function.
What’s the thinking behind using low-capacity autonomous vehicles?
Using autonomous vehicles makes it cost-effective to transport passengers in lower capacity vehicles than would otherwise be possible with a human driver. This means that vehicles can be much smaller than the one shown below. On less popular routes this allows a high frequency of service to be provided whilst maintaining high occupancy, and on popular routes, passengers can be selectively grouped based on common origins/ destinations to perform a minimum number of stops.
Importantly, vehicles cannot be so small that the stations fail to cope with the high number of vehicles. Vehicles with a very low occupancy rate (e.g. 2 passengers), referred to as personal rapid transit, will almost certainly result in congestion within the system, mirroring today's roads unless very complex and expensive stations are designed. The number of vehicles required would also be high, driving up the cost.
Won’t it be decades until autonomous vehicles are good enough to operate safely?
This is likely true for road-going level 5 self-driving cars that must operate amongst unpredictable human drivers, pedestrians, and a poorly maintained road network (particularly road signage). However, I argued that a network of segregated (and tunnels), where only CamCross vehicles can operate, would be required to insulate the system from road traffic.
This vastly simplifies the complexity of the autonomous driving challenge by removing many of the uncertainties that make achieving level 5 autonomy so difficult. There are no unpredictable pedestrians, cyclists or drivers - it's coping with these edge cases that is so challenging, the vehicle control is comparatively much easier. Building a system that can do this is within the realms of possibility with technology that has already been proven today.
The self-driving control performed by many conventional cars on highways is arguably more challenging than what is necessary here. What is more, all vehicles could wirelessly communicate their position and speed to each other, sensors could also be installed throughout the pathways to detect any trespassers, and the vehicles could be supervised by humans at a central control centre (in a similar way to Air Traffic Control).
Surely there is still technical risk with this argument? What if the vehicles don't work for some reason?
There is definitely still some technical risk present, as there is with any vehicle development programme, and it would not make sense to start building expensive infrastructure until confidence in the vehicle’s performance has been confirmed.
Thus, my suggestion would be to invest a small amount of money (~£5-10 million) to build several prototype vehicles as soon as possible. These prototypes would act as a technical demonstrator and could be made to run autonomously on an airfield or proving ground to validate the autonomous driving systems show the limits of the systems (e.g. Zoox have done similar things with their test mules). This could likely be done in the space of a year or two with a sufficiently well-funded, and motivated team.
The initial infrastructure designs (particularly pathway routing and initial station design) could be developed in parallel with these prototypes. There is a lot of work required to identify where stops should be, what the paths connecting these should look like and with obtaining building permissions/ buying land etc. - more on this next week
If the vehicle tests went as intended, then detail work can begin on station design with the confidence that the vehicles will operate as needed. If the vehicle's performance is less promising than originally hoped, then the design could be reworked before fully committing to detailed parts of the system design (such as station dimensions, pathway widths etc.) By performing the vehicle prototype testing and initial infrastructure design in parallel this helps to ensure that the launch date is as early as possible (ideally before 2029).
How do you stop the system’s infrastructure costs from becoming extremely expensive?
Connecting Cambridge to the surrounding satellite towns will require several hundred kilometres of pathways at grade, as well as around 10km of underground tunnels. A system of this expanse would become unaffordable if large rolling stock is used, so vehicles were designed according to several infrastructure cost reduction principles:
By using these principles, it allows tunnel diameters to be reduced from 6m+ to under 4m. As the cost of tunnelling scales with the area, the savings are large in doing this.
What could a CamCross vehicle look like?
There are many potential vehicle concepts that I believe might work in this system, and narrowing down a final design requires significantly more work. Here, I’m presenting a reference design for a vehicle with certain novel features aimed at addressing the key issues that would otherwise stop a system of small vehicles from working.
Reference Vehicle Key Features
Vehicle dimensions: 6.0x2.28x2.43m (lwh)
Total Capacity: 24 (16 seated + 8 standing)
Proposed Max Speed: 80km/h
Tunnel Diameter: 3.7m
Width of pathways at grade: 2.5m
Projected Vehicle Cost (Including Autonomous driving technology):
~£100,000 (based on current prices for similar-sized autonomous pods)
How were the maximum dimensions of the vehicle determined?
Constraining the frontal area of the vehicle is required for the cost of tunnelling and building pathways to be kept low, but it still needs to be large enough for passengers to be comfortable. I drew a range of geometries illustrated to the left to generate bounding boxes for vehicles in different sizes of circular tunnels. Building a vehicle with rounded top corners would allow the bounding boxes illustrated to be exceeded slightly.
I have also designed things so that the tunnels include an escape pathway down which passengers could walk in the case of a total vehicle systems failure when underground. A 95th percentile male mannequin has been drawn in this exit pathway.
A range of vehicle aspect ratios was explored for each tunnel size. A 3.7m tunnel chosen to be the most suitable size as this allows for vehicles tall enough for standing passengers whilst keeping tunnelling costs to a minimum.
The cost of stations, particularly underground stations, is primarily driven by the required platform length, and this is dictated by the length of individual vehicles used. Thus, there are cost advantages to keeping vehicles short, in addition to the flexibility offered by transporting passengers in smaller numbers.
I found that a maximum length of 6m, as suggested by members of the CAM TAC, worked well for the reference design.
What does something of this size look like? And how do you make sure there is enough space for passengers on board?
The reference design is for a vehicle which is not dissimilar in size to a minibus – a Ford Transit 15-seater minibus is slightly smaller at 5.8x1.9x2.26m (lwh) (checked in May 2020).
Whilst a capacity of 15 would probably be workable, I have tried to increase the passenger capacity as this would reduce the number of vehicles required in a fleet (and therefore cost).
The minibuses mentioned above use internal combustion engines, and much of the footprint of the vehicle is dedicated to housing a large fuel tank, engine, transmission, and driver controls. Moving to an electric vehicle, allows a low-profile “skateboard” platform to be used, and increasing the height of the roof allows for standing passengers allows for more commuters to comfortably fit within the same footprint.
Where would you get one of these vehicles?
The UK is a great place to develop and build these vehicles given the wealth of expertise within EV design, and low volume high-value manufacturing, particularly in areas such as the West Midlands.
Low-profile skateboard architecture with the powertrain constrained to the base of the vehicle has become commonplace for electric vehicles and results in a vehicle with a low centre of gravity (good for handling and passenger comfort), and lots of design freedom/ space within the passenger cabin.
Engineering risk could also be reduced as there are companies which sell flexible platforms upon which custom vehicles can be built. For example, Delta Motorsport (one of our partners), has pioneered their “S2 Platform” for autonomous vehicles, and this could potentially have been adapted to work for one of these vehicles. This would further accelerate how quickly a prototype could be developed.
What elements of the vehicle design help the system from becoming overwhelmed and simply causing an underground traffic jam?
The largest challenge presented by using smaller vehicles is ensuring that bottlenecks at major central stations and along high transfer rate pathways do not develop. This is a major criticism of systems such as the Boring Company’s “Loop” which uses standard Tesla cars, and I can see it being an issue with any single- or two-seater pods in Cambridge as so many are required to achieve meaningful passenger flows.
To prevent a bottleneck from developing, I focused on designing the vehicle to allow passengers to alight and board vehicles with a short dwell time at stations – this helps to maintain a high throughput of vehicles. This, in combination with the station designs I’m proposing should address many of the concerns surrounding capacity.
What steps were taken to help reduce the dwell time?
Allowing passengers to alight from both sides simultaneously and through as large a door aperture as possible helps reduce the dwell times required.
I considered sliding doors, but this results in a limited opening size as these need to slide somewhere within the vehicle (see the minibus image above). Gullwing, and swinging doors were also considered but rejected because of the large amount of platform space that they need when the doors swing open.
I took inspiration from commercial vans and chose roller shutters as the preferred concept because these allow for an aperture almost the entire length and height of the vehicle. As they open, the shutters roll up into a space in the roof above seated occupants.
With roller shutters, both sides of the vehicle could be opened within 2 seconds with a motorised system, enabling a short dwell time. Transparent polycarbonate roller shutters used for securing commercial properties could be integrated into the door to serve as windows.
Seats located in the centre of vehicles
The wide door aperture and passengers alighting out both sides meant that I could not place the seats along the periphery of the vehicle and have standing passengers in the middle of the vehicle (commonplace amongst rolling stock such as the London Underground). I also tried to offer as many seats as possible whilst still allowing space for wheelchairs, prams, bikes, and large suitcases.
I decided to design two rows of 8 seats with passengers arranged back-to-back, facing outwards. The vehicle was too narrow to fit a row of standing passengers on both sides of the seated passengers. I decided to stagger these seats down the length of the vehicle to maximise standing space whilst keeping the vehicle balanced. All passengers are modelled as 95th percentile men to ensure there would be ample room with a random sample of passengers.
The vehicle’s battery pack is designed to be packaged underneath the seats, where there is sufficient space for up to around 300kWh of lithium-ion batteries. Electric motors, steering rack and suspension systems are located in between the wheels on either end of the vehicle.
How can you achieve high passenger flows with such low-capacity vehicles?
High passenger transfer rates can be achieved through convoying, where vehicles autonomously follow each other to form “trains” several vehicles long with just a couple of metres between each vehicle. For example, a passenger transfer rate of 3600 pphd (higher than short term demand estimates) could be achieved with convoys of just 3 vehicles with 60 seconds headway between each convoy. Higher transfer rates can be achieved by reducing the time between each “train” of vehicles or increasing the length of a convoy.
The vehicles would not be coupled mechanically, so vehicles can split up to take different routes or stop at stations independently of the other vehicles – this is key to enabling the flexible on-demand service. These modes of driving are reasonably complex and should therefore be demonstrated using the initial prototypes to understand the limits of the system.
What if a vehicle breaks down in a tunnel and people need to leave the vehicle?
In the event of a vehicle failure within a tunnel, the front and rear doors could be opened to allow fast egress at both ends of the vehicle, and out the sides. I did not include any seats over the wheels to allow access for egress.
How fast are these going to go? And are they going to be comfortable?
In my view, public transit should not optimize purely for speed and shorter journey times. I believe passengers would much rather a slightly slower but more comfortable journey than a very-fast ride which feels like riding a roller coaster to work.
Therefore, I have proposed a maximum vehicle speed of 80kph which is not dissimilar to the speed of buses on the current guided busway. Vehicle speed could also be limited in the tunnelled section if required. However, the maximum comfortable speed (and acceleration) is something that should be tested with the prototypes. The slower speed also helps to reduce the energy consumption of the vehicle.
Is this your final vehicle design?
This is just an illustrative concept design and by no means a final vehicle design. I tried to identify the main issues that would cause a system like this to fail and presented a solution that addresses these. There are also some issues such as battery charging that I have not covered here but have had some thought.
There is further work required to take this concept through to a prototype stage and beyond, but I believe that it is the solid foundation for further development. The vehicle concept presented here, combined with the network layout, and novel station designs mean that CamCross is something that I think will offer a very convenient service, whilst being affordable and practical for the CPCA to build and operate within the next few years.
Thanks for reading this blog – hopefully, my vehicle design ideas give a little food for thought and I’d love to hear from you in the comments (or by email!). The vehicle only represents one part of the solution, with the network layout and station designs key to enabling this system to work. Follow me on Linkedin or Twitter to be notified of when I release the next blog.
Thanks to Dr David Cleevely, Professor John Miles and Dr David Braben from the CAM TAC. This work would not have been possible without their support and built heavily upon their ideas for such a system.