Automated Vehicles

Driverless (or "autonomous") cars are all the rage at the moment for their promise of improved safety, convenience and road-carrying capacity (although public policy will need to develop to ensure these benefits are delivered).

But there is also a lot of hype, over-promising and even dangerously rushed implementation from companies such as Tesla (which – thanks to a "poor culture" led by Elon Musk – is putting "self-driving" systems on the road that are clearly not ready), given there are many barriers to implementation of fully automated cars, not least safety issues (especially in bad weather at speed), or the cost & aesthetics of segregated or overhead infrastructure.

So initial applications will probably need segregated corridors, but that increases costs (not least for land), so commercially viable corridors will need reasonably high demand, which of course tend to be those most suited to public transport.  Subsequently cars might venture off these corridors onto fenced-off, feeder lines (such as out-of-CBD automated parking stations), before being extended to roads shared with pedestrians.

Regional Australia could also be suitable for early applications, given the huge benefits to safety for areas where public transport is almost non-existent and road accidents are an order of magnitude higher than in urban areas.

Meanwhile, in public transit land, there are umpteen new ideas out there to literally reinvent the wheel/car with driverless Personal or Group Rapid Transit vehicles (PRT / GRT, where PRT has compact vehicles for only one or two people and GRT has up to minibus-sized, shared-use vehicles with space for a dozen or so passengers).  e.g. see:

Driverless transit offers the prospect of changing the cost/speed/service-frequency/weight trade-off & optimisation problem - enabling safe operation of smaller, closely-spaced (low waiting time) vehicles with high total capacity, but without the penalty of more drivers or larger, heavier trains and infrastructure (with more costly & less flexible routing).

This has potentially huge implications for the future pattern of urban development beyond the traditional confines of commercial centres, as discussed in my attached slides from 2004.

But whilst the promoters of these technologies typically highlight the benefits of an automated city-wide network with the "anytime-anywhere" convenience of the private car, none of them seem to have any realistic commercial way of getting there - in a staged way.  No city is going to dive straight into building an entire network of a new technology to replace the car.  As with driverless cars, the most viable initial applications are likely to be GRT on moderately-high capacity corridors where traditional public transport is too expensive or slow, &/or where GRT offers the potential for more flexible future expansion.  PRT seems to have little advantage over driverless cars in terms of network capacity, but may offer the prospect of lower cost, light-weight overhead infrastructure and faster vehicles, as Skytran claim.

It seems pretty clear that automated cars and public transport are on a path of technology convergence.  

The issue then is what vehicle design is likely to be most viable?  I suspect the best commercial strategy, at least initially, is to use incrementally-customised versions of standard cars (like my design below), as they could build on the existing huge economies of scale available from the car industry, and could also allow car owners to drive on and off automated high-speed corridors for the first and last leg of their journey.

Also, whether driverless cars or PRT/GRT, a key requirement is advanced network optimisation and control systems, which, though undoubtedly feasible, are not cheap to develop.  Luckily Uber and others are leading the way - again with initial applications for ordinary cars/taxis.  Application to shared-use and then driverless taxis are just 'down the road' (excuse the pun), if there is effective management of cleanliness (e.g. assisted by smell sensors) and inter-personal safety, e.g. by automatically partitioning the front & rear of vehicles when there are only two passengers (with more than two there is safety in numbers).  Any applications of driverless cars or PRT/GRT need to extend on this existing network-optimisation technology and high-value commercially-viable transport market applications that Uber and others are currently pursuing.  Meanwhile automatic-train-control systems for metro rail offer a starting point for safely controlling closely-spaced cars or GRT vehicles on medium-capacity corridors.

Ultimately, as with electricity distribution networks, optimisation of the total transport network may require a distributed intelligence system, perhaps using local "price signals" in every vehicle and at every intersection, rather than having an all-knowing, super-powerful centralized control system. With the right "market design", the prices at each node could produce "transactions" that produce the best local outcome, and, just like Adam Smith's "invisible hand of the market", an optimal outcome for the overall system.

With appropriate corridor safeguards & emerging control technology, it should soon be quite feasible to build an automated, light-weight elevated lane above a freeway median strip for regular-sized, "on-demand", guided cars/vehicles to safely travel over 200kph (given the Tesla Model-S has a "governed" top speed of 250kph), which, being more flexible than conventional rail, could cost-effectively traverse the difficult terrain over the 160km from Sydney/Parramatta to Newcastle in under an hour, and also enable future branch lines to take people closer to their ultimate destination at each end of the line.  Although this approach would not nearly match the capacity of conventional heavy rail, a plausible vehicle capacity of 9 persons with a spacing of 20 seconds would provide a reasonable capacity of 1,620 passengers per hour (comparable to a well-serviced aviation route) and with superior service and relatively low infrastructure cost would have a much better chance of being able to charge premium fares and deliver an appropriate return on investment.  Over time, improved control systems could potentially reduce safe vehicle spacings to about 2-seconds and hence increase line capacities by a factor of ten.

As for incrementally-customised versions of the car, below is my idea for a detachable road-anchor for safe, high-speed, lane-changing vehicles to operate in all weather and road conditions (which won me the 1992 Churchill College Engineering Essay competition).  It would only require one non-precision tube per lane buried beneath the road, with an edge-tracking detector on the anchor constantly adjusting its position to avoid touching the tube edges - except if the wheels lose their grip, when it would provide a physical anchor to keep the car on the road.  Two anchors - one each near the front and rear of the car - enables one to always keep the car safely anchored to the road, even whilst changing lanes - as illustrated in the series of drawings below.

I originally thought the buried tube would include power and communications lines with a pick-up on the car anchor, and even though battery and mobile communications technologies are now so vastly improved, with costs as low as €1m/km for this Swedish system, it may well still be economically viable (especially if the tube doubles as a convenient duct for utility cables that are required anyway).