Sunday, November 8, 2015

International Driverless Cars Conference, Adelaide, 5-6 Nov, 2015

As a computer programmer, I was probably not typical amongst the 300 attendees at the International Driverless Cars Conference in Adelaide.  However, after reading the Columbia University Earth Institute study,  Transforming Personal Mobility, I'd became interested in simulating a shared fleet of autonomous cars in Canberra.  So, off I went to Adelaide to learn more.

SA Premier Jay Weatherill and Adelaide's Lord Mayor Martin Haese started the conference by issuing the following challenge to delegates: help us use this technology to reduce congestion, road accidents and demand for new roads, provide cheap door-to-door and on-demand mobility for everyone, reclaim public and private urban space from decades of car-domination and make Adelaide the world's first carbon neutral city.

Haese urged the audience to ask "What is possible?" with what he described as "this most profound technology of the 21st century". A line-up of international and local experts from industry and academia did just that over the following two days.

David Homburg, president of the Australian Institute of Architects SA Chapter and principal at Hassell Adelaide discussed how self-driving cars could reverse the damage inflicted by car parking and garaging, ever-wider roads and rail right-of-ways to the urban form. He noted the typical double garage takes the same space as a hotel room, and increasingly dominates suburban street-scapes. Just the savings from tighter packing of self-parking cars would release 25% of car-parking space, and fewer but shared cars would allow cities to reclaim vast amounts of space allocated to multi-storey car-parks and street parking. In considering how the coming technology would affect transport options, Homburg said it wasn't "one size fits all". He noted that Singapore, with high density and efficient rail, are planning to use self-driving cars primarily for "the last mile" between homes and rail stations. But with an analysis that has implications for Canberra's transport planners, he suggested that many cities such as Adelaide already have a road infrastructure with the capacity to use the more efficient driverless cars as mass transit.

In a QandA session, former WA Transport Minister and current Federal Member for Perth, Alannah MacTiernan raised this issue again: will the arrival of fully autonomous cars call into question the appropriateness of traditional major road and rail infrastructure projects?  The consensus seemed to be somewhere between "it depends on circumstances" and "yes".

Professor Alan Stevens, chief researcher at the UK Transport Research Laboratory and Chairman of Intelligent Transport System and Paul Grey, CEO of Cohda Wireless detailed the benefits of vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communications in avoiding accidents and optimising traffic flows. Stevens described three driverless car trials supported by the UK government and industry in Greenwich, Bristol and Milton Keynes. [Milton Keynes is of interest to Canberra residents, as it is a planned city designed around a hierarchical grid road system and the private car in the 1960's. High population growth has raised the spectre of congestion. Earlier this year, Milton Keynes discarded plans for a light-rail in favour of a trial of autonomous "pods".]

Paul Grey characterised the safety, fuel efficiency and congestion improvements facilitated by V2V and V2I, and how his SA company's vehicle-to-everything (V2X) technology will soon start appearing in General Motors. He also unveiled their V2X radar technology which allows local mapping capabilities to piggy-back on V2X communication transmissions.

Oliver Cartsen, Professor of Transport Safety, University of Leeds, discussed driver behaviour, the causes of accidents and how safety could be improved with relatively "low tech" application of V2V systems to alert drivers to dangers and perhaps to automatically take action to prevent accidents.

In related presentations, Professors Raj Rajkumar and Ramayya Krishnan, both from Carnegie Mellon University, stated the motivations behind their work: to reduce the 1.2 million traffic deaths annually, to reduce the costs of congestion, and to assist independence and self-esteem by providing mobility to all. They reported the challenges, both technical and societal, which must be addressed before driverless cars are commercialised. Rajkumar stressed that the huge benefits on offer have generated great competitive pressures between automakers and large IT companies, and between countries and states vying to gain advantage through early adoption, which in turn has spurred extremely rapid progress. Krishnan described how relatively simple signal optimisations had been demonstrated to produce savings of around 25% in travel times and 20% in vehicle emissions in urban settings. He raised the importance of public policy in ensuring private gains to those using driverless cars are not offset as costs to others. For example, an occupant of a driverless car is free to use their travel time productively and will not suffer as keenly the costs of congestion to which they contribute.

Gerard Waldron, Managing Director of the Australian road research group, ARRB outlined the importance of collaboration in gaining both the early and long-term benefits of intelligence transport systems. To this end, ARRB has established the Australian Driverless Vehicle Initiative (ADVI) which will facilitate cooperation between industry, academia and government on technology and policy.

Representatives from Volvo, Tesla, Bosch and Navya ( a new manufacturer of driverless shuttle buses) described their technology and directions.  After a raising expectations of an exciting announcement by first showing a slide full of fine-print with terms such as "safe harbour" and saying "Tesla doesn't normally do this", the Tesla presenter turned out to be a sales rep and showed a couple of marketing videos most attendees had already seen.  He also repeated Elon Musk's assertion made in October that Tesla will deliver a fully autonomous car by 2018, qualifying somewhat by describing that date as "Elon time".  I'm sure he was meaning his boss was optimistic, not delusional.  The apparent consensus from presenters was that fully autonomous cars capable of operating in all environments would probably not be commercialised until 2020 at the earliest, and possibly as late as 2030, although specialised operation in certain settings would almost certainly be available before unrestricted autonomy.

Philosopher Robert Sparrow from Monash went over some well-discussed ethical and commercial issues raised by cars that could be programmed to kill their passengers.  Gerard Waldron, Managing Director, ARRB and Rita Excell also from ARRB described the rationale and operation of their new Australian Driverless Vehicle Initiative (ADVI).

Robert Anderson and Jeremy Woolley from the Adelaide University's Centre for Automotive Safety Research discussed benefits of auto safety systems and made some interesting points about safety spending.  For example, it is known that sealing a road shoulder helps drivers maintain control should they stray off the main road; but as more cars are fitted with better brakes and stability control, is this the best use of public funds?

A major theme to emerge during the conference was the care required during the transition from full human control of cars through the increasing levels of automation to fully autonomous operation. Many speakers discussed the vital but difficult "human factors" problem of making it clear what the car is doing and who is responsible. Simon Henderson, a pilot and manager of Virgin's Fleet Standards discussed autonomous aviation systems, warning that great care must be taken in their design as otherwise monitoring them can require more knowledge than flying. Apparently, one of the most frequent final things heard in flight-recorder post-mortems is "What's it doing now?"

Another common theme was that the synergy of autonomous driving technology, V2V and V2I communications, electric vehicles and on-demand travel in a shared fleet of cars, together forms intelligent transport systems (ITS) capable of delivering much cheaper, cleaner and more convenient transport. (Yes, "ITS" is a "thing" you can expect to hear a lot about in the coming years.)

As Andrew Somers from Transoptim Consulting observed "transport has not been seriously disrupted for some time", but that period of stability is about to end.

Whether the great potential of commercially available fully autonomous cars is realised within 3 years as claimed by Tesla CEO Elon Musk, or within 5 years as predicted by Ford CEO Mark Fields and by Google, or even with 10 or 15 years is almost immaterial, because urban infrastructure is planned, built and maintained over generations.

The internet transformed the world of communications and ideas by facilitating the very cheap and near instantaneous dissemination of information. Autonomous cars promise a correspondingly dramatic revolution in the movement of people and physical goods, with far-reaching flow-on effects on where and how we live. The better prepared we are, the greater the benefits we will derive.