How Can We Make Self-Driving Cars Actually Happen? 2

Wednesday, August 10th, 2016
liked this article
Engineering Education Australia – 300 x 250 (expire Nov 30 2016)
FavoriteLoadingsave article

“You sit, you relax, you don’t have to do nothing. It knows when to stop, it knows when to go.”

That is how a lady identified by her first name of Annie describes her experience on Google’s promotional video after riding what are fast becoming known as self-driving cars or autonomous vehicles (AVs).

Annie’s response also articulates some of the benefits which companies who are investing in autonomous vehicle technology would like to promote.

At an individual level, these include improved safety features associated with the elimination of ‘human error’ when driving as well as greater convenience through being freed of the need to perform active driving tasks whilst in the course of their commute. Unveiling his company’s master plan for the future on July 20, for example, Tesla CEO Elon Musk outlined a vision of autonomous vehicles being 10 times safer when compared with cars controlled by human drivers.

Longer term, opportunities for improved traffic management are possible where any vehicles which are connected to the network could, for example, be rerouted in such a way as to avoid sections of road which are impacted by accidents, breakdowns or excessive congestion.

Nevertheless, the extent of work which still needs to be done was underscored on May 7, when 40-year-old Joshua Brown tragically became the first known driver to die whilst riding in an autonomous vehicle after the vehicle’s sensor system failed to distinguish a large white 18-wheel truck and trailer crossing a highway in Florida and attempted to drive under the trailer.

That raises questions about how we are going to go about delivering upon technical standards and rules which will be needed to enable vehicles of different make and model to communicate seamlessly with one another and some of the challenges which need to be overcome before ordinary Australians will be able to step into an autonomous vehicle and be confident of arriving safely at their destination.

Further questions surround whether or not vehicles will actually be networked together en masse or whether cars will continue to operate independently of each other as vehicles with human drivers do and simply be equipped with the sensors, cameras and mapping technology which are required so as to enable them to manage themselves on the road without interference from human occupants.

On that last point, Queensland University of Technology associate professor Michael Milford says the jury is still out as to which way the industry will go. What is clear, however, is that self-driving vehicles will need to be equipped with their own individual independent capabilities so as to enable the vehicle to be driven outside of major urban areas where major networked infrastructure is unlikely to be available, Milford says.

In terms of technical standards, Milford says these may not necessarily prescribe detailed methodology but will more likely focus around the standards which the vehicle will have to achieve from a performance perspective. Manufactures may have to prove, for instance, that their vehicle is able to demonstrate 99.99 per cent reliability when driving over the course of a million miles in wet conditions, or that sensors within the vehicle can reliably ‘see’ or detect other objects within given distances even when there are adverse conditions.

Milford says difficult questions will also need to be resolved regarding human ethics and values, such as the priority decisions which an autonomous vehicle should be programmed to make when confronted with a situation where an accident is unavoidable.

Overall, Milford puts the odds of autonomous vehicles being commonplace within a decade at roughly even money. He says commercial imperatives dictate that the development of self-driving cars is inevitable irrespective of any setbacks which occur.

“The commercial and economic pressure to develop autonomous cars is so incredible – it’s a trillion dollar market – that regardless of the ethics or how many people get killed or what happens along the way, it’s almost inevitable assuming that we don’t have a complete meltdown of society,” Milford says.

Dr Charles Karl, national technical leader, congestion, freight and productivity at the Australian Road Research Board (AARB), offers a slightly different perspective.

With current experimentation efforts on the part of manufacturers remaining primarily focused around their own vehicles, Karl says, the first phase of the autonomous vehicle revolution is likely to revolve around the ability of individual vehicles to operate without input from the driver. The linking together of individual vehicles will likely constitute the next phase of development after that, he says.

When it does come to the question of linking vehicles together via a network, Karl says considerations surrounding interoperability will need to be addressed from a number of angles.

First, there is the question of technical interoperability and how we are going to ensure that vehicles produced by different manufacturers can communicate and interoperate seamlessly with one another whilst on the road from a technical perspective.

Next, there are questions of ‘business interoperability,’ which is needed to enable corporate customers such as transport operators, service fleets or the police to seamlessly procure and maintain fleets of vehicles which may cross different suppliers or models or to enable emergency roadside assistance personnel to perform work on vehicles irrespective of their make and model in a similar way in which they do now with respect to ordinary vehicles.

Finally, there is the less obvious level of semantic interoperability. If someone were to say that Melbourne’s West Gate Freeway is ‘cactus,’ for example, exactly what ‘cactus’ specifically means will need to be understood across the network. Likewise, if the current condition of the road (due to weather conditions and the like) is determined to be a ‘Grade A’ or a ‘Grade F’, exactly what this means needs to be fully and specifically understood across the autonomous vehicle fleet if network connected vehicles are to be re-routed accordingly.

Karl says questions surrounding how autonomous vehicles will evolve will largely be answered through field testing. He says rather than having one overarching set of rules for all types of vehicles on the road, it is likely that different rules will apply to different vehicles in a similar way to which the Civil Aviation Authority has different rules for different types of aircraf, such as military aircraft, commercial aircraft, gliders, private aircraft or drones. Fully autonomous vehicles, for example, could be used in certain corridors, such as shuttles which go around airports, universities or golf courses or transfer commuters along special ‘last mile’ corridors. For other classes of vehicle, such as heavy trucks or light vehicles, individual rules might dictate the conditions under which that type of vehicle could be allowed to operate on autopilot.

Karl says autonomous vehicles also pose interesting questions from a regulatory perspective. These include whether or not different procedures are required for autonomous vehicle registration as compared to regular vehicle registration, or whether or not special licences should be required to operate autonomous vehicles as opposed to normal ones.

It is also possible that road authorities and toll road operators could potentially push for the ability to have some control over when and where auto-pilot modes within a self-driving car are able to be activated, and may wish to retain an ability to disallow this in cases where road conditions are deemed to be overly hazardous.

Mark Whitmore, head of enterprise marketing APAC at multi-national digital mapping and navigation outfit HERE Maps says a number of features need to be in place in order for autonomous vehicles to be able to operate seamlessly and naturally on everyday roads. As well as sensors which are sufficiently intelligent to sense everything around them, vehicles will also need a ‘digital horizon’ which will be provided by high definition (HD) digital maps. These maps will enable vehicles to ‘see ahead’, allowing them, for example, to position themselves in the correct lane in preparation for an upcoming turn.

Just as important, however, is the humanised level. Autonomous vehicles will need to learn not only about safe and efficient operation from a technical standpoint, but also about aspects such as the tolerances and speeds at which their human passengers will indeed be comfortable. Cars need to learn to differentiate between what is possible and what is acceptable to an everyday driver if he or she is indeed going to be prepared to cede control of the wheel. Cars will also need to learn about our travel preferences. Rather than travelling along more direct and congested routes, some drivers may on occasions prefer to take longer back street pathways.

When self-driving cars become connected to a network, Whitmore says, they will become mobile information probes, sending, responding to, interpreting and collecting real world information. The intelligence gleaned from sensors within the vehicles and uploaded into the cloud will find its way back to drivers through improved services.

Upon approaching a speed limit sign, for example, the vehicle could compare this to what is understood to be the speed limit by the centralised mapping system and trigger an update if indeed there is a difference. Vehicles could also alert other vehicles behind them (and have them slow down) about an animal on the road or a child on a bicycle. Even the simple act of putting the windscreen wipers on could be used to alert other drivers of approaching weather conditions.

In terms of who makes up the standards and rules, Whitmore and Karl both believe the technical standards are likely to be developed overseas and that any thought of Australia making up its own standards would be impractical (especially when most of our vehicles are being manufactured overseas). Having said that, Karl says we will then need our own operational framework to ensure everything works together and local rules and conditions are adequately accounted for.

In terms of when all this will become reality, Karl acknowledges that things will not happen overnight but adds that we should not underestimate how quickly change can happen.

“The iPhone was introduced in 2008, Karl said. “So it’s not even 10 years (since the release), it’s eight years. Look at how much our lives have changed with the smart phone.

“Some people have said automation is like smart phones on wheels. I’m not sure about that. But you know, it’s not going to be 40 or 50 years. If you look at how quickly we have embraced the smart phone from 2008 – even young people are going around in schools, within eight years.

“How this is going to work with cars, I dare not guess.”

FavoriteLoadingsave article


 characters available
*Please refer to our comment policy before submitting
  1. Barry B.

    We actually don't need to worry about this at all as Australians – we just need to wait for the big tech companies in North America to perfect the technology.

  2. matt

    Driver less cars are a twenty year pipe dream. Waste of time thinking about is as we deleted the Australian car industry anyway. No benefits for Australia here, lets move on