Disrupting urban mobility: Trends transforming the future
This paper is a summary of an invited talk presented at the inaugural iMOVE Transport of Tomorrow Symposium, 26-27 March 2019, in Melbourne, Australia. To find out more about Professor Dia, his work, and his background, read iMOVE’s interview with him, Hussein Dia: tackling the urban transport future.
The 2020s are predicted to be a decade of transformation for urban mobility. There are at least six forces that are expected to disrupt the urban mobility landscape. From self-driving vehicles and the sharing economy, through to vehicle electrification, mobile computing, IoT and blockchain technologies, each of these trends is quite significant on its own.
But the convergence and coming together of their disruptive forces is what will create real value and provide urban mobility innovations. Once converged, they will enhance the travel experience for millions of people and businesses every day.
The future of transport
- How will the future of transport look in the coming years, with autonomous on-demand shared electric vehicles?
- Will personal car ownership decline, giving way to Uber-style fleet-operated autonomous shared vehicles?
- Will mobility be offered as a subscription service?
- Are we edging closer to a vision of ‘zero accidents, zero emissions, and zero car ownership’?
- Will we still be allowed to drive in the future? Can it all possibly go wrong?
- And what do we need to do today to prepare for a future in a highly automated world?
This article shares some insights on the role of digital innovations and disruptive technologies in shaping the future of urban transport. It starts by exploring the changing landscape of urban mobility, the emerging trends that are shaping how we get around in our cities, the opportunities and business models presented by disruptive technologies, their potential impacts and the key challenges and barriers for their deployment.
Technology growth
The application of technology in the transport field has seen strong growth over the past 20-30 years. The three dimensions of tech-enabled urban mobility included consideration of the convergence of infrastructure, technology, and travellers with the aim of creating smarter mobility with meaningful technology for a better user experience.
Substantial benefits were realised in the productivity and efficiency of our transport systems over that period. Some of the technology interventions have resulted in significant community benefits compared to the traditional ‘building out of congestion’ approaches that focused on increasing supply and capacity.
Rather than continually expanding the infrastructure capacity, the technology approaches focused on better utilisation of existing assets through the use of sensors for monitoring and measuring the performance, and through advanced data analysis and algorithms for making better use of the data and converting it into actionable policies and strategies. Examples included automated incident management systems, freeway traffic control, corridor and network management systems, and intelligent transport systems applications which delivered good benefits.
Even simple approaches like optimising traffic signal control using advanced algorithms have been found to result in larger benefit-to-cost ratios, compared to traditional approaches of increasing capacity.
NEW APPROACHES IN URBAN MOBILITY
In recent years, these trends have continued, accompanied by renewed thinking about how we provide mobility and access to jobs and economic opportunities in our cities. Some of it has been partly recognising that past practices have met with limited success and that new approaches are needed. And some of it is due to the widespread use of technology and innovations, and through the changing context for how we want to build future cities – smart, healthy and low carbon.
These encouraging trends recognise that the ultimate goal of mobility is to enhance access to jobs, places, services and goods. The narrative is changing – the focus has shifted from ‘transport’ to ‘mobility’, and more emphasis is given to ‘accessibility’. Rather than focusing on the infrastructure we need to move people and goods around, the focus is on providing the mobility we need to access economic opportunities.
And instead of giving priority to building additional infrastructure, the focus is shifting to understanding and managing the demand for travel, maximising efficiency of existing assets, and improving their reliability and resilience. These trends are also increasing the focus on the social dimensions of transport to ensure that mobility benefits are equally and fairly distributed for all income groups.
In my view, the most significant trend in recent times is the challenge to car ownership models, and in particular car sharing and ride-sharing options that have been made easier and more popular worldwide through mobile technology platforms.
Peak car?
Then there is “peak car”. A number of researchers have looked at our travel trends and found evidence that car passenger-kilometres per capita are going down in the world’s developed cities, and even in some of the emerging cities in China and India.
In Australia, the available data shows that peak car occurred in all major cities in 2004. Some have argued that this was due to the Global Financial Crisis, but the data shows the downward trend had started much earlier than 2007 and had continued to go down since then.
It is anyone’s guess whether this trend is set to continue, but at the same time it cannot be assumed that mode shares will remain business as usual, and that we should continue to build our cities around private cars. These norms may even get more disrupted through technology and digital innovations.
There are a few possible causes for these trends. The first one is the growth in public transport usage. In Melbourne, for example, the growth in public transport use has been around three times the growth in private transport use. Over this time the growth in car use has been slower than population growth.
RAPID URBAN POPULATION GROWTH
Another megatrend that will have a profound impact on mobility is the rapid urban population growth. Today, our cities make up only 2% of the earth surface, yet they accommodate around 54% of the world’s population, account for 75% of the energy consumption, and are responsible for around 80% of the emissions and pollution.
And the percentages are growing. By 2050, two-thirds of the world population will be living in urban areas. Transport energy consumption is also forecast to double to meet the travel demand in the world’s future cities. As more people move into cities, we need to think of new solutions to make transport more sustainable. This challenge is amplified by the ageing infrastructure in many cities around the world.
The opportunities
Technology and innovations continue to surprise us with their fast pace of breakthroughs and advances which continue to unfold on many fronts. There are at least six forces which in my view will have big impacts on urban mobility over the next 5-20 years. From self-driving vehicles and the sharing economy, through to vehicle electrification, mobile computing and blockchain technologies, each of these trends is quite significant on its own.
But the convergence and the coming together of their disruptive forces is what will create real value and provide innovations. Once converged, they will enhance the travel experience for millions of people and businesses every day. One of the biggest disruptions expected in this sector is the impact of artificial intelligence (AI) which will underpin this new ecosystem of urban mobility, and which many have labelled as the fourth dimension of tech-enabled urban mobility.
Business models: The kilometre as a utility
Some of the exciting emerging trends in this area are the business model opportunities. Today’s wave of disruptive mobility has been driven by innovations and new solutions to optimise ’excess capacity‘ of assets: namely ridesharing and car sharing.
We are seeing Silicon Valley extending its reach into the auto industry, and vice versa. Uber is testing self-driving vehicles, and Google is running ridesharing trials. Meanwhile, Tesla also announced that its vehicles will become part of a network of autonomous car-sharing service to give owners a way to generate revenue from their electric vehicles.
The past few years saw significant investments poured into ride sharing services. The biggest investors have been the auto manufacturers and technology companies.
The world’s most powerful companies, including auto manufacturers, are no longer interested in making a one-off transaction with consumers through the sale of a vehicle. Instead, they are targeting a new business model in which they would offer consumers seamless mobility services in which the kilometre of travel will become the main utility.
In Australia, for example, households spent around $65.8 billion in 2016 on private vehicle travel. They also spent around $2.7 billion on public transport. Together, this constitutes a $70 billion pool of money that these companies and businesses are targeting, regardless of the mode of travel. Consumers will have a wide range of options and solutions available to choose from rather than relying solely on car ownership.
Take Uber, for example. The company has not made a cent of profit since its establishment in 2010, but is still receiving huge investments. More than 70% of the company’s cost goes to drivers. It is no surprise then that it is testing self-driving vehicles. For the time being, a test driver will still be behind the wheel. In the long run, and by taking the human driver out of the loop, the company expects it would be able to provide mobility as a service at a very competitive cost.
Today, the average cost of passenger-kilometre of travel using a privately-owned vehicle is still much lower and more appealing than ridesharing. But huge investments have already been made in automation. When self-driving vehicles arrive, the return on investment will be substantial and could reduce the cost of on-demand mobility services to nearly the same cost of owning and driving a car. This assumes no carpooling. With successful carpooling, the cost per passenger per kilometre is going to put car ownership even under greater threat.
SELF-DRIVING TECHNOLOGY
Over the past few years, self-driving vehicles have captured people’s imaginations and have also inspired some visions of a different future, as well as a great deal of hype. Considerable research still needs to be done to distinguish between the hype and reality, particularly their likely impacts on urban mobility.
Road safety
One thing that most people agree on, though, is that self-driving vehicles can have a positive impact on road safety. Nearly 1.2 million people die in road traffic crashes worldwide every year. This is the equivalent of 15 wide-body aircrafts, each with a capacity of 200 passengers, falling out of the sky every single day and killing everyone on board. This wouldn’t be accepted in air travel and it is shocking that it continues on our roads today.
In addition to the pain and suffering, these crashes are estimated to cost more than $500 billion each year globally. 70-90% of these crashes are caused by human error. A large proportion could be avoided by using self-driving vehicles and there is compelling logic in removing humans – the key source of error – from the driving equation. Driven by Artificial Intelligence, these vehicles will not make errors of judgement the way a human driver does.
The road to full autonomy
There are 5 levels of automation that are now widely accepted in the industry. Most people agree that we will soon have Level 2, but the real disruption that will totally remove the driver and transform the economics of mobility, comes at Level 5.
The journey to Level 5 is a long one that has only just begun. Full automation is hard because it requires solving problems beyond technology. A fully automated vehicle needs to handle all situations including its own equipment failure. In commercial aircrafts, for example, equipment failure is handled through multiple redundant systems which makes the aircraft systems very complex and also very expensive.
No one knows yet how to scale an aircraft’s level of redundancy to an affordable mass-market vehicle. It is also often argued that the biggest barrier to Level 5 automation is Level 3. The prospects for L3 are probably in doubt today because of the problem of recapturing the attention, in an emergency, of a driver who has zoned out or who has fallen asleep. I think Google was ahead of the curve in that regard and that could be the reason why they decided to leap forward to L5 directly.
The technology today
What would things look like in the future when we have Level 5? Well, the future is already here. The sensing technologies and devices are quite mature. In some vehicles, the hardware needed to fully self-drive itself may include 8 video cameras that give the vehicle 360-degree peripheral vision up to 250 metres range. One radar, 12 ultrasound sensors; and a liquid-cooled supercomputer the size of a lunchbox. The on-board computer can perform more than 24 trillion operations per second, which is equivalent to 150 MacBook pro computers put together!
But still no mention of redundancy. If any of these critical components fails at L4 or L5, what would the vehicle do? This also doesn’t mean that travellers can flick a switch and turn self-driving on yet. Two things are not ready yet: First, the self-driving software still has to be validated before the new features can be enabled, and second, the regulators will have to approve the system. Even if consumers buy the technology today, it is not clear when they will be able to use it.
POTENTIAL IMPACTS
A large body of literature is now available on how fully autonomous vehicles could impact our lives. These range from road safety, through to congestion, public transport, insurance and car ownership. They also cover a wide range of social, economic and cyber security issues.
Impact on urban mobility
The impacts of shared fleets of autonomous vehicles have been studied in a number of simulation projects. At Swinburne, we have used agent-based traffic simulation models to estimate the impacts of shared autonomous mobility-on-demand systems in a suburban context in Melbourne. We started with modelling a base case scenario representing the existing situation with single-occupant private vehicle trips during the morning peak.
We then simulated new scenarios representing different options for autonomous shared mobility. The results showed that this would lead to around 80% reduction in the fleet size, 83% reduction in the space required for parking but at an increase of 30-80% in total vehicle kilometres travelled (VKT) depending on whether the vehicles are car-shared or ride-shared with another passenger. These results are consistent with some findings from international studies.
I should point out that we can achieve these results today if only we can get people to carpool. We don’t need to wait for self-driving vehicles. The key difference with autonomous vehicles when they arrive, is that they will lower the cost substantially because there will be no driver cost. This will create other issues and concerns about the loss of jobs for drivers and this is something that we need to discuss as a community to determine the kind of future we want, and how to address the negative impacts of automated technologies.
The key, then, in the case of self-driving vehicles, is that they will need to be shared – otherwise we risk a nightmare scenario where these vehicles would simply reinforce existing norms. We should also start early planning for these to be used as a form of public transport mainly for first and last kilometre solutions.
Regulations
The final part of this article looks at some of the barriers, particularly regulations. There is huge momentum around this in many countries around the world, including the work that is currently being completed in Australia. I only want to cover an area that has not received a lot of attention to date.
The vehicle brain
The self-driving software that many companies are developing today is based on artificial neural networks and includes millions of virtual neurons that mimic the human brain. The neural nets do not include any explicit ‘If-then’ programming. Rather, they are trained to recognise and classify objects using millions of images and examples from data sets representing real-world driving situations.
There are a few benchmark data sets used today to test the performance of neural nets in terms of detection and classification accuracy. The KITTI data set, for example, has been extensively used as a benchmark for self-driving object detection. Baidu, the dominant search engine company in China, which is also a leader in self-driving software, is reported to have achieved the best detection score of 90% on this data set.
But is this enough to certify that the AI software, the vehicle brain, is safe enough? After all, this test only shows how well the software can detect objects. But we all know that the driving task is much more complex than object detection, and detection is not the same as understanding. Autonomous vehicles should not only detect and recognise humans and other objects, they also have to be able to interact with, understand, and react to their behaviour.
Visual Turing test
A group of researchers have come up with a test based on machine vision, which may be well suited to doing the compliance test for self-driving vehicles. The basis of the Turing Test is that a human interrogator is asked to distinguish which of two chat-room participants is a computer, and which is a real human. If the interrogator cannot distinguish computer from human, then the computer is considered to have passed the test.
But instead of evaluating textual or verbal information, the self-driving test would be based on a framework in which computers would answer complex questions about a scene. First, human test-designers develop a list of attributes that an image might have. Images would be hand-scored by humans on given criteria, and then shown to the AI, without the ‘answers’, to determine if it was able to pick out what the humans had spotted. That’s good news but putting together such a test would not be easy.
The future directions of urban mobility
So where do we go from here? Well, there is already strong momentum around development of regulations and the challenge here is to ensure that they are robust yet flexible and are outcome-focused. Trials and demonstrations will also go a long way to demystify the technology and increase public acceptance.
In conclusion, not everyone will be excited by this vision, and many would be sceptical and disagree that we are at the cusp of a transformation in mobility. Others still want to drive and not everyone is likely to want to rideshare or carpool on a daily basis.
Many might also argue that better investment in public transport would achieve similar outcomes. Whether you embrace or object to these scenarios, the reality is self-driving vehicles are coming and they will have socio-economic impacts and other effects on our society – some good and some bad. I see them as having a role in delivering new shared public-transport-like mobility solutions (first and last kilometre) as part of a holistic approach to improve road safety and promote low carbon mobility.
The real challenge will be to ensure that they don’t simply reinforce existing norms. This will require a shift in attitude in travel behaviour and encouraging commuters to share excess capacity in vehicles of all types.
This is too important to be left only to market conditions and commercial interests. The time is now to think about the kind of future we want to have in a highly-automated world, and ramp up the work today to shape the future directions of urban mobility.
Discover more from iMOVE Australia Cooperative Research Centre | Transport R&D
Subscribe to get the latest posts sent to your email.
A concise and helpful summary. A few comments. One of the key infrastructures, once on moves from a vehicle-centric perspective, is the combination of satellite constellations and wide access integrated ground system with encryption channels safety of life reliability and business-secure APIs.
Coincidentally this is offered by the EU Galileo+Egnos system.
So I’d add to Hussein’s piece a bit more in terms of:
1. What I have attributed to Galileo/egnos style systems as a generic infrastructure. The clouds of LEO satellite systems will offer this globally soon, and will cover non-urban and remote environments … the impacts and potentials are obvious.
2. Mobility is not necessarily accessibility (which is now seeing a resurgence in attention from the person perspective). The latter is a great deal more complex as it also has to include personal perceptions as well as the technical capacities.
I’ve written on these aspects and others that expand the mobility/ technology axis to a wider set of actors for a number of decades, and am delighted to see at least some of these integrative and multidimensional perspectives and technical capacities slowly emerging at last (this is not yet the place to contribute to the pricing issues … distributional equity has been part of my work since the 1960s …).
Thanks for sharing your comprehensive thoughts, Hussein Dia. A couple of complementary considerations:
1. According to my understanding, with regard to the future of mobility, we need to consider the global impact of mobility offerings&technologies over their entire lifecycle including sourcing, production, utilization and disposal. In addition, the perspective must go far beyond technology and include economic, environmental and societal aspects and related side-effects (such as: would people prefer to live in rural areas, if they were enabled to work from home or drive to an urban office quickly and without traffic jams). You may question, if the consumption of resources for construction of individual passenger cars (with any power train technology) as well as the provision of roads and charging infrastructure stands in a reasonable relation to the value add of this concept.
2. Autonomous vehicles’ data is growing even faster than the global data sphere thanks to autonomous test vehicles, which generate between 5TB and 20TB of data per day and vehicle. All this data must be received, stored, protected, analyzed in real-time – and retained for research and legal information. Cameras tend to generate 20 to 60Mbps, depending on the quality of the images that are captured, as well as sonar (10 to 100kbps) radar (10kbps), LIDAR systems (10 to 70Mbps) and GPS (50kbps). The key is to ensure that sensors are collecting the right data and it is processed immediately, stored securely and transferred to other technologies in the chain. This huge amount of data is a considerable cost factor.
3. I miss in the article statements on the protection of privacy of users of CASE mobility. According to Apple CEO Tim Cook there is probably more information about you on your smartphone, than there is in your house. This includes intimate conversations, calls made and received, text messages, photos, videos, contact lists, calendar entries, internet browsing history and personal notes, as well as financial and health data, food preferences, medication addiction, user-IDs and passwords to access to your e-mail or banking accounts and websites like Amazon, Facebook, Twitter and Netflix (to mention only the rather unsuspicious). Do we really want to further erode users’ privacy through CASE mobility?
4. The great importance of 5G in terms of the nationwide introduction of CASE mobility (Connected, Autonomous, Shared, Electric) and the networking of “things” in the Internet of Things (IoT or Industry 4.0) is repeatedly emphasized in the media, among other things due to high data rates of up to 10 Gbit/s and latency times of less than 1 ms. However, the devil is in the detail, as three simple examples show:
▶︎ 5G mobile radio networks require many base stations because the high frequency bands have only a small range.
▶︎ In order to achieve high bandwidths, 5G base stations must be connected to the telecommunications network with fiber optic cables, which is expensive and time-consuming.
▶︎ Extremely short waves are relatively susceptible to interference: Even rain, snow and fog above 10 GHz attenuate strongly; raindrops are no longer negligible as disturbing objects at a wavelength of 1 cm.