Smart Moves: Exploring AI's Impact on the Future of Mobility

Over the past five years, the mobility sector has witnessed an unparalleled surge in the adoption of Artificial Intelligence. However, alongside this remarkable progress, we must recognise that significant technological advancements bring not only opportunities but also challenges. 

In this episode of the Anticipate Podcast, Terry Smith, Intelligent Mobility Lead at WSP Middle East, is joined by Dr Raj Kamalanathsharma, Intelligent Mobility Advisor, to unravel the intricate web of AI-powered innovations and discuss the challenges and opportunities they present. The talk also delves into envisioning a future where mobility seamlessly integrates with advanced intelligent systems.

Want to hear more insights from leading industry experts? Subscribe to our podcast via Spotify, Apple Podcasts, and SoundCloud to stay up to date.

Transcript

0:00
 
From autonomous vehicles and smart traffic management to logistics optimization and predictive maintenance artificial intelligence is revolutionizing how we move people and goods. Over the past five years, the mobility sector has witnessed an unparalleled surge in the adoption of artificial intelligence. However, alongside this remarkable progress, we must also recognize a significant technological advancements bring not only opportunities, but also challenges, it is crucial to acknowledge and address potential negative impacts that accompany these transformations. My name is Terry Smith, intelligent mobility lead at WFP in the Middle East, and in this episode of anticipate podcast, I'm delighted to be joined by Dr. Raj Camelon Sharma, intelligent mobility advisor. In our talk today, we want to unravel the intricate web of AI powered innovations, discuss the challenges and opportunities that present and visit a future where mobility seamlessly integrates with advanced intelligent systems. Hi, I'm Raj, and welcome to the anticipated podcast.

0:57
 
Thanks for having me, too.

0:59 
 
So I suppose we will start maybe with a quick introduction, I think it's fair to say that in the world of mobility and AI, it's quite a niche area. Most of us working in this space have had quite interesting career journeys, typically from an engineering background, and eventually we land in the world of mobility and AI. I think it'd be great if we could maybe give our listeners a brief introduction about your background, your career today, and perhaps your current role.

1:25 
 
All right, well, I was a civil engineer, before a pivoting to mobility space around 15 years back. At that time, intelligent transport and mobility Analytics was still growing. So I did a master's and PhD from Virginia Tech us, focusing specifically on intelligent signal control and vehicle control for connected and automated vehicles. It was still growing at that period, around 2008 timeframe. Being part of the Virginia Tech transport Institute's research team really gave me a chance to indulge in algorithmic development, testing, implementation, and etc, of several Intelligent Transport Systems. Suppose that I stayed there, did some more research and then moved to Booz Allen Hamilton, where I was leading their transport analytics center for over four years that I've supported clients like US Department of Transportation, US Postal Service, etc. In their transition to connected and automated vehicles. In 2018, I moved to Dubai, I was part of RTS Public Transport Agency, I was the chief data scientist and spearheaded agency's Big Data and AI program. During that time, we had around 20 Odd initiatives every year from a broad range of visualizations from sentiment analysis, image and video processing to detect anomalies, predicting operational performance, etc. To improve overall Dubai mobility landscape. Since last year, I've been advising mobility leaders in the region on intelligent transport systems and initiatives, specifically around analytics and AI. I'm still working with Dubai Roads and Transport Authority, and also work with neon land mobility, UAE Ministry of Energy and infrastructure on the same topics.

3:08

Thanks very much. We have a very impressive journey into space to date. I suppose if we if we start to look at what is AI and the benefits, I think we've all heard of the term AI in its simplest form, you know, in our day to day lives, we see you know, our use of Alexa for voice commands, or Alexa learns or voice recognizes and carries out commands. on our behalf. We see Netflix provides personalized recommendations, or indeed, when we interact with chatbots for customer services, as most kind of public services now have an interactive Chatbot. But when it comes to the mobility sector, would you perhaps explain how the concept of AI and its relevance in the mobility sector?

3:50 
 
Sure, AI in a nutshell, is actually just a simulation of human intelligence in machines. So in my opinion, mobility sector has a lot to gain from embracing analytics and AI. And this is primarily because, you know, the people and goods movement have a lot of variability in terms of the time they move space, true purpose mod route compliance, and several other factors. And this variability makes sensing and decision making very challenging if technology is not used. That's why we can see that technology has been part of any new traffic management systems for at least 2030 years now. So AI is currently used in traffic management specifically to sense and recommend the right traffic management plan. It's also used in incident management to predict traffic incidents and hotspots, public transport planning and operations for applications such as predictive dispatch, demand responsive transit, and one of the most widely used application of AI is in the shared mobility space. We all use Uber and such e hailing apps. So a location of the vehicles around hotspots where the demand is analyzing driver and passenger behavior and enforcement through you know, anomaly detection. All these are some of the applications of AI mobility, and it's very relevant to mobility.

5:14
  
Thanks, Raj come if we maybe delve a bit deeper down into debt, assign a VA and this is no subject we've had a lot of discussions me and you over the years offline and various projects. It's fair to say I think most transport agencies are most cities to produce a wealth of valuable data from the various transport modes. If we take the Dubai example on deed, you know, any capital city the multimodal network typically includes you know, metro, tram bus taxi. And of course, we obtain real traffic data from various sensors deployed on the network. And often sometimes data from from third party providers who provide you know, GPS data for the road network. Typically, or often this data is stored, you know, in silos, often in data warehouses is typically used for offline or post event analysis standard reports. So it's often the case where the user manual may run the report daily or weekly. But what about unstructured data? So when we consider unstructured like, you know, CCTV, video images, social media feeds, sentiment analysis? How can transport agencies make better use of this data, and deploy AI solutions that can extract value from this valuable asset?

6:28

You're right, I mean, transport data is very diverse. It can come from any of the systems like the AVL systems in public transport vehicles, ticketing system, fleet management systems, and roadway sensors. And we also recently have a suite of new sources of data, primarily around mobile phone data services grow people movement data from their mobile phones, it's also becoming very common in the transport decision making cycle, of course, due to the diverse nature of the data. But one of the key aspects of making them useful is standardization. And to some extent, it's done, for example, the GTFS real time feed is standardized for public transport, the mobility data specification for shared mobility, and there are protocols such as TCP IP and Qt T, etc, for, you know, traffic control data, IoT sensor data, etc. But given that we are in this region, and the standardization is specific to different locations, we procure product from around the world, some things from us some things from UK, some things from China, etc. So we do have a mix of standardized and non standardized data feeds that's coming to the transport agencies in this region. And therefore, sometimes there is a lot of processing that's required prior to using them to be structured or unstructured. Some of the processes that we do is data cleansing, data interpolation smoothing, and even internal standardization. So for example, domain, Dubai may have its own standardized schema that may not work well, in Riyadh or so and so forth. So I would say this is one focus area that agencies and standardized standard organizations in this region needs to work together and develop specific standards.

8:22

Thanks version, I definitely agree in your thoughts on you know, ensuring that the data to be used, you know, it's of good quality, it's cleanse, and it's readily available. Prior to any specific use case development. From previous experience, it's often the most time consuming part is making sure the data is of good quality and that you can get the data even though it may be in the same organization. Getting that data, I suppose into one, you know, central repository is often want to time consuming parts of it. If we look down, I suppose at autonomous vehicles, it's clear that you know EVs will form a key part of the future of mobility be it you know, autonomous public transport such as the Dubai Metro, which has been operating, you know, seamlessly since 2009. Here in Dubai, are indeed self driving cars. Would you perhaps explain to the audience the role of AI in self driving cars, and what what benefits such AI can bring to drivers? And I suppose to order you know, let's call them road users.

9:22

Sure. In general, an autonomous car is actually just a moving rack of technology. And it has two main components a sensor suite that detects and predicts the movement of objects around it through leaders, radars and other sensors and the decision making system which paints tactical and real time decision constantly, such as what speed to drive what, what lateral deviation to use within a lane, which lane to drive on, etc. This gets very complex when you have to share street with conventional cars, pedestrians and other vulnerable road users. But in general, the benefits that we have from autonomous vehicles is very clear. They are more safe because obviously autonomous vehicles, they don't get distracted like us humans and economic efficiency, because you know, when open autonomous vehicles drive itself, we can use our time for something else for our work to watch Netflix or whatever, there are other benefits as well, that's highlighted in many research such as reducing traffic pollution, etc. But that all us depend on, that all depend on, you know, whether we provide a dedicated space for self driving vehicles to work or not. See, the current roadways are designed for human drivers, and the heterogeneity of human drivers will always look at the potential for self driving cars.

10:39

Yeah, I think you make a very good point on, you know, current roadways being designed for human drivers, not particularly for autonomous vehicles. You know, I think that may be a key factor in the mass deployment of autonomous cars. I know there's been a number of instances of driverless cars been involved in accidents or incidents during during trials. Many of these are attributed to the car ride, and you're misinterpreting the environment that it senses or seeds, or indeed, the cars encountering a scenario which has not been programmed. You know, I think I think personally, the human always has will have a part to play going forward, we have our natural intuition and ability to respond to any given scenario. So yeah, I think there's still a way to go before we see mass deployment of fully autonomous cars, but it certainly it's coming. I think there are many benefits to integrating AI into mobility systems, you know, board retrofitting to existing systems, or but also including in new mobility systems and solutions, would you be able to maybe outline some of the key benefits of integrating AI into mobility systems. Of course,

11:48
 
one of the obvious benefit from using AI in transport is the efficiency. So, as I mentioned before, transport is a complex ecosystem. And taking all the data into consideration to make intelligent decisions, whether it's traffic management, whether it's public transport control, requires a lot of computational capabilities, algorithms that learn from our behavior, etc. So, if we have an AI based system, obviously, you will improve the efficiency. For example, we once had fixed route fixed schedule public transport, and we can see that at times versus are too full or too empty. Now with demand responsive transit that runs on AI, we don't have that we have reduced inefficiencies. Now without higher efficiency comes lower cost. For example, AI based public transport systems have shown to have reduced cost by up to 10%. This is primarily by reducing that debt kilometers where you know the segments where we don't need to run buses or we need to run smaller buses, etc. Or intelligently routing public transport and taxis to areas with higher demand. There are several examples of this. Around the world where AI systems can sense and adapt so that our operational costs are lower. One of the other key example, especially when it comes to self driving vehicles, as well as driver assistance systems is reduced accidents and improving road safety. For example, some estimates show that full deployment of autonomous vehicles, we will see up to 85% reduction in traffic crashes. And you can already see that some of the systems such as forward collision warning, adaptive cruise control, etc. have already proven to reduce accidents that are caused by distracted drivers. The last one, I would say is the use of AI and enforcement systems such as use of traffic cameras, fatigue monitoring system, driver risk scores, etc. This is also aimed at improving road safety and and improving compliance basically.

13:56

Thanks very much for those insights. And maybe would you be able to provide some real world examples of success stories, or AI has made a significant impact in the mobility industry.

14:07
 
There are quite a few around the world. Some of the most common ones, I would say is in the field of sensing sensing of vehicles through traffic cameras, or plate number recognition systems, sensing of fatigue in drivers by fatigue monitoring systems, sensing of wrong way parking by enforcement system, sensing of cars that didn't pay for parking through a parking violation systems etc. So I would say sensing is the biggest or most widely available use case around the world. And there are several examples for that. And then it comes to the examples where decision making by AI is there. And for example, digital twins specifically associated with active transportation and demand management. These are systems that simulate different traffic management plans across A city and chooses the right one based on train data. Once that system is city brain, for example, it's widely adopted in over 23 cities in China. Another obvious example, as we discussed before, is the autonomous car. There, again, coming to almost all cities by the end of this decade, including the bison, shared mobility companies as well, like Uber, Lyft, etc. They use AI to predict demand, relocate vehicle, and also to use them to understand what kind of risk scoring drivers have, what kind of fraud scoring passengers have, etc. One other main example I've seen in most public transport organization is predictive maintenance. Increasingly, buses and trains are coming up with IoT based sensors to understand the vibration, the temperature, etc. And these sensors are used to predict failures before they happen during operations. Examples are, for example, Trenitalia, they have a predictive maintenance system developed by SAP, or there is a UT implementation that was done maybe a year back in London Underground trains.

16:14

Thanks, Raj. So we've talked about what is AI some of the many benefits that AI brings to mobility. But we must also consider the challenges and particularly the ethical considerations that this technology brings. It's not all good news. Recent research has outlined that AI revolution, particularly, you know, the use of tools, like chat GPT may result in job losses down the road. It's not quantified. But we've we've been pre warned that, you know, job losses may come is something particularly as mobility professionals that we should be concerned about.

16:48

Well, like other jobs, AI might replace jobs with new ones. The example I usually use in the context of mobility is the traffic police. So perhaps maybe two, three decades back intersections, even if they have a signal system, they used to be manned by police, who should stand in sun pollution. And to make sure that intersection is clear, and people follow the traffic rules. In some cities, the practice still exist. But easily this is something that we have started and we can avoid by having some sort of violation detection algorithms, a cube at intersections, they can give to a high level of positions, who is what making the violation and what action is required on them. So now this job is not necessarily gone. But it's replaced, because now there is a new system in this intersection. So we need the technicians for that we need remote inspection engineers or inspectors who need to look into it. So obviously, when some jobs are gone, new jobs come out, most of them will be technology driven. And specifically in the context of mobility. I think, in the near future, there won't be a reduction in mobility based jobs.

18:07

Okay. And and if we, I suppose, if we looked at data privacy, and the protection of personal information, this is obviously utmost importance, not alone, you know, in traditional IT systems, particularly in implementation of AI solutions, which often rely on new sensors to detect and track users. I think it's fair to say that quite a lot of mobility AI solutions do rely on images, typically CCTV cameras, with, you know, analytics running on top of them. Number periods are often captured as people's faces are on view. people and vehicles are trapped. So we understand movements of people and goods around the city. Would you be able to elaborate for us on some of the the ethical concerns that AI brings?

18:49

Yes, as you rightly mentioned, I think one of the main use of AI currently isn't sensing whether it's sensing vehicles, or sensing commuter movements or understanding the origin destination where people travel within a city using track readers. This is literally the one of the main use of AI. And in traditional sense, that traceability so understanding if it's the same person going from point A to point B and then to point C, this traceability is of importance as well to mobility professionals, so that we can improve the service, but the identifiable information base, the ID ID, this is not of that importance. And rightly, most of the privacy protection laws, they mandate removal of the personally identifiable identifiable information at the source itself. In addition, most Mobley mobility analytic data providers, they also use techniques to make sure that data is anonymized. They do have techniques such as trajectory to trimming some providers trim the first 200 To 300 meters, and the last one to 300 meters of trajectories, aggregation and other algorithms to make sure that the data is not traceable to a person. Of course, with advent of AI, this challenge is only growing. For example, GDPR is one of the most comprehensive protection laws, and even GDPR mandates data collectors to give the ability to users to request removal of data. But this doesn't mean that the data is removed from the algorithms that are trained on it. So definitely, as data protection laws, we are not there yet to deal with AI. But some of the basic protection, such as removing identifiable information is still there.

20:47
 
Thanks very. So we know obviously, there's data privacy and protection issues, potentially, in some of these AI solutions. How can we perhaps address some of these concerns to ensure responsible and equitable implementation of AI technologies in the mobility industry?

21:04
 
Well, that is where humans come into play. See, for it's important for us as mobility professionals to understand the can do and cannot do some of the systems that we implement. Algorithms are often blinded by the data they have. I have had a project where an AI algorithm or bus planning, it kept suggesting stops to remove from operations, because they had very little ridership, because the algorithm was trying to reduce cost. But on the long run, this will only result in one of two things. One, perhaps that bus stop may have been a livelihood for a family, or worse the algorithm over time, it will reduce the public transport coverage altogether. So hence, it's important for us mobility professionals to understand how the AI operates, whether they are getting the full picture of what's on the field, and more importantly, whether the decisions that they make match our expectations and experience.

21:59

Thanks, Raj. And now we're supposed to the million dollar question, what is the future of AI in mobility? Obviously, very hard question to answer. We know that AI is here to stay, the pace of development within the AI industry is unprecedented. We're seeing new tools released weekly, fantastic tools, they charge GPT. But how do you see this technology impacting our day to day mobility in the future? Do you see the day where we are all being driven around and you know, level five, fully autonomous vehicles driverless? A day where accidents don't happen, as AI is controlling the entire network? And the decisions are being taken away from the human? Or is there a place for human control and intuition?

22:43

I think yes. And I'll tell you why I think that see mobility 4.0 was the period 2010 to 2020, where technology was deeply used in mobility, but it was not essential to its existence. 2030 and beyond, we will have mobility 5.0, which is a timeframe when technology becomes essential to transport and mobility. Because we are moving towards smarter electric cars, intelligent Mobility Systems demand responsive shared mobility modes, we will need technology to move us around past that. If this growth continues by 2050, we will be in a fully tech driven mobility ecosystem. And I expect that we would have cities or at least districts where the mobility system design itself will revolve around advanced technologies such as connected and autonomous and shared demand responsive electric vehicles. So yeah, I think yes, we'll be there, maybe three, four decades in the future.

23:42

That's good to hear to still play for us humans. To thanks, Raj. I think we've probably had our time limit on the podcast, but it really has been a fascinating discussion. We've learned that AI holds immense potential to revolutionized the mobility industry, it already is, and they can see this, you know, being enhanced forward in the future. It provides enhanced safety, efficient traffic management and personalized travel experiences. However, we've also discussed the many challenges that come with integrating AI such as data privacy and the need for robust security measures. So I just like to thank Dr. Eyes for joining us today and taking the time to share his expertise and answering our questions and all things mobility and AI and to our audience. Thanks for listening all the way through. please do leave us a comment if today's episode has sparked her interest. And don't forget to join us again in two weeks for a new talk. Stay tuned.