CAV

Protect and survive


Automated Driving Systems are an interesting example of the application of Artificial Intelligence to transportation. Bob McQueen explores some of the associated challenges and opportunities

Artificial Intelligence (AI) is of growing interest in the transportation field. The ability for information technology to provide decision-support and perhaps even take over some of the human functions associated with transportation is both intriguing and concerning.

As the capabilities of AI evolve the focus moves from menial, repetitive tasks to more sophisticated jobs that were once thought to be the exclusive domain of the human. Tasks that require perception and judgment are coming within the reach of AI and automation. I mention these two things in the same breath as I believe it is necessary to be smart, and also capable of implementing decisions based on the new insight that comes from being smart. It’s not good enough just to know about prevailing conditions, threats and opportunities, you have to be able to react appropriately.

“A smart city requires great sensing abilities to identify threats and opportunities. It must also have the intelligence to react appropriately in the light of the information derived from the data”

As I often joke with my clients and partners that “You know that your city is smart if you poke it with a stick and it reacts appropriately”. This may seem like a lighthearted or even flippant approach to defining a smart city, but the statement is embedded in some practical truth. A smart city requires great sensing abilities to identify threats and opportunities. It must also have the intelligence to react appropriately in the light of the information derived from the data. It is also crucial to move efficiently from data to information to insight to action. Currently, the judgments and decisions based on the insight coming from sensing is conducted by a human operator or manager. Typically, the human considers information coming from multiple sources to inform decision-making regarding appropriate strategies and responses. In future, decisions may be made autonomously by a sentient management system.

Like most new technology developments, connected and autonomous vehicles are being implemented in a step-by-step rollout. While full automation remains a mighty leap, new vehicle features include the fruit of Automated Driving System development programs. Features such as self-parking, lane keeping, and distance-keeping are being offered on the latest car models.


This also includes connected vehicle technology that enables a two-way exchange of information between the driver and back office. General Motors has been a pioneer in this respect with the OnStar system which has been available since 1996 and has had a major impact on safety and user experience. More recently Ford have introduced the SYNC system that supports two-way communication between the vehicle and back office.


This tells us that the introduction of Connected Vehicle technology in advance of Automated Driving Systems can move us towards our safety, efficiency and user experience goals. Connected vehicles can also have a terrific impact on transportation planning and operations by providing probe vehicle data regarding origin, destination, instantaneous vehicle speed and vehicle ID. This basic set of data can go a long way to improving the transportation community’s understanding of prevailing transportation conditions and the fluctuations in the demand for transportation within urban areas. Probe vehicle data can also make a considerable improvement to advanced traffic management systems such as traffic signals and freeway management.

“The progress being made in automating the vehicle is pushing the boundary of Artificial Intelligence from decision-support to complete automation”

Moving the conversation back to autonomous vehicles, more typically referred to as Automated Driving Systems these days. It is obvious that the progress being made in automating the vehicle is pushing the boundary of Artificial Intelligence from decision-support to complete automation. AI provides the ability to sense the huge variation and prevailing conditions witnessed by drivers. This is combined with automatic control of the vehicle to offer an effective alternative to human operation. It is worth pausing at this point to think about what we consider to be a successful outcome of the application of advanced technology to replace the driver.


Safety is the primary goal, as nearly 1,250,000 people die in road crashes worldwide, each year. According to the US Department of Transportation approximately 80% of crashes can be avoided through the use of Automated Driving Systems. Internationally, there is a multinational road safety project known as Vision Zero. The project provide support for a strategy to eliminate all traffic fatalities and severe injuries while increasing safe, healthy, equitable mobility for all. Automated Driving Systems will play an important role in moving towards Vision Zero.


Which brings me to an interesting point regarding a successful outcome of Automated Driving System introduction. Are we seeking perfection from automated driving or just a substantial improvement over the human? Recent experiences suggest that Automated Driving Systems are not infallible, but does this mean that we shouldn’t press ahead at full speed to gain safety improvements that are currently achievable? This brings to mind an old saying: perfection is the enemy of the good. While safety must remain paramount, at what point do we start missing out on safety improvements by not implementing because the system is not perfect just yet?

While working on a federal project a few years ago a system engineer told me that safety and security can be considered as two different perspectives on the same issue. Safety involves protecting the external world from the effects of the system. Security involves protecting the system from the effects of the external world. A comprehensive approach requires that both perspectives are thoroughly addressed.


Taking a look at the security perspective, it seems to me that the transportation community has made an important shift in considering the importance of system security. In the past, transportation systems have flown under the radar, working quietly in the background with a reasonably low level of threat from attacks on the system. However, as our transportation systems become more prominent and as they become successful in improving safety, increasing efficiency and shaping the user experience, then their profile rises.


As more people become aware of the silent systems operating in the background, the threat increases. Automated Driving Systems are at the pinnacle of new high-profile transportation systems. It is important that we incorporate the very best practices with respect to system security. It is obvious that these will be drawn from beyond transportation from other markets and disciplines were lessons have already been learned and tools have already been developed.


I’m not an expert on system security but would offer the observation that it will be important to consider the entire chain of connections from the vehicle to any roadside infrastructure, on to a back-office operation and beyond. This should also include elements of the chain that may not be obvious on first look. For example, Michigan DOT has examined the legal consequences associated with the maintenance of connected and autonomous vehicles. What liability does the car mechanic have if the autonomous vehicle fails in the field after maintenance under his supervision?


There is a related question regarding what security procedures are in place when the autonomous vehicle resides in the maintenance workshop and is accessed by a range of electronic tools for maintenance purposes. Are these tools also subject to the same high and consistent level of security as the other elements in the chain? It is obvious that it’s not enough to have a high level of protection for the individual elements in the chain of connections, such as the vehicle, the telecommunications network and the back office, but a systemwide approach must be taken to develop an appropriate level of security for the entire chain or system. This could be a challenge if different organizations are responsible for different elements of the connection chain.


Processes and procedures within the back office also need to support a consistent level of security and I would note that this should include physical as well as cyber security. There is not much point in having a highly sophisticated cyber security system if someone can simply walk up to a computer and gain access to the system. Of course, this applies to Automated Driving System-equipped vehicles also. As there will be millions of these, the risk of unauthorized access is high.


A few years ago, the Dutch government, leaders in congestion pricing research at the time, noted the change in the risk profile when gas tax is replaced by a distance-based congestion charge. The collection of tax revenue by oil companies involves a relatively small number of collection points that are already protected by point-of-sale technology. There is a small number of companies involved in collecting the tax and sending it to the government.


For vehicle-based taxation, there will be a significant increase in the number of collection points (every vehicle) creating a higher risk profile. It is also obvious that the whole cash flow of the operation of the operation is turned on its head. The revenue is collected locally and then sent to the government rather than being collected by the government and distributed locally. That could be the subject for a whole book never mind an article.

“There is a related question regarding what security procedures are in place when the autonomous vehicle resides in the maintenance workshop and is accessed by a range of electronic tools for maintenance purposes”

In addition to security issues, there are privacy concerns associated with Automated Driving System-equipped vehicles. The ability to track vehicles and exchange data between the vehicle and the back office introduces a new set of challenges with respect to the rights of the individual to privacy. There is an interesting conundrum associated with this. On the one hand travelers expect higher and higher levels of service, focused on an almost customized set of services, based on a detailed awareness of their current needs, as well as prevailing transportation conditions. On the other hand, many people expect to remain anonymous. The balance is vital to the success of future transportation operations. Customization of services is also a particularly important aspect of transportation service delivery these days.

“On the one hand travelers expect higher and higher levels of service, focused on an almost customized set of services, based on a detailed awareness of their current needs, as well as prevailing transportation conditions. On the other hand, many people expect to remain anonymous”

We need to dispel the illusion that transportation customers are influenced solely by public transportation service providers. The reality is that transportation customers’ service expectations are being set by organizations that deliver services related to transportation and beyond. These include Uber, Southwest Airlines, Netflix, Amazon, AirBnB and FedEx, to name a few. Transportation customers experience customized service from these organizations and many more. This sets the benchmark for what they expect from transportation service.



There is no question that Automated Driving Systems will represent a challenge with respect to security and privacy. The availability of superb big data management and analytics technology, combined with access to a wide number of data sets, means we have the ability to circumvent privacy by combining data sets to reintroduce personal information that has been redacted. As my grandmother used to say, “just because you can do it doesn’t mean that you should do it”. This is where the role of data governance becomes crucial. There are proven approaches to the creation of data governance plans that define who has access to the data, how the data will be stored, how the data will be protected and define things that you will and will not do with both the system and the data.

“We need to dispel the illusion that transportation customers are influenced solely by public transportation service providers. The reality is that transportation customers’ service expectations are being set by organizations that deliver services related to transportation and beyond

In spite of all the challenges, I am a big proponent for Automated Driving Systems. In my opinion this is the best time to be in transportation as new technologies are enabling new insights and new capabilities. Such is the value of these new technologies that I believe it justifies the effort to overcome challenges and barriers.

Sometimes our reaction to change is initially negative until a deeper understanding of the value and benefits convinces us otherwise. I don’t believe that there are any unsurmountable barriers associated with the introduction of Automated Driving Systems. Diligent and comprehensive planning and design should enable us to reap the full benefits of this amazing technology, while overcoming the challenges and managing any undesirable side effects.


I look forward to the day that my car will be a better driver than me.

FYI


Bob McQueen is principal of Bob McQueen & Associates and North American Bureau Chief for H3BM

Share now...