Article credit: Written by Nathan Eddy for TU Automotive with contribution from Dr Siddartha Khastgir. Original article here.

Development and testing of autonomous vehicle technology is data intensive, requiring solutions for the management and storage and of rapidly expanding volumes of data.

The primary reason for this explosion is the collection of video and LiDAR data, which has to be high-resolution. Naturally, the more sensors one has on the AV, the more data the AV would generate. An AV sensor suite may comprise of eight cameras, two LiDAR sensors and two radars sensors or it may comprise of just one LiDAR and a couple of cameras.

Depending on the operating conditions, for example the operational design domain (ODD), of the automated driving system (ADS), the manufacturers’ choice of sensor suite may differ substantially. For example, the resolution of data required for a low-speed ADS application in a business park (constrained ODD) may be different to a high-speed application on highways.

“Vehicles are collecting data 24/7 as they create a 3D world via LiDAR and video and there is no easy way to compress data,” explained Ken Obuszewski, global general manager of NetApp’s automotive vertical. “Traditional methods of data reduction don’t work in this context.” He said when it comes to managing this data, it is critical to have rich metadata about the data you have captured in order to optimize the processing of the data, while data tiering and archiving to the cloud are necessary to store and retain massive amounts of data.

Data management methods such as tiering and archiving allow for proper insight into what data gets stored, and what data gets discarded. For storing the data, Obuszewski said that because these huge volumes of data are generated in multiple locations, the challenge is how to provide and access that data where and when you need it. In short – flexibility and locality matters.

“Virtual desktop infrastructure is important, sometimes it’s easier to bring the engineer to the data,” he said. “The flexibility and delivery of data is made more efficient by the ability to create a data fabric across the data pipeline, a way to store and move data from the edge to the core and cloud and back again.” Obuszewski said some of the “heavy lifting”, such as AI processing needs, should be centralized for efficient data movement, noting data must be moved in a secure and efficient manner.

In the next 18-24 months, he sees more data processing being done at the edge, capturing event-based data. He also predicted an evolution towards more efficient use of the data, such as being able to identify and keep only what they feel is valuable and using AI at the edge to both label and identify event-based data to reduce the amount of data being retained, moved, and managed. This includes data intelligence strategies that use AI at the edge and being more targeted and determine what is being captured. “These concepts will make the data more valuable and increase the quality of the data you train against,” he said.

Obuszewski noted cloud technologies would also continue to play a major role, with data being centralized within the cloud, providing compute resources and state-of-the-art tools and applications in the development environment. “The cloud is the environment of choice for the data scientist, because it provides flexibility and availability, as well as long term data and cold data storage, offering cost efficiencies, for example, and legal and compliance needs,” he said. “Automakers are looking for ways to efficiently store this explosion of cold data. The cloud provides that option.”

Dr Siddartha Khastgir, head of verification and validation, intelligent vehicles at WMG, University of Warwick, pointed out that depending on the operating conditions, i.e. ODD of the ADS, the manufacturers’ choice of sensor suite may differ substantially. Khastgir explained the crucial aspect of AV testing is not in understanding the quantity of data but rather it resides in understanding “what data” is useful. “Data storage and management is about storing the ‘right’ data and the challenges are associated within the process of defining the right data and subsequently extracting it,” he said. “Thus, there is now a shift in focus from storing ‘all data’ to storing the ‘right data’.”

Michael Erich Hammer, who leaders the Big Data intelligence team at powertrain development, simulation and testing firm AVL, said at the moment, it is still unclear what the main strategy will be on handling and storing such large amounts of data. “The costs for hot storage of data in the Petabyte range is huge,” he noted. “Therefore, a continuous trade-off between scalability, performance and costs between cloud and on-premise storage has to be expected.”

He noted transfers of large amount of data (>10Gbit/sec created in the vehicle) will not be possible via the network for quite some time. “The workflow contains shipment of physical storage devices from the vehicle to back-office environments,” he explained. “This ‘long loop’ from data creation to information extraction, with potential for delays lasting more than a week, needs to be supported by a ‘short loop’ for live fleet monitoring during operation of the vehicles on road.”

Like Obuszewski he said when it comes to scalability in processing information out of the data, there is “no doubt” about the advantages of cloud-based services. “We expect to see hybrid solutions where workloads are executed on premise and in cloud environments,” he said. “One important point is to first reduce the data necessary to store.”

This means extracting relevant driving situations, or so-called “scenarios”, which are key to validate the AVs. “For example, highly dynamic scenarios with more traffic participants and also vulnerable road users involved are considered more interesting than just simple, constant driving scenarios,” Hammer said. “This includes scenarios where an unexpected behavior of the environment or other traffic participants as well as the AV occurs, would be of great interest.”

Khastgir pointed out that depending on the system architecture for the AV application, the cloud could play an important role in data storage and management, and like Hammer, said the challenge currently lies in being able to get the data off the AV and get it stored in the cloud quick enough. “High speed connectivity solutions like 5G may offer a solution to this challenge, along with data compression methods,” he said. “However, coverage and signal strength of 5G connectivity may pose a challenge for such an approach.”

He added that standards like ASAM OpenLabel, which are focusing on ways to label sensor data, are a big step towards getting the industry to speak the same language for raw data. There are some activities which are taking place (or have just taken place) like Data Storage System for Automated Driving (DSSAD) at UNECE, ASAM OpenLabel and the BSI PAS 1882 to provide some guidance on this aspect.

“However, a more detailed description may be required to help manufacturers overcome data storage challenges,” Khastgir said. “I do see more efforts on both technology and standardization fronts happening for off-board storage.”

Article credit: Written by Graham Jarvis for TU Automotive with contribution from Dr Siddartha Khastgir. Original article here.

Definitions are often important – no more so when it involves safety.

This is why there was a call for evidence in August 2020, and a confirmation by the UK Government confirmed in April that automated lane keeping systems (ALKS) technology “could be legally defined as self-driving, despite insurance industry concerns around safety”, says the Insurance Times. Matthew Avery, director of research at Thatcham Research, explained in the article that “there is still a lot of work needed by both legislators and the automotive industry before any vehicle can be classed as automated and allowed safely on to the UK roads. ALKS as currently proposed by the government are not automated. They are assisted driving systems as they rely on the driver to take back control”.

Clearly defined technological path

He says the Society of Automotive Engineers (SAE) has created a clearly defined technological path to automation, which involves a five-step development of technology can functionality. At Level 1 there is adaptive cruise control and at Level 5 there is full automation which permits a vehicle to go almost anywhere without human intervention and without driving controls for a human driver. He explains that at present most vehicles offer Level 2 assisted driving to enable “an engaged driver to take their feet off the pedals and take their hands off the steering wheel for brief periods of time”.

This may involve enabling a vehicle to stay in a lane, maintain a safe distance from the car in front, and momentarily control the vehicle under the driver’s direction. “A popular example of this is Tesla’s Autopilot function, which is not actually automated whatever drivers and marketing spiel may suggest,” he emphasizes.

At Level 4, automation allows the driver to safely disengage. In certain circumstances, such as on motorways, Avery says the driver could let go of the control to watch films, read a book, or even go to sleep. “The car is in full control, and able to safely navigate any obstacle or hold-up – including roadworks and inclement weather,” he explains before adding: “This Level 4 automation meets peoples’ expectations of automation: e.g. the car can do what a competent and engaged driver can do. The problem with Automated Lane Keeping Systems is that they are Level 3 technology, which is defined as conditional automation.” So, at Level 3, the vehicle can in certain situations navigate and it can drive without the oversight of the driver. This is nevertheless restricted to a certain speed and conditions. It therefore has its limitations.

Safety limitations of ALKS

ALKs can’t move a vehicle between lanes, even in the case of an emergency. The driver always needs to be ready to take back control.

He therefore argues: “This fundamentally contradicts our view of what automation should be, in that the doesn’t replace the driver. The safety limitations of ALKS suggest that the car would not be able to respond to a driver emerging from a broken-down vehicle, or automatically change lanes to avoid debris, nor could it find safe harbor in the event of an incapacitated driver. So, the driver of the vehicle will potentially be exposed to danger and could cause a hazard to other drivers, as their vehicle could be left stationary in lane if the driver fails to take back control within 10 seconds.”

Avery believes the safety questions are compounded by the vehicles ability to operate in specific environments. Automated driving systems such as ALKS may not, for example, be able to read specific UK road signage – leading to traffic rule infringements, which could lead to a driver being fined.

Insurance liability

He says there are also challenges around insurance liability, particularly in fully autonomous Level 5 vehicles, raising questions about whether a ‘driver’ is liable, or whether a manufacturer or technology provider is liable when something goes wrong. This will affect decisions about who pays out for what. So, there will need to be access to vehicle and driving data to fully understand who was in control. Was it the system or the driver at the time of a collision? Without this data it would be impossible to identify fault and liability.

A spokesperson for the ABI comments: “While the insurance industry fully supports the development towards more automated vehicles, drivers must not be given unrealistic expectations about a system’s capability. It is vital that ALKS, which rely on the driver to take back control, are not classed as automated, but as assisted systems. By keeping this distinction clear we can help ensure that the rules around ALKS are appropriate and put driver and passenger safety first.”

Driver education

Siddartha Khastgir, head of verification and validation, intelligent vehicles at WMG, University of Warwick in the UK agrees with the viewpoints expressed by ABI and Thatcham Research in the feature article about public’s perception of automated driving technology being a key barrier to the successful roll-out of ALKS technology. “While I wouldn’t go as far as classifying ALKS as ADAS, I will definitely say that driver education is key to ensuring safe use of ALKS and other automated driving technologies,” he says.

This why the WMG is calling for ‘informed safety’ for the users of automated driving systems, which is about informing the users about the “true capabilities and limitations of the technology to the driver”. Unfortunately, this has rarely been done to date. He adds that there is a need to prevent any misuse of the technology. Problems arise when users overly trust systems which have limited capabilities. So, by making them aware of the systems limitations it should be possible to prevent accidents.

He adds: “One of the fundamental things to be communicated to the user for an ALKS is the constrained Operational Design Domain (ODD) of the ALKS. ODDs are fundamental to safety of ALKS and other automated driving systems.” By creating an understanding of the ODDs, drivers or users can fully understand the capabilities of the automated systems to use them safely. “For example, an ALKS has its ODD as motorways and should not be used or get activated while on urban roads”, he warns.

Thatcham Research has worked with its EuroNCAP partners to define three clear states to enable the average driver to understand automation. Avery concludes that Level 3 systems are open to too many questions, which leading to drivers misunderstanding the capabilities of the systems – citing the incidents involving Tesla cars as an example. He, therefore, argues that automation needs to be abundantly clear and “robust enough to accommodate the vagaries of real-world traffic such as roadworks, fog, or broken-down vehicle.”. Furthermore, vehicles can only be considered as being fully automated when they operate using Level 4 systems.

So, for now, he says Thatcham and UK insurers believe there should be no moves beyond today’s assisted driving technology because it offers all the benefits with fewer risks. He argues that Level 3 automation should become the next ‘genetically modified crops’ scandal, causing consumers to reject the benefits because poorly designed or immature systems could lead to vehicles crashing.

He, nevertheless, concludes that automation will bring significant societal benefits. For them to be realized, he believes there is a need to wait until the systems can cope with all driving scenarios, just as an engaged driver would. With careful attention to defining connected and autonomous vehicle technologies, this will be achieved – safely. By educating defining these technologies, it becomes possible to understand their limitations, and then to inform drivers about them. With a clearer understand of their capabilities and limitations, there will be fewer accidents and insurance claims.

Article credit: Written by Graham Jarvis for TU Automotive in collaboration with Dr Siddartha Khastgir. Original article here.

Traditionally insurance policies focus on the vehicle and its users.

Even with fully autonomous vehicles (AVs) they will concentrate on the vehicles. However, liability may rest more with the automotive manufacturers than with an autonomous car, van or trucks users. What’s missing, says Tony Fish, chief executive of Digital20 and an artificial intelligence expert, is how policymakers, carmakers and insurers cover people outside of them in an accident, such as pedestrians and other road users.

He says there needs to be polarity in the thinking and that it’s not about a chicken-and-egg situation: “We can move away from the safety and protection of the driver, to super lightweight, highly energy-efficient vehicles that protect the public. Stop designing and building infrastructure to mimic what we have accepted, which was a poor idea at the start.”

Talking about when the law will change to protect people outside of the vehicles, he says he’s not convinced that there will be any radical change. In his view this is because there is a need to go through a step-process. He explains: “At the moment we insure the person who’s driving the car and the insurance is for that person inside the car. What we want to reach is for the protection of the person outside the [autonomous] car. We should insure both. Do we first insure that car, or the person driving the car, which is closer to New Zealand’s model.”

“In NZ, the car is insured, so anyone can drive that car, or any car that’s insured because the car has the insurance and not the individual. I suspect that the law won’t change overnight, but the law needs to think about what the journey needs to look like. I think there is no motivation to do it and so the question is why? There doesn’t seem to be timetable, a framework to have it in place in, say, ten years so that insurers can follow the timeline.”

Different models

He rightly comments that every country has its own view out how insurance should work. However, he argues that this is a tad confusing, perhaps exacerbated in the UK by insurance premiums often being paid up front for a year. There is, nevertheless, a market for premium finance and some insurance policies can be paid over the course of the year on credit. Despite this he asks whether the upfront payment model is the right one “in light of time of use-based insurance”. Fish adds that the law doesn’t seem to want to have that conversation, partly because they have built a model based on risk and premium’s being paid up front.

However, Siddartha Khastgir, head of verification and validation, intelligent vehicles at WMG at University of Warwick, points out that in 2018 the UK’s government brought into force the Automated and Electric Vehicles Act (AEVA) 2018. He comments: “The AEVA 2018 takes into consideration people outside the car and insures them in case they suffer damage owing to an automated vehicle being driven in automated mode. This is an important aspect of the Act, which enables automated vehicles to be deployed on UK roads.”

With regards to the safety and insurance cover of people outside of, for example, autonomous vehicles, Khastgir thinks there is a need for a systems approach in order to succeed. That may involve technology development, regulation, business models or insurance. “Therefore, the law not only needs to consider the safety of the people inside the AV but also the safety of the actors (pedestrians, cyclists, other vehicles etc.) around the AV,” he says.

“However, some autonomous may be designed to operate in areas where other actors are not allowed. In such a situation, it may be acceptable to consider the safety of those inside the vehicle only. Having said that, the AV will need to be able to implement a mechanism to continuously detect if other actors are present or not and safely cease operation in case of the presence of another actor in the surroundings.”

Setting standards

In response Fish comments that it’s not just about mandating law. To him it’s about setting standards to protect people. He claims that when we set standards to protect the person within the vehicle, lots of wasted energy is created in terms of designing something that’s not required, including wasted raw materials. This means, for example, a car may reach the end of its life without having been involved in a crash. “The level of waste we’ve created is pretty bloody mad and this has created a race to become bigger not smaller,” he suggests.

From an ethics perspective, Khastgir believes that both the person or persons inside the vehicle and those outside of it warrant cover: “For autonomous vehicle technology to succeed, it needs to be both. It is imperative that the safety of both in-vehicle occupants and actors outside the vehicles is considered in all technical development, regulatory and liability discussions.”

“An AV will interact with its operating environment. Therefore, one cannot design a safe AV without considering the interactions of the AV with its surroundings, for example with pedestrians, cyclists, infrastructure etc.” Subsequently vehicle manufacturers’ safety assurance processes for AVs will need to consider a variety of test scenarios to see how the autonomous vehicle handles oncoming pedestrians, cyclists, other vehicles. He says, as ‘actors’, they have to be part of the operational design domain (operating conditions) of the AV.

Economics of value

Fish nevertheless enquires: “Where are we having the discussion: whose life do we value more?” He believes the conundrum of whether the person in the vehicle if more of value, or the person outside the vehicle is more of value, falls down to the economics of value. “It’s the same question that COVID-19 has raised”, he suggests before adding:

“Both have an economic value impact on society, so why have we made a distinction between those two values. The insurance companies set a value on life, whether you die or kill someone while driving. That person who’s been hit then becomes a cost to society. So, why don’t we make cars safer for the people we hit? That’s the observation.”

Changing behaviour

He wonders what would have happened if seatbelts hadn’t been made mandatory. Would people have changed their behavior? He also considers whether lives would have been significantly saved and whether there would be fewer deaths if we had to drive slower than we do. “We have to deal with the infallibility of the system we want, versus the fallibility of the human,” he remarks.

He concludes that people are far too accepting of what they already have: “We are poor at looking back and asking whether we’d done anything different, would we have had a better outcome. People don’t want to lose their agency; they don’t want to give up driving their cars, even if there would be a better outcome. It’s a discussion that becomes very political, and individuals put themselves ahead of society.”

He also argues that there needs to be more polarity between genders, arguing that even seatbelts were designed with the average man in mind, making women an afterthought. So, from an insurance perspective, and to protect everyone, policies do more than cover the individuals with a traditional or an autonomous vehicle. To gain polarity, it’s ethically right for those outside it to also be covered.

However, this may lead to what he calls Peak Paradox, which he explains “sets off individuals against society and survival against work.” From an automotive perspective, he says driving [the ability to have agency], is put against a society that wants to have safer transport. He calls for a compromise and polarity can’t be gained without one. At the moment he declares there isn’t one but it’s the outcome that everyone should want. At the moment, we’re not heading in that direction.

Article credit: Written by Graham Jarvis for TU Automotive in collaboration with Dr Siddartha Khastgir. Original article here.

Daimler predicts that connected, autonomous, shared and electric (CASE) ‘intuitive mobility’ will succeed in turning the automotive industry upside down.

It thinks a true revolution is afoot, combining each component into one comprehensive, seamless mobility package. The company adds on its website: “For many years, Daimler has been investing in mobility, which goes beyond the actual vehicle. In the future, we will continue to expand this commitment. Therefore, we look forward to innovative companies and start-ups willing to work with us.” So, there will be a brief discussion about the role that start-ups can play in offering automakers innovative solutions.

In a press release Sajjad Khan, member of the board of management of Mercedes-Benz CASE comments: “With our products and services, we play a key role in shaping the mobility of tomorrow – and we are always aware of our responsibility for sustainability. We are convinced that the future is connected, electric, shared and autonomous. Our vehicles and mobility solutions show the course we already have set for tomorrow and you can be sure, there is a lot more to come.”

User-centricity

Siddartha Khastgir, head of verification and validation, intelligent vehicles at WMG, University of Warwick says intuitive mobility is about putting the user at the center of the mobility system. It’s the important to design solutions with the user in mind, right from the start. He argues intuitive mobility can only succeed if it responds to two key factors: desirability and safety.

He explains: “For society to reap the many benefits of new technologies like autonomy, electric propulsion and connectivity, it is essential to ensure the technology can be reliably accepted and adopted. The key to desirability is building trust, while ensuring that that the technologies are safe and used in a safe manner. At WMG, University of Warwick, we created the concept of “informed safety” which enables the development of appropriate trust and also ensures safe use of technologies. Informed safety means that the users are aware of the true capabilities and limitations of the system. While no system, or technology, will ever be 100% safe, one can still reap the benefits of the new technology if we use it appropriately and within the designed operating boundaries.”

Identify operating boundaries

To achieve this there is a prerequisite to identify the operating boundaries, and then it becomes a necessity to convey them to the user in an intuitive manner. To him, this remains a key challenge for engineers. “Thus, the concept of intuitive mobility needs to take a systems thinking approach while combining design concepts, safety and new technology development,” says Khastgir.

However, to Mercedes-Benz it’s also about sustainable mobility – about the environment. The company’s Mercedes EQ release comments: “The outlook is clear for Mercedes-EQ: we understand the limits for our planet, and to ensure innovation. This is why Mercedes-Benz AG has had its climate protection objectives scientifically verified by the Science Based Targets Initiative (SBTI). This means that the company is in line with the requirements of the Paris World Climate Accord. Achieving these objectives requires sustainable products.”

“The CASE Intuitive Mobility concept will change the way public use and interact with mobility solutions,” argues Khastgir who thinks there is potential for change in the vehicle ownership model. Instead of owning a car, for example, users will be able to purchase journeys rather than buying a vehicle for their sole use. He nevertheless adds: “However, that is a utopian dream and not something that will happen in the near term.”

Proliferation of CASE

“In the next few years, we should see a gradual proliferation of each of the CASE technologies, and the introduction of new services and business models”, says Khastgir who adds: “For example, over the last nine months, e-scooters have gained traction in the UK with various trials being conducted.”

He says there are various ‘gigafactories’ being established in the UK and in Germany to meet the increasing demand for electric vehicle batteries. This is because he finds that the uptake of electric vehicles has increased substantially over the last few years. Yet, he underlines that there is some way to go in terms of autonomous, automated technologies.

He therefore comments: “The key to the success of CASE technologies is going to be conducting safe public trials of these technologies to not only introduce them to society but also to understand user interaction with the technology in a real-world setting. Incorporating learnings from such trials is essential for the uptake of these technologies when offered as commercial products.”

Immense challenge

To enable the revolution, start-ups can play a role in helping carmakers to discover innovative solutions. Khastgir warns that the challenge of realizing CASE intuitive mobility is immense. It can’t rely on just one organization alone to find all of the solutions and to answer all of the questions surrounding it. “Based on this, the principle of collaboration is essential if we are to collectively realize the vision. It is no longer an option, it has now become a necessity. In this mobility ecosystem, start-ups play a key role in understanding ever-changing user needs, creating disruptive solutions and, owing to their nimbleness, are even able to adapt their solutions faster than the larger corporate organizations.”

So, the next generation of mobility solutions will see disruption and innovation going hand-in-hand. Khastgir believes that automakers “will need to continue to partner with start-ups to bring in the disruptive mind-set and jointly develop innovative solutions”. Through collaboration, they can bring together the best of both worlds. For example, he reveals that car manufacturers have “acquired various autonomous driving technology start-ups to boost their stable and get themselves a head start”.

Beyond the actual vehicle

Gradually the mobility ecosystem will move beyond the actual vehicle and, as this happens, the software content will continue to increase exponentially. “This is an area where many start-ups are flourishing,” he says. To develop intuitive, sustainable mobility, there is a need to be realistic based on the current state of various CASE technologies.

Khastgir explains: “While the ultimate ambition for CASE will remain a connected autonomous shared and electric mobility service, we are just not there currently. More near-term targets will include electric vehicles with connected applications improving battery range of the vehicles.”

Nevertheless, he predicts that over the next five years there will be some autonomous or automated features deployed in limited operating conditions, which is also known as an operational design domain (ODD). There may also increasingly be applications and services such as autonomous valet parking and autonomous last-mile delivery. They are on the industry’s near-term horizon.

Regarding a post-pandemic era, he thinks the nature of the ‘new normal’ is still unclear. What is apparent is that there will be more people working from home. Yet, there will still be many averse to car-sharing, further disrupting the shared and services segment. Khastgir therefore concludes by calling for the introduction of new business models and further disruption in the mobility ecosystem.

Siddartha Khastgir, Head of Verification and Validation - WMG.

Over many years we have seen the rise of hype around Autonomous Vehicles (AVs), but more recently the reality of the associated challenges has become more visible to the AV community.

One of the key challenges the community is dealing with is to prove that AVs are safe. Some studies have suggested that in order to prove that AVs are safer than human-driven vehicles, they need to be driven for over 11 billion miles. While this seems to be an improbable proposition, a more compelling argument is that using the number of miles driven as an individual metric is not a meaningful way to judge AV safety. Driving billions of miles up and down deserted pristine roads is of limited value if you want to deploy AVs in centre of London or in the notoriously rainy British weather. Thus the focus of evaluating AV safety needs to switch from number of miles to the types of scenarios encountered by the vehicles, which is much more important.

However, due to the unbounded number of scenarios that an AV may encounter in its lifetime, it is unreasonable to expect they will ever be 100% safe on their own. As a result, the concept of ‘absolute safety’ is actually a myth. Having said that, we can still reap the potentially enormous benefits of AVs (e.g. reduced number of accidents, lowered emissions, improved traffic throughput etc.) by creating “Informed Safety”. Informed Safety essentially means that the users are aware of the capabilities and limitations of the AVs. An aspect of Informed Safety involves understanding “conditions” in which the AV is capable of operating safely. An industry acronym already exists for this: ODD – Operational Design Domain. ODD has dominated a lot of the AV safety discussion in recent months.

What is an ODD?

ODD has been formally defined by SAE J3016 as “Operating conditions under which a given driving automation system or feature thereof is specifically designed to function, including, but not limited to, environmental, geographical, and time-of-day restrictions, and/or the requisite presence or absence of certain traffic or roadway characteristics”.

ODD essentially means the “operating conditions” under which an AV can safely drive itself. For example, the operating conditions (ODD) of a low-speed shuttle could include a city centre or a business park, and not a four-lane highway. On the other hand, the operating conditions of the recently announced Automated Lane Keeping Systems (ALKS) will include a motorway and not a city centre. They may additionally include specific lane dimensions and road markings too, along with weather conditions. In order to deploy AVs safely, not only do the AV manufacturers need to define their systems’ ODD, but they also need to convey that to the regulators and local authorities in a way that is understandable to them.

The need for ODD standardisation

As ODD is fundamental to AV safety, there is a requirement to have a common understanding and a way of expressing the operating conditions between various stakeholders. Various standardisation bodies (national and international) have acknowledged this and initiated projects on standardisation of ODD.

One such project was initiated by the British Standards Institution (BSI) who recently published the PAS 1883, which provides a hierarchical taxonomy for ODD definition. The project was sponsored by the Centre for Connected and Autonomous Vehicles (CCAV), HM Govt. and led by WMG at the University of Warwick along with a multi-stakeholder steering group. Other activities in this space include the international (ISO) standard ISO 34503 (led by WMG), which builds on the PAS 1883 work and outlines a high level definition format in addition to making the PAS 1883 ODD taxonomy internationally relevant.

Understanding ODD through categorisation

According to BSI PAS 1883, ODD is comprised of scenery (geo-stationary attributes e.g. roads, traffic lights, traffic signs etc.), environment (e.g. weather information) and dynamic elements (e.g. traffic). The scenery attributes are further classified into zones, drivable area (DA) (which may be roads or pedestrian pathways for low-speed shuttles), junctions, special structures, fixed road structures and temporary road structures. Expanding one of these attributes further, the drivable area is further classified into DA type, DA geometry, DA lane specification, DA signs, DA edge and DA surface.

Awareness of ODD attributes

While having a common taxonomy and language is important for AV safety, it is only part-of the puzzle. From an AV’s perspective, it needs to be aware of every ODD attribute used to define its safe operating conditions (e.g. rate of rainfall, lane dimensions, drivable are type etc.). An AV may get this information from on-board sensing or from external sources (e.g. HD maps, traffic management systems, weather stations or other road infrastructure).

When external sources are used to get the ODD attribute information, it is essential to ensure that they provide an accurate description of the “current” operating conditions. This means that the HD maps providing information about the road conditions need to be kept updated to reflect the changing road conditions. Similarly, any road side infrastructure providing information about lane specifications (closed lanes or lane markings) need to reflect the real-time conditions. ODD awareness is essential from an AV’s perspective as it will allow it to establish whether it is safe to operate or not. If the awareness of the attributes mean that the AV is outside its defined ODD, it may decide to come to a safe stop or hand control back to the driver.

ODD Awareness: The role of infrastructure

With a focus on safety, understanding an AV’s operating conditions relies on data, regardless of whether this is making decisions based on scenery, environment or dynamic situations. To do this effectively, the road infrastructure is crucial. This is one aspect in AV safety that has unfortunately received little attention but it is a pivotal factor in enabling ODD awareness and subsequently AV safety. As mentioned earlier, in order to measure or monitor many of the ODD attributes, an AV might need sophisticated and expensive sensing systems, which could potentially make them economically unviable.

However, infrastructure has the capability of both providing much of that information and also ensuring the information reflects the current state. For example, roadside weather stations can provide information about the rainfall rate to the AV. Similarly, infrastructure can also provide information about any accidents or lane closures or speed restrictions, or any local changes in road surface condition or road markings.

Connected and reliable roads for safe autonomy

Not only will we need to maintain our road infrastructure, but we need to be able to communicate their conditions to enable AVs to decide if it is safe to operate in those conditions or not. Connected road infrastructure will thus play a pivotal role in ensuring safe operation of AVs.

Find out more about WMG's Verification and Validation research here.

Article originally published in the Road Safety Markings Association's 'Top Marks' magazine here.

I am no fan of Lewis Hamilton (due to my Ferrari allegiances), but developed immense respect for him on a personal level when I read about a conversation between Lewis and Toto Wolff (Mercedes Formula 1 boss):

“He (Lewis) once asked (Toto) me: ‘Have you had the active thought that you are white?’ I said: ‘No, I have never thought about it,’ and he said: ‘I need to think about it [my colour] every day because I am being made aware of it.’

As a six-time world champion, Lewis has achieved almost everything there is in the sport. If he is made aware of his colour, every day, imagine what others less privileged than he, feel like or are made to feel like every day. A lot has been said in recent years and (more so) in recent months about diversity; but the question remains, why are under-represented sections of society still forced to deal with the effects of under-representation every day of their lives?

Unfortunately, we live in a society that talks the talk about diversity but rarely walks the walk. Most conversations around diversity have a siloed approach. As a society living in the 21st century, we should be ashamed that we need a campaign like #BlackLivesMatter when by now we should actually have established that Black lives must matter just as much as White lives and all other lives. No one should have to fight to get what is rightfully theirs.

Given that we live in a society which has failed us, symbolism, role models and diversity champions are needed to raise awareness. However, championing diversity without creating a meritocratic environment will reduce that championing to a token gesture and a tick-box exercise, something most organisations are conveniently doing already.

Awards: A Polarising Subject

I volunteer some of my time for an engineering charity. Back in 2014, I floated the idea of a “woman engineer of the year award” to celebrate woman engineers who have reached great heights in their careers despite the societal and organisational hurdles and stereotypes they might have experienced. For me, awards not only recognise contributions but also create role models.

I was a bit naive in thinking that setting up of the award would be a no-brainer for the organisation’s awards committee. Little did I know that a bunch of white men (‘the awards committee’) sitting in a room in central London would decide that creating an award to celebrate women engineers would be discriminatory!

This is even when I got a major British engineering giant to sponsor the award, so money wasn’t even an issue! To my surprise, the more I spoke to engineers (of all genders) in the UK and internationally, the more I realised that this is a polarising topic. Some of the white woman engineers were passionately against the idea of this award. To be fair, they have been fortunate enough to be brought up in families and organisations where bias was limited and people progressed on merit. But my biggest issue with them is that they wrongly (and perhaps naively) assumed that everyone else has been as fortunate as them.

This was six years ago, and to date, the organisation doesn’t have and doesn’t want to have an award for woman engineers!

The Flawed Argument

I was following the recent uproar on Twitter against the Institute of Physics (IoP) following an ignorant blog post by their Chief Executive and the lack of diversity in their honours and awards. I don’t belong to the institute and neither am I familiar with their way of working. But one statement from that blogpost jumped out for me: “nominate more people from under-represented groups for awards”.

At the face of it, the statement makes perfect sense and yes the responsibility lies with us. But let me give you another fact. A recent study on bias in science, of over 1.2 million PhD recipients found that ‘underrepresented groups have to innovate at higher levels to have similar levels of career success’.

Nominating more people from underrepresented groups wouldn’t help when the bar for giving them recognition is much higher than their peers. Isn’t this blatant discrimination? Just having more nominations will become a hollow act and sooner rather than later, people will not be motivated enough. This is an issue with our entire academic world, corporate world and society in general.

Applying Systems Engineering

Just speaking about diversity is a reactive approach to decades of the wrongdoings of bias. While I absolutely believe that we need to be vocal and promote diversity in our teams, organisations and society, the potential impact of all our work could be far-reaching if we are able to create a culture of meritocracy. Recognising people for their contribution (irrespective of their skin colour or race, religion, gender etc.) can be immensely empowering and motivating.

One big thing I have learnt in my career applying systems engineering is that context (by which I mean the environment in which something happens), is important. Actually, context is everything! A command from a controller or a piece of faulty code may be safe or hazardous depending on the context in which they are exposed.

Similarly, circumstances under which individuals’ contributions and achievements were made need to be contextualised. When I say, judge the person on merit, I include their journey towards their achievements (equality is not the same as equity). If we consider just “what” has been achieved, it means we are assuming that everyone in the society has had equal opportunities. And in our current societal setup, I would really like to meet any person who can guarantee that!

There is a need for those at the top of organisations to foster a truly meritocratic environment, and for those within the organisation to celebrate and embrace diversity and create role models. Unless we have this combined approach, there will always remain a glass ceiling for the under-represented and we would have failed to deliver as a society.

Should meritocracy be chosen over diversity?

Truth is I don’t know the answer. However, I believe it’s the trait of a great leader to be able to contextualise the merit (laid on paper) with the fight an individual has had to endure to achieve it, to level up the playing field.

A recent TED post on qualities of an effective team player mentioned hunger as one of the three qualities (the other two being humility and emotional smartness). A few months back, I was speaking with a friend of mine who had to fight throughout her life (due to our societal bias and prejudices), but now that fighting spirit and the drive she has, is her biggest strength. So much so that she wants her son to also undergo some level of hardship to develop that inherent hunger and self-motivation. And I completely agree with her as too much lip service is being made in the name of diversity without creating a meritocratic environment for people to grow or get recognition.

In times to come, our biggest achievement as a society would be if we have created a society where everyone has had equal opportunities and we don’t have to make this decision of meritocracy over diversity or vice versa. Diversity will then be a by-product of meritocracy. However, until then, we need to ensure that we create and celebrate role models who have cut through the hypocrisy of our society and achieved due to their hunger and desire.

An appeal to the research community and industry

When we evaluate applications (be it for awards, research funding, and recruitment), we need to evaluate them in the context of the society we live in and the obstacles the applicants have had to overcome. The academic world is especially poor at this. Unfortunately, the ‘system’ doesn’t really allow for considering what people have experienced. However, the UKRI Future Leaders Fellowships is a scheme which I think does that as it is so focussed on the individuals. And I am not just saying that as a chosen fellow. I still vividly remember a UKRI FLF video (from last year) from Professor Fiona Watt (Executive Chair, Medical Research Council), saying this scheme is for “people who don’t fit neatly into boxes!” Never before, have I heard something like this in an academic world. Kudos to UKRI for creating such a programme.

Before making a decision on applications, just imagine what they could achieve if they were given the same resources and infrastructure (because of their fighting spirit) … If we do this, the number of awardees/lecturers etc. from under-represented sections of the society would increase dramatically, which in turn would inspire many others. However, this is a short term measure.

In the long term, we need to take both a top-down and bottom-up approach to get over the decades-long wrong of ‘biases’. Recognising that the playing field is uneven and that we are not all coming from the same starting point and appreciating this in our decision making would be a major step forward. I hope we can create research and organisational culture which has its foundations on meritocracy enabling diversity.

It is a marathon and we need a few 100m sprints wins to keep our spirits high to drive the long-term commitment to a holistic approach….

................

I must admit, I wasn’t sure how this blog will be received. I was so worried (borderline scared) that I reached out to a few of my friends and colleagues to get their opinion to gauge if I should even publish it. Big thank you to Professor Paul Jennings (WMG), Dr Poonam Goyal (my wife), Corrina Urquhart (WMG), Dr Izzy Jayasinghe (UKRI Future Leader Fellow, University of Sheffield), David Webb (CCAV, HM Government), Michaela Hodges (WMG), Jennie Martin (ITS UK), Nicola Jennings (University of Bath), Luke Logan (BAE Systems), and Dr David Bott (WMG) for their feedback on this blog.

I am an Indian national and a bit outspoken (especially on matters I deeply care), and have been told that if I do the right thing, I would have a good night’s sleep (however long or short it might be!).

This blog represents my personal opinion and not that of my employer or any organisation I am (even remotely) associated with.

**Original article written by Siddartha Khastgir, published to Medium here.