One way of accelerating the deployment of autonomous vehicles, explored by Graham Jarvis.

“Dramatic benefits are expected to be reaped by society because of the advent of autonomous vehicles (AVs) have a big caveat which assumes a 50% to even a 90% usage penetration of autonomous vehicles.” That’s the view of Swaminathan Gopalswamy, professor of practice and director of the connected autonomous safe transportation (CAST) programme at the Department of Mechanical Engineering, Texas A&M University.

Societal benefits

According to the Eno Report of October 2013, the societal benefits include:

·         Dramatic benefits from autonomous vehicles is in terms of lives saved per year (with a 90% penetration, 21,000+ lives saved just in the US alone;

·         Fewer crashes (4.2M), economic cost savings ($100+Bn);

·         Travel time savings (2700+M Hours);

·         Fuel Savings (700+M Gallons), etc.

The caveat is that these benefits are based on significant penetration of AVs (in the above example, 90% penetration)

Techno-business proposal

So, what is he proposing? “It’s a techno-business approach and concept for autonomous vehicles that we believe would accelerate the deployment of autonomous vehicles,” said Gopalswamy. “Indeed, it is overly optimistic. In fact, that is the premise of our concept! It’s not realistic to expect anything like this. In the US are 250 million vehicles on the road, and so we are not going to be able to switch them off and then turn on autonomous vehicles.”

 So, it’s going to take a really long time. Autonomous vehicles will be limited to begin with. There will be different levels of autonomy in 2025 but not sufficient to achieve the full societal benefits of autonomy. “My expectations are that, if we go the way we are doing it today, sufficient penetration will happen perhaps in 50 years’ time. If we say we can leverage the infrastructure much more, then this could make a significant difference in commuter-traffic areas.”

Mindset change

“There is a mindset change that is required. Infrastructure is led by government agencies – public organisations. There has to be an incentive for private parties to jump into the game. New players that can come in from a commercial angle to drive infrastructure and the economy. The concept that we are making is infrastructure-enabled autonomy, requiring new players, who will be driven by business value propositions and not governmental policies.”

This demands the partitioning of the intelligence required to drive autonomously and it is subject to how this intelligence is distributed among the key autonomous vehicle players: e.g. the carmakers, the infrastructure providers and third-party players. Furthermore, to disseminate his new concept he is working with his colleagues on a report. To give you a taster of what’s to come once it has been published, he has shared with TU-Automotive part of the report’s abstract and introduction – all of which appears under the title of Infrastructure Enabled Autonomy: A Distributed Intelligence Architecture for Autonomous Vehicles.

Current approaches

Gopalswamy et al claim, for example, in the report’s abstract that the current approaches to autonomous driving, whereby the automotive manufacturers are required to shoulder most of the responsibility and liability associated with “replacing human perception and decision-making with automation, are slowing down the penetration of autonomous vehicles. In turn this means that the societal benefits of autonomous driving aren’t being realised as quickly as they could be. They argue that the carmaker’s responsibility and associated liabilities must be re-balanced, and that by sharing them with other players within the autonomous driving ecosystem it may become possible to accelerate the deployment of autonomous vehicles. This requires infrastructure enabled autonomy (IEA) and the deployment of a framework that is akin to a Bayesian network model for assessing the risk benefits of such distributed intelligence infrastructure.

The additional benefit of such an approach, according to Gopalswamy and his colleagues, is that it enables an autonomy-as-a-service model, while still permitting the private vehicle ownership.  Furthermore, under the current regime, he believes the benefits of autonomous driving will be realised in 50 years’ time and under his IEA framework he claims that could be accelerated to permit them to be seen within the next 15 years from 2018 – by 2033 rather than by 2068.

Packaging intelligence

He adds: “Commercial autonomous vehicle initiatives are all heavily relying on packaging all the driving intelligence (situational awareness generation, decision making, along with drive-by-wire) within the vehicle. However, this introduces software functionality that cannot be deterministically predicted and analysed. Correspondingly, this places a huge liability on the automaker. It appears that automakers are responding to this by retaining the ownership of these vehicles, and by deploying autonomous vehicles through shared fleets.

“While there is indeed a positive response to ride sharing, as a society we are far away from relinquishing private car ownership in any significant way. What this implies is that we are very far away from any meaningful penetration of AVs. A critical way to promote usage penetration is to promote privately owned AVs but this requires a dramatic reduction in the liability associated with autonomous vehicles.”

Vehicle and infrastructure

“We propose that one way to reduce that liability is to redistribute the driving responsibility from being completely inside the vehicle, to include the vehicle and the infrastructure. Our concept leverages the rapid growth in connectivity (from Wi-Fi to DSRC to 5G and more) and the huge potential for edge computing to propose an architecture that will change the current business model of autonomy for privately owned autonomous vehicles.”

With IEA, he explains: “Autonomy is not just supported by the infrastructure but actually enabled by the infrastructure. In the IEA concept, autonomy is provided as a ‘service’ and, as such, people would pay for autonomy by usage – on an as-needed basis. In our concept, the driving intelligence is split between the infrastructure (which will provide situational awareness), third-party vendors embedded with the OEM (who will provide decision-making) and the traditional automotive OEMs (who will provide the core powertrain drive-by-wire capability).”

New business opportunities

“We believe that our techno-business concept will accelerate the deployment of autonomous vehicles because it improves the attractiveness of autonomous vehicles to the end users, reduces the potential risks for traditional automotive OEMs, introduces new business opportunities for infrastructure operators and creates an entrepreneurial eco-system driving decision making and user engagement.” He concludes by suggesting that his concept will “significantly increase the penetration (excess of 15%) within 10 years after the core technology reaches maturity over the next 3 to 5 years.”

[Auto.Jarvis.2018.01.12]

 

Connected & Autonomous Vehicles

14 May 2018 - 17 May 2018, Santa Clara, USA

From vehicle electrification and infrastructure to the evolution of ADAS and vehicle automation to enhanced connectivity and new mobility models, no rock will be left unturned.

Related Reads