Philosophical penguins show the road to safety for smart cars
Eureka moments in academia are a rare thing. And when they come they often occur in the strangest of places. Take Archimedes for instance. The Sicilian-born physicist and mathematician discovered the theory of hydrostatics in his bathtub, while Sir Issac Newton’s life-changing epiphany, which led him to the theory of gravity, took place in a Lincolnshire garden and not a laboratory or classroom.
Some 300 years later, Professor Yiannis Papadopoulos, a leading computer scientist at the University of Hull, describes how he and his colleague, Dr Youcef Gheraibia, a lecturer at the University of Souk Ahras in Algeria, had their ‘A-ha moment’.
Recounts Professor Papadopoulos, “It was a Sunday evening in Algiers and after thumbing through the TV listings, Dr Gheraibia settled on a nature documentary. While everyone else in the room was fascinated by the leading-edge camera work, which captured a group of penguins searching for food in the open ocean off Antarctica, Dr Gheraibia saw something else. He noticed that penguins who worked in large groups were far more successful at detecting and capturing prey than solitary hunters, or when only a few birds worked in concert to find prey.”
(PICTURE) Dr Gheraibia in his living room in Shouk-Ahras, Algeria, where he had his penguin epiphany
Suddenly, Dr Gheraibia had a flash of inspiration. What if he could bottle the penguin’s collaborative approach to hunting for food and apply it to computer science?
But watching an episode of David Attenborough’s award-winning documentary series ‘Life in the Freezer’ only scratched the surface. Inspired by nature, Dr Gheraibia embarked on primary research to see if the natural world could solve practical engineering problems.
It was then that Professor Papadopoulos, who is Leader of Hull University’s Dependable System’s Research Group, had his lightbulb moment. Professor Papadopoulos, by chance, stumbled across Dr Gheraibia’s work. For years, his group had been developing a state-of-the-art model based safety analysis and design optimisation tool called HiP-HOPS, which uses algorithms to identify and correct faults in complex engineering designs. Papadopoulos saw Gheraibia’s study as the missing piece of the jigsaw.
“Bio-inspired technologies could one day revolutionise software design and could do that for the smart vehicles of tomorrow”, he says.
Papadopoulos explains, “As we enter a fourth industrial revolution, technology is moving at an incredibly fast pace. By 2035, it is likely that many of the largest cities in Europe, North America and Asia will have embraced Connected Vehicle landscapes and Level 4-5 autonomous driving technology.”
Continues Papadopoulos, “Going forward - the big challenge for the industry is to ensure that the software that runs on selfdriving cars, the code that allows vehicles to communicate with each other, pedestrians, cyclists and roadside infrastructure, does what it is supposed to do. Automakers, legislators and, society as a whole, are demanding that these systems deliver high integrity. They want to know that the software in a connected, autonomous vehicle works correctly, uninterrupted by failure and, in an age of cyber-terrorism, is protected from hackers. Importantly, they expect that systems in a connected environment continue to interact safely in infinity of possible scenarios that cannot all be scripted and assessed a priori”.
(PICTURE) Professor Papadopoulos holding a ceramic statue of a penguin handcrafted by his son Oisin
More code than you can count…
The obstacles faced by researchers like Professor Papadopoulos and his team are numerous as they were complex. Take a new car today for example. While it will probably have some form of in-car computer system on-board, the level of interoperability is fairly basic compared to a vehicle sold in two decades’ time.
However, according to Professor Papadopoulos, “in a modern luxury vehicle, it is claimed that 100 million lines of code exist, and in the future, with the advent of Level 5 autonomous vehicles the amount of code is likely to increase three-fold (see IEEE Spectrum article)”.
Explains Papadopoulos, “With so much software already on board and moving towards autonomous and connected cars, we face enormous challenges. The increasing complexity and density of software increases the potential for functional defects. Frequent software design changes increase the probability of design errors, and the increasing complexity poses serious questions: how do we succinctly, powerfully and quickly navigate through enormous bodies of software code? How do we structure and deliver this code? And, how do verify that this code is dependable?”
“In addition, autonomous and connected cars introduce completely new hurdles. In connected systems, the configurations that the total system of vehicles and infrastructures may assume are unknown and potentially infinite. As vehicles and infrastructures connect, emergent system behaviours may arise in ways that are difficult to predict from simple superposition of the behaviour of individual elements. How does one design for safety, and ensure safety, in such systems?”
How a hungry penguin inspired change…
But the question many of you might still be wrestling with is, how did the hunting behaviour of a small, black and white torpedo-shaped flightless bird, located in Antarctica some 1,238 kilometres from the nearest car (Ushuaia, Argentina – in case you’re interested), provide a solution to one of the most complex conundrums facing carmakers today?
Says Papadopoulos, “Observing and learning from the natural world has unlocked many doors for scientists and engineers in the past. Take penguins for example. They live in large densely-populated rookeries which sometimes number a million birds.
“If the whole colony is to survive – it must consume a vast amount of food. Each penguin, for instance, needs to eat one kilogram of fish each day to live. So a colony of a million animals would require 1,000 tonnes of fish each day to sustain itself.”
“Therefore, the hunting strategy that these penguins developed through evolution, must be highly efficient – otherwise the colonies would be a lot smaller or, in the worst case scenario, the birds extinct.”
"We realised that this strategy has generic elements which can be abstracted and be used to solve other problems," he said. "Such as determining the integrity of software components needed to reach the high safety requirements of a modern car."
Penguins dance to the HiP-HOPS beat
Researchers at Hull, in conjunction with Dr Youcef Gheraibia have developed tools based around the collaborative foraging behaviour of penguins that can help optimise complex software to make sure it meets safety requirements. These algorithms have been integrated into the HiP-HOPS tool.
"We can use penguin algorithms to solve difficult design problems" Papadopoulos said. “Early applications include determining the optimal software integrity of subsystems and components in cars as well as the architecture and replication of software tasks on controllers.”
(PICTURE) HiP-HOPS: The HiP-HOPS interface within Safety Designer (a tool by ESI-ITI GmbH)
Papadopoulos added that groups of simulated penguins can quickly search through the innumerable ways software can be specified and ordered to find optimal solutions in terms of safety and cost. This work, which has already been demonstrated in several case studies, seeks to improve and automate the application of ISO 26262 - one of the most important and exacting safety standards in the automotive industry today.
And as cars reach a greater level of automation – as they begin to communicate with other vehicles and the surrounding infrastructure, “this goal-based international requirement, which sets safety standards for a vehicle’s electrical and electric systems, will take on an even greater significance”, thinks Papadopoulos.
He explains, “The more advanced that cars become, the greater the technology requirement and pressure on automakers to ensure that safety requirements are captured early in the design phase,” he begins. “The sub-systems, i.e. the actuators, sensors and controllers, which make up the highly complex vehicle architectures of tomorrow will also need to be integrity checked at an early stage of development. Currently, this is mostly done manually using engineering judgement and at best semi-automated techniques. But, in the very near future, when each vehicle is likely to contains three hundred million lines of code connected to code of other vehicles, that approach is neither cost effective nor safe. If the carmakers cannot guarantee safety, they no longer have a sustainable business model. Hence the penguin system.”
A digital certificate with a difference…
Professor Papadopoulos, who estimates the combined cost of the two projects to be around three million pounds, has also secured EU funding to develop vehicle integrity certification which he says “would aim to enhance safety and other attributes of dependability for self-driving vehicles in connected environments”.
So what is the science and reasoning behind these digital dependability certificates?
Says Papadopoulos, “Digital certificates can be crucially important in connected and autonomous vehicle landscapes, as cars will constantly need to make split-second decisions. The big question that the industry does not seem to have a comprehensive answer to, is how are these decision are being made?”
“Our aim therefore, in this new EU project, is to create a concept of digital certificates that will facilitate safe operation in open systems. You can imagine these certificates as modular, composable, and executable specifications of dependability for components and systems that will span the lifecycle of those systems. They are produced semi-automatically during safety assessments, with system certificates utilising component certificates, and provide a basis for certification of a system before it can participate in a connected environment. We believe that such certificates must also be executable. That they can function in real-time is an absolute pre-requisite, as they will need to be able to monitor and evaluate every possible scenario aiming to continually certify a system for operation within its current context.”
“Finally, the certification would cover every vehicle component and so it would be valid and enforceable not just for OEMs but for the tier 1 and tier 2 companies too. This credential is incredibly important in a connected vehicle landscape,” he adds.
And how long does Papadopoulos think it will take before the project is fully executable?
Continues Papadopoulos, “The initiative is at concept stage at present so it could be many years before we are ready to share the certificate with EU regulators and wider industry.
“We are facing a number of challenges including merging traditional dependability models, data analytics and penguin algorithms for search, in order to create new algorithms that can operationalise these digital certificates. The next phase will be to seek out and begin building the technology infrastructure that can help us realise this exciting goal in the context of the DEIS Horizon 2020 project.
“DEIS stands for Dependability Engineering Innovation for cyber-physical Systems, but the Latin translation is ‘to the gods’. We may need some help here,” chuckles Papadopoulos.
Looking back to look forwards…
With our interview drawing to a close, I tell Professor Papadopoulos, who hails from the city of Thessaloniki in Greece that I intend to begin the feature by referencing, Archimedes, one of Ancient Greece’s most famous scientists.
“Why not end the piece by quoting one of our most renowned philosophers then,” he jokes.
“How about this one from Aristotle?” laughs Papadopoulos.
“He once said, ‘Everything you can imagine already exists in the sphere of the potential’ ”.
“In other words – for every challenge we encounter in life - a solution already exists.
“A beautiful Aphrodite pre-existed in the block of marble from which she was carved” says Professor Papadopoulos, and “like sculptors or a hungry colony of penguins,” he continues “we scientists are privileged with being able to occasionally catch some fish or carve an answer out of a difficult problem.”
Over to you then, Professor Papadopoulos and Dr Gheraibia …