The development of connected autonomous vehicles is gathering pace. Modern cars, while yet to be considered fully autonomous, are being launched with ever more features which provide autonomous functionality.
Cybersecurity is one of the biggest fears surrounding this drive towards full autonomy. If connected autonomous vehicles are networked together on the road, even relatively low-level cybercrime could wreak havoc on society. Theoretically, all vehicles connected to the network could be simultaneously hacked and driven nefariously to cause dangerous problems.
Another obstacle is the trust that society has in the concept of a connected autonomous vehicle. Media coverage of accidents involving driverless cars hasn’t helped to build public confidence in vehicles being driven by data rather than a ‘trusty’ human being.
The role of mathematics in connected autonomous vehicles
Cybersecurity is improved by increasing data security, either by making data more difficult to access when it’s being stored or making it more difficult to intercept when it’s being transmitted. The technology which increases data security typically involves applications of cryptography. This often involves the use of abstract algebraic principles to encrypt data in such a way that makes it difficult to read what the data represents without knowledge of the encryption used.
The use of data to drive a vehicle often depends on machine learning algorithms, which apply statistical models and probabilistic techniques to data gathered by a car in real-time. This enables the computers driving the vehicle to determine what to do next.
It’s clear that the application of mathematics has a major role to play in the development of connected autonomous vehicles. However, obtaining patents to protect technology underpinned by mathematical techniques can be tricky — here’s why.
Mathematics must improve a technical process
UK and European patent law excludes mathematical methods as such from patentability. Additionally, US patent law doesn’t view abstract subject-matter with any great deal of positivity. So does this mean that those innovators developing technology to make connected cars more secure or ‘intelligent’ on the road should avoid filing patents?
It really does depend on the invention. If your ‘invention’ results in better mathematics or a solution to a long-standing mathematical problem (say, the Riemann hypothesis) then it’s likely to be very difficult to get a patent. However, if the better mathematics has an impact on a physical process then a patent becomes more likely, as the mathematics improves a technical process.
If applied to cryptography, guidance from patent offices indicates that a new algorithm which makes a code more difficult to crack is more likely to get a patent than a new algorithm which enables the factorisation of a large prime number without any other technical benefit (as beautifully elegant as the mathematics may seem!).
If applied to machine learning, a new statistical algorithm which uses LIDAR data to determine the likelihood of black ice on the road ahead is more likely to get a patent than a change of statistical distribution in an existing algorithm which, while making the mathematics more interesting, provides no further technical benefit than the existing algorithm.
Last year, the European Patent Office (EPO) clarified its approach to artificial intelligence and machine learning. It stated that the application of these principles to improvements in a technical process would not be excluded from patentability as a mathematical method. In the example given above, it would be difficult to discount the detection of black ice as a non-technical matter, but changing the underlying statistics of the algorithm with no additional benefit to the outcome would likely be viewed as non-technical.
This guidance should encourage the protection of innovation in this area or, perhaps more usefully, provide guidance as to how decision-makers should conclude whether to file a European patent application.
This guidance also shows that the EPO is preparing itself to examine patent applications in this area — and that it’s already drawing lines in the sand to give stakeholders an idea as to how their technologies will progress before the EPO.
Not sure whether your innovation is patentable?
The application of mathematics can have many people running for the hills — but it will be key to the development of connected autonomous vehicles. If you’re innovating in this area, you shouldn’t fear filing patent applications directed to technology which is underpinned by mathematical principles.
If you’re not sure whether your innovation is patentable, it’s best to get a preliminary opinion on whether to pursue patent protection from an intellectual property attorney. This will provide you with informed guidance, offering your best chance of cost-effective protection without pursuing any dead-ends.