The Combat Air AI Challenge received a huge response from across the breadth and depth of academia, micro-SMEs, and medium-scale start-ups. The challenge comprised of two streams:
-
Stream 1 - to utilise deep reinforcement learning to inform sensor resource management across the ISANKE super-system by modelling behaviour of the blue force platform whilst exposed to a range of operational conditions.
-
Stream 2 - to explore potential deployment of deep integrated sensor fusion across the ISANKE super-system using machine learning.
Across these two streams, we identified two nominees for the pilot-projects: Decision Lab for Stream 1 and Origami Labs for Stream 2. We have been working closely with both the nominees to develop innovative insights at pace for each stream since the pilot-projects kicked-off in mid-July.
As the pilot-projects near completion, Harita Joshi, Head of Capability, Future Combat Air at Leonardo UK, who conceived and led this initiative, reflected on the importance of technology innovation and partnerships for large-scale product development endeavours: “The GCAP programme provides a once in a generation opportunity to explore and expand the technology boundaries of the art-of-the-possible. We wanted to test out a model whereby we can leverage expertise from wider UK industry and academia to join us on that journey. The success of the challenge is based upon the principles of agility, fast-paced innovation and collaboration, which have enabled the delivery of substantive results within a year”.
Elaborating on the partnerships, Amber Illman, Procurement Officer, Future Combat Air, added: “The collective UK SMEs/Academia community represents the cornerstone of a future resilient supply chain for us. We explored non-traditional procurement routes to rapidly expand our reach and engage with the best UK has to offer, allowing us access to cutting edge research, technology advancements, and skilled talent”.
The Challenge, a joint endeavour with Ministry of Defence (MOD), supports Leonardo UK’s work within GCAP (Global Combat Air Programme). Leonardo in Italy and the UK is playing a major role in GCAP, including Leonardo acting as the UK and Italian leads for the core platform’s Integrated Sensing And Non-Kinetic Effects (ISANKE) and Integrated Communication Systems (ICS). ISANKE is a complex network of capabilities and, whilst co-ordinating numerous sensors, assimilates information into a common situational awareness picture as part of an integrated offering.
As part of Stream 1, Decision Labs have been working on developing a Reinforcement Learning system as a game-theoretic, non-myopic decision making solution within a simplified air combat simulation. The simulation pitches blue platforms against a range of operational scenarios and tests continuous and discrete actions in a complex action space through a curriculum learning approach. This could support automated or assisted decision making which drastically reduce a pilot’s cognitive load under challenging combat air operational scenarios. Decision Lab CEO, David Buxton, highlighted: “We’re thrilled to be working with Leonardo. This challenge has given us the chance to build our working relationship and advance cutting-edge technology to meet new and emerging challenges. As the project continues, we look forward to developing and refining new capabilities.”
Stream 2 sees Origami Labs explore the challenging problem of multi-platform multi-sensor fusion using machine learning to develop a solution that allows rapid implementation, evaluation and visualisation of different data fusion architectures to enable comprehensive trade space analysis. The solution considers the context of pertinent domain-specific constraints such as small, low-resolution datasets for operational relevance. This would provide valuable insights into how the deployment of machine learning approaches can enhance a pilot’s situational awareness. Thom Kirwan-Evans, co-founder and director of Origami Labs, commented: “Leonardo have been very flexible, given our size and age as a company. The project is being developed collaboratively in an agile manner, evolving to account for learnings as research should. This has been a great opportunity; as a new company formed to make a real-world impact, we could not have asked for much bigger. The challenge is helping us to grow, expanding a UK SME capability.”
Commenting on the overall success of the endeavour, Jonathan Smith, VP Capability and CTO, Future Combat Air (GCAP), Leonardo UK, said: “It has been great to see this initiative unfold into effective partnerships that deliver technical insights into a complex problem space.”
Leonardo will continue to develop the technical innovation and partnership themes that have emerged out of the Combat Air AI Challenge.