Hadean secures £20M synthetic environment & AI deal with MoD

By Sam Cranny-Evans, published on 5th December 2024. 

Hadean, a UK defence tech company, has secured a UK MoD Enterprise Agreement under Defence Digital worth up to £20 million (€24.16 million), according to a 4th December press release. This partnership will allow Hadean to develop its AI and synthetic capabilities to support the MoD in areas such as training, simulation, and mission planning and will give the MoD access to Hadean’s synthetic environments and incorporate them into its existing capabilities. 

On the same day, Hadean announced three other agreements across industry, including a partnership agreement with BAE Systems under Project OdySSEy, which provides users with a synthetic environment that simulates the real world using Hadean’s AI capabilities such as its machine learning to simulate and emulate patterns of life. The goal is to allow users to experience more natural behaviour from simulated people, rather than following set paths that have been hard coded into them. Project OdySSEy is run by BAE combining a range of industry partners to provide virtual training environments for armed forces. It is designed for training scenarios that are difficult to carry out in a live environment for reasons of risk, cost, or difficulty. OdySSEy also uses BISim’s VBS4 simulation software, which you can see in the video below. 

In addition to the BAE partnership, Hadean announced two others with Deloitte and Moody. With Deloitte, Hadean will support the NATO Modelling & Simulation Group’s capability development initiatives, providing its technology to empower a common synthetic environment for NATO partners. And, in its work with Moody, Hadean will be combining its synthetic environment with Moody’s data to understand logistics and supply chains as a means to understanding risk and the potential for disruption arising from natural disasters and man-made events. 

In October this year, Hadean demonstrated an AI-enabled training capability that utilises large language models (LLM) to populate a synthetic training environment with ‘human terrain’ that mimics human interactions, in a trial conducted for the British Army. They used multimodal LLMs to create a more immersive and realistic training environment by integrating realistic simulations of audiences, actors, adversaries, and enemies (A3E). The capability was designed to simulate the complex interactions between soldiers and civilians and enhance the realism of live and simulated training.

Multimodal LLMs are a type of AI that can process and understand information from various sources, such as text, audio, and images. Unlike traditional LLMs, which primarily work with text, multimodal LLMs can create more comprehensive and realistic content, making them ideal for applications like synthetic environments.

This video shows the BAE Project OdySSEy, which runs on software developed by BISim. Hadean’s synthetic environment technology will be integrated into this system to generate more realistic actions from the non-playable elements. Credit: BISim. 

Calibre comment – synthetic environments

What is a synthetic environment? And why are they so useful? A synthetic environment is a computer-generated world that simulates real-world scenarios. It’s like a virtual reality experience, but often used for more practical purposes. These environments can be as simple as a virtual room or as complex as a simulated city or battlefield.

 A typical use case is for battle simulation and training as they can allow a large force to train together without the safety restraints that they would face in the real world. ‘It’s easier to practice realistic tactics in a simulation than on the training ground,’ a serving soldier once told Calibre. The reason for this is that safety requirements can often prevent a unit from training in the way they would fight, they may have to seek permission from a safety officer before firing, for instance. 

Synthetic environments also allow AI algorithms to be trained on simulated data and then tested in a synthetic environment to assess how they perform. This is pretty important for defence applications as armed forces have to figure out how their algorithms will perform, especially if they are used in decision-making or target selection. Algorithms must always be tested in the real world, of course, but synthetic environments can add a lot of value by providing training scale at a much lower cost.