Draft:Trillium Technologies

{{subst:submit}}

Trillium Technologies

Company type	Research and Development Company
Industry	Artificial Intelligence
Founded	October 2015
Founder	James Parr
Website	trillium.tech

Trillium Technologies (Trillium) is a research and development company with a focus on intelligent systems for public benefit, planetary stewardship, space exploration, and human health

History

Trillium was established by James Parr with the goal of applying artificial intelligence (AI) to systemic space and Earth science domain problems through a public-private partnership model with space agencies. Trillium was originally created as a UK-based sustainability consultancy called Imaginals Ltd, which developed grand challenges for the Obama administration in the U.S.A. and provided management support for NASA’s Asteroid Grand Challenge in 2015. Also in 2015, Imaginals was rebranded into Trillium Technologies with a focus on intelligent systems.

As an outcome of NASA’s Asteroid Grand Challenge, Trillium worked with NASA’s Office of the Chief Technologist to conceive and deliver the Frontier Development Lab (FDL U.S.) in 2015, an applied AI research lab focused on space applications. FDL U.S. was operated in partnership with the SETI Institute, NASA Ames, and NASA HQ.

In 2018, Trillium partnered with the European Space Agency (ESA) Phi-Lab to create the Frontier Development Lab Europe (FDL Europe) which has run annually since its inception in partnership with ESA ESRIN, ESA ESOC, ESA ECSAT, and ESA ESTEC.

Other partners of Trillium include Oxford University, the U.S. Department of Energy, and UN agencies such as UNICEF and UNOSAT as well as others via the Frontier Development Lab initiative.  

In 2020, Trillium built the online platform SpaceML to develop and distribute open-source AI research and data and invite space and Earth science researchers to more easily reproduce results and develop derivative investigations. During the same year, it also launched a research program in Australia called Data Quest in partnership with the Australian Space Agency and other partners.

In 2023, Trillium launched FDL MED to focus research efforts on AI for exploration medicine. In the same year it also launched Trillium Applied Services in order to formalize its AI, space, and healthcare services for the private sector.

Trillium has achieved numerous demonstrations of ML onboard spacecraft. These are organized under the sub-brand NIO: Networked Intelligence in Orbit.

Structure

Trillium Technologies is a multinational group of three separate entities headquartered in London, U.K. These entities are Trillium Technologies, Ltd. in the U.K.; Trillium Technologies, Inc. in the U.S.A., and Trillium Tech Pty., Ltd. in Australia. Trillium's workforce is international in nature and includes researchers, scientists, designers, developers, project managers, and AI specialists.

Products and Applications

Independent of the operations of the Frontier Development Lab, Trillium also produces AI technologies and peer-reviewed research in partnership with NASA, ESA, USGS, the U.K. Space Agency, and the U.S. Department of Energy.

These products and applications include:

• Kessler: Kessler is a Python package for simulation-based inference and machine learning for space collision avoidance and assessment.
• Karman: Benchmarking thermospheric density estimation in collaboration with Oxford AI4Science Lab.
• ML4Floods: ML4Floods is an end-to-end machine learning pipeline for flood extent estimation: from data preprocessing, model training, model deployment, to visualization. FloodMapper is built upon the open-source ML4Floods toolkit.
• NIO: Networked Intelligence in Orbit (NIO) is a family of research projects that leverage machine learning in space. Related peer-reviewed research includes “Towards global flood mapping onboard low cost satellites with machine learning” and “In-orbit demonstration of a re-trainable machine learning payload for processing optical imagery” in Nature Scientific Reports.
• RaVAEn: RaVAEn is a lightweight, unsupervised approach for change detection in satellite data based on Variational Auto-Encoders (VAEs) with the specific purpose of on-board deployment.