ARIA: Standards and benchmarks for dexterous robotics

ARIA funds and supports research teams, known as Creators, to achieve breakthroughs. They could be anyone, from individuals doing their own research and PhD students at a university, to startups and large organisations.

Our programmes deliberately reach across disciplines, sectors and institutions – we welcome applications from across the R&D ecosystem, including individuals, universities, research institutions, small, medium and large companies, charities and public sector research organisations.

Applications will be stronger where they bring together a broad range of organisations, including overseas partners and influential developers, manufacturers, research groups, end users, test facilities or standards organisations.

For the vast majority of applicants, we require the majority of the project work to be conducted in the UK (i.e. >50% of project costs and personnel time).

We can award funding to applicants whose projects will primarily take place outside of the UK if we believe it can boost the net impact of a programme.

Projects will run for up to 24 months from the agreed contract start date. Teams may request up to £3 million in ARIA funding per project for this period. Proposals addressing a focused subset of the RfP are also in scope, with the funding requested proportional to the proposed work. However, preference will be given to consortia with a credible plan to encourage wider uptake across the global robotics community, rather than small groups working in isolation. We expect to make up to four awards.

Matched funding is not required but is permissible. Applicants may include cash or in-kind contributions and should identify their source, status and any conditions. The absence of matched funding will not in itself disadvantage an application.

We seek proposals addressing a clearly defined barrier that affects multiple developers or users of dexterous robotic systems. Proposals may focus on one area or combine several. Applicants should justify a coherent scope that can deliver useful, validated outputs within two years; they are not expected to address every area below.

Examples of work in scope include:

• Tactile data standards: common data formats, representations, interfaces, calibration methods or communication protocols that enable tactile data from different sensing technologies and geometries to be understood and used across systems. Approaches may use physically defined quantities, learned representations or other methods, with clear treatment of uncertainty and sensor limitations.
• Benchmarking and performance evaluation: frameworks that define the dimensions of dexterity and connect component performance, manipulation capabilities and application-level outcomes; repeatable test methods, metrics and reporting protocols; and ways to assess reliability, robustness, adaptability and readiness for deployment.
• Shared testing and adoption infrastructure: reproducible test rigs, datasets, tools, coordinated trials across laboratories, accessible test facilities or benchmarking services that allow others to evaluate and improve their technologies.
• Other enabling work that addresses a significant shared barrier to dexterous robotics and manipulation, with a clear explanation of how it will benefit the wider field.

Applicants may build on previous work, including work funded by ARIA, but prior involvement in ARIA-funded work is not required. All proposals should explain what they will reuse, improve or connect, and why further work is needed.

Successful projects will be expected to:

• Submit a clear project plan for what they plan to achieve, with specific milestones.
• Define the problem with the developers and users affected by it, and show how the proposed outputs will improve their ability to develop, compare, integrate or deploy robotic systems.
• Deliver practical, reusable outputs with measurable success criteria. Frameworks and proposed standards should be exercised through test methods, implementations or trials appropriate to their scope, with sufficient documentation for others to use them.
• Validate their work across independently developed technologies and with different organisations and test sites. Evaluation should establish reproducibility, limitations and relevance to real manipulation tasks. For tactile data, this could include the effort and performance involved in adding a new sensor; for benchmarking, consistency across sites and the relationship between measured capabilities and useful task performance.
• Engage relevant ARIA-funded creators and the wider robotics community, while making clear which access arrangements and partner contributions are already agreed and which remain to be secured.
• Demonstrate a route to international adoption during or after the project timeline, including participation by non-UK and/or international organisations able to implement, test, maintain or promote the outputs. Where appropriate, engage established standards bodies and related initiatives, and explain how the work will complement them.
• Explain who will own IP generated through the work and on what terms others can use it. We favour arrangements that support broad international adoption while capturing lasting value for the UK, which may include licensing revenues for UK organisations.