Background
Autonomous robotic systems are physical platforms with motors and sensors enabling them to move through and perceive the environment around them. On-board and cloud computing processes make use of sensor information to apply autonomous control algorithms to navigate, make decisions and undertake actions in the real world.
It is of critical importance that the actions of autonomous robotic systems are predictable, remain under control and any rogue systems (or subsystems) detected and effectively managed immediately.
A detailed risk assessment is required to understand and manage the risks of autonomous robots and thereby provide trust in safe and reliable operation.
Currently, this risk assessment may involve training the product within a computer-simulated environment. However, there is always a significant gulf between the simulated and true environment. For example, unexpected events may occur in the real world, such as human or animal activity disturbing a system.
In addition, real-world autonomous robotic systems are vulnerable to cybersecurity threats, which further complicate the creation of trustworthy autonomous robots by requiring consideration of the deliberate and malicious compromise of systems.
Sandpit
EPSRC, in collaboration with National Security Technology and Innovation Exchange (NSTIx) and UK government partners, are inviting applications to attend a joint sandpit in the area of trustworthiness of autonomous robotic systems for national security and defence.
The aim of this sandpit is to generate new, innovative and transformative approaches to the creation and operation of trustworthy autonomous robotic systems with specific focus to addressing the needs and challenges within defence and national security.
The sandpit will be an intensive, interactive, inclusive and free-thinking environment, where a diverse group of participants from a range of disciplines and backgrounds will work together for five days. The aim will be to immerse participants in collaborative thinking processes and ideas sharing in order to construct innovative approaches.
It will be led by a director, who will be supported by a team of mentors. Professor Robert Richardson (University of Leeds) will be the director for this sandpit.
The director, mentors and a small number of stakeholders will attend the sandpit but will not be eligible to receive research funding. Instead, their role will be to assist participants in defining and exploring challenges in this area.
The process can be broken down into several stages:
- defining the scope of the challenges
- evolving common languages and terminologies amongst people from a diverse range of backgrounds and disciplines
- sharing understandings of the challenges, and the expertise brought by the participants to the sandpit, and perspectives from relevant stakeholders
- taking part in sessions focused on the challenges, using creative thinking techniques
- capturing the outputs in the form of highly innovative research projects
- a funding decision on those projects at the sandpit using ‘real-time’ peer review.
Participants should be able to apply their knowledge, skills and experience across disciplines to develop innovative research arising from different perspectives, with the potential to deliver new ideas focused on trustworthiness of autonomous robotic systems.
As the sandpit progresses, participants will build up thoughts on how the identified ‘challenges’ may be addressed and develop their innovative ideas and activities into research projects.
Projects should contain genuinely novel and speculative research. The director and mentors will act as independent reviewers, making a funding recommendation on the projects emerging from the process through a ‘real-time’ peer review process.
Scope
The sandpit will include inputs from a variety of sources, bringing together a wide variety of disciplines and innovators to address the research challenges associated with understanding and developing trustworthiness of autonomous robotic systems.
The sandpit will address key concerns, such as those given below, in relation to specific operational contexts:
- how can we minimise the gap between how a robotic or autonomous system (especially its artificial intelligence systems) behaves in a simulated environment and in a real-world operational context
- how can we ensure robust cybersecurity against malicious attempts to hijack, neutralise or otherwise tamper with a robotic or autonomous system’s functions
- how can we quantify the risk or confidence deficit that remains after minimising (but never eradicating) the simulation-reality ‘gap’ and the cybersecurity issues?
Through the sandpit, participants will build up thoughts on how these challenges may be addressed and develop their innovative ideas and activities into research projects.
The aim of this sandpit is to generate research proposals that:
- take into account the needs of customers and stakeholders and consider co-design with end-users
- form new collaborations between researchers, innovators and users in diverse research areas
- create new and transformative research ideas in trustworthiness of autonomous robotic systems
- allow researchers to pitch projects for seed funding to test and de-risk novel ideas
- address the key research challenges that are identified
- cultivate a common language between disciplines.
Achieving the sandpit aims will require participants from an appropriate mix of diverse backgrounds and relevant disciplines. Researchers from a diverse range of disciplines are therefore encouraged to apply to attend this sandpit.
We are not defining the disciplines that should be represented but asking potential participants to indicate how their expertise can address the challenge of trustworthiness in autonomous robotic systems in the broadest sense.
Applicants need not have worked on the problem before. However, emphasis will be placed on working across disciplines to foster new collaborations and bring new thinking to the problem.
Funding available
It is expected that up to £3 million of funding will be made available to fund research projects arising from this sandpit.
We aim to adhere to the key dates as published, however we reserve the right to make modifications in the event of exceptional unforeseen circumstances.
