Robot Dexterity

Clarification questions

Answering your questions

If you have questions related to Robot Dexterity, please reach out to clarifications@aria.org.uk.

We’ll update this page twice a week with answers.

Nb: clarification questions should be submitted no later than 4 days prior to the relevant deadline date. Clarification questions received after this date will not be reviewed. 

What counts as a robot?

For the purposes of this programme and opportunity space, “robot” means any engineered device or machine with complex computational, perceptual and sensory capabilities, plus complex capabilities for autonomously acting on the physical world. We do not place any requirements on the materials or technologies used for a robot.

 

Are molecules in scope?

When determining if an application is in or outside scope, the Programme Director will decide whether capabilities count as “complex”. As a concrete negative example, consider an artificial molecule that performs similarly to an NMDA glutamate receptor: it opens to permit ion flow when both a ligand is bound and potential difference exceeds a threshold. The molecule thus arguably carries out a physical action after performing a logical computation on sensory inputs. However, we do not consider either the computation or the action complex enough to qualify this molecule as a robot in itself. A molecular motor could still be within scope under the “Novel components” aspect of the programme, since it could potentially be incorporated into a “sufficiently complex” robot as one of several components. 

 

Are non-autonomous devices in scope?

For us, autonomy is a defining feature of a robot (Figure 1). Pre-programmed or automated machines, or those tele-operated by humans, are not true robots. However, since this programme is focusing on novel hardware, we will consider non-autonomous or semi-autonomous devices if there is a sufficiently high degree of novelty in the hardware or the dexterous abilities, especially if the intent is to move towards increasingly autonomous operation. This applies for example to surgical robots. The programme director will decide what counts as “sufficient”, and may define particular applications as out of scope at Stage 2. 

 

Are exoskeletons in scope?

Wearable devices such as exoskeletons, which are attached to humans and never intended to operate autonomously, are not in scope as use-cases. However, we would be open to, for example, funding work by an exoskeleton company on a novel actuator which could benefit both robot dexterity and exoskeletons. In such cases, please explain your own motivations and how these align with the programme goals.

 

Are protheses in scope?

Devices such as prosthetic hands are an edge-case. Clearly there is overlap between the need for strength and dexterity in both prosthetic hands and robot end-effectors. However, prostheses have additional constraints which do not transfer directly to autonomous robots, for example acceptability to, and the ability to be easily controlled by, their human users. Any applications involving prosthetics should therefore be careful to explain how the proposed work advances the programme objectives.

 

What do you mean by dexterity? Is this different from manipulation?

We are not making any meaningful distinction between dexterity and manipulation. In our usage, both refer to skilful, precise manipulation of specific objects, usually with a high degree of flexibility. This will usually be through physical contact, but we do not require this. Flexibility is key. Humans display dexterity when they knit or sew manually, but a knitting or sewing machine does not display dexterity in the sense intended by the programme. Rather, these are machines designed to remove the need for dexterity on one specific task. 

The Programme Director will determine the level of skill, precision and flexibility required to count as “dexterity”. This will depend on the situation and task, but will require a clear advance over the state of the art along some axis (not necessarily performance alone, but also reliability, cost, sustainability and anything else where a case can be made for societal benefit). 

 

You exclude defence/military. Does this include things like private security, police, border patrol, security services? 

Yes, work that is specifically for such use-cases is out of scope.  

 

What about upstream manufacturing and supply chain logistics for defence and military? What about military healthcare?

We recognise that a dexterous robot that could help with civilian manufacturing, logistics and healthcare could necessarily also assist with these tasks in defence/military contexts, but we will not fund any tests or field trials in these contexts.