MOTION – Morphological Computation of Perception and Action

Partners: Imperial College London (lead organization), Cambridge University, University of Surrey, Shadow Robotics Company, Rethink Robotics, Southern Scientific Ltd.

Duration: 3 years (July 2016 – July 2019)

Funding Body: Engineering and Physical Sciences Research Council (EPSRC), UK

Amount: £1.2 million

tumblr_inline_nxn3vzX9VD1r53j17_400Motivation: The recent Ebola crisis highlighted the need to have better accessibility to patients to address epidemics that can even pose a threat to the lives of healthcare providers. It made us think about a future possibility of having a safe robotic proxy through which a doctor located at a remote location could feel the physical conditions of a patient. We identified few main technological breakthroughs that would make this possible. First, we have to make robots safe enough to come into contact with a patient, then we have to find a good way to present the patient’s condition to a doctor.

Scientific questions: The overarching objective of this project is to improve the efficacy of soft robotic probes for soft tissue examination by deriving deeper insights into the role of mechanical impedance of a soft robot in regulating the morphological computational basis of coupled haptic perception and action. The notion – morphological computation – in soft robotics views the mechanical circuits in the embodiment as a computational resource for both perception and action. Since the mechanical impedance of the embodiment changes the functionality of those mechanical circuits, it is important to understand its role in regulating perception and action of soft robots. The intended application is to use a soft robotic probe to remotely examine the abdomen of a patient like in Ebola response and elderly care in their own households.

The nature of sensorimotor coupling and its implications on the very nature of computation of action-perception arbitration in soft robots and biological motor control is not yet well understood. From mechanoreceptors in the fingertips to the tendons in muscle-bone interface, human tactile perception and proprioception (constituents of haptic perception) are conditioned by the state of the surrounding soft tissue. Since soft-tissue properties and muscle tension change with action, haptic perception and action are intrinsically coupled. This enables the neural control to: (i) influence perception by conditioning muscle fibres around haptic sensor receptors, and at the same time (ii) directly solve local computational problems to do with motor control through the mechanical circuits of the embodiment itself. A deeper understanding of this coupled process of sensing and action will open up new opportunities to design efficient and robust soft robots for a variety of applications including remote physical examination of patients.

Therefore, the main objectives of this proposal are: 1) To systematically describe the information criteria that can be used by a soft robot to optimise the quality of haptic perception and the stability of physical examination of a soft object with variable stiffness; 2) To understand how human regulate joint stiffness through muscle co-contraction in a task that involves action to perceive a set of hidden information in a soft tissue; 3) To test a soft robot with a shared embodiment for haptic perception and action in a collaborative teleoperated physical examination scenario of palpating a phantom abdomen to identify an abnormality of an internal organ. Here the “shared embodiment” refers to soft tissue used to actuate the joints as well as to mount the sensors analogous to the way biological muscles are used to actuate as well as to accommodate spindle and tendon sensors.

For inquiries, collaborations, contact the principle investigator Thrishantha Nanayakkara