Bio-mechatronics and robotics

Exploration robotics

Wormbot

We are developing robotic systems with the capability of exploring natural and artificial environments that cannot be safely or easily accessed by humans.

There are many scenarios where exploration robotics have the potential to have significant impact:

  • Improved manufacture of complex engineering systems (aircraft, nuclear power stations, ships) through inspection and monitoring of inaccessible locations
  • Improved surgical outcomes through developing technologies to monitor and measure patients internally during surgery
  • To discover new knowledge of archaeological sites, whilst preventing damage
  • Saving lives through quick identification of victims trapped within buildings following a natural disaster.

Some locations are difficult to access due to their physical location or size, whilst others maybe hazardous to human health due to heat, air pressure, toxic chemicals, nuclear radiation or human action such as in war zones.

Exploration robots are often small requiring miniature mechanisms, actuators, sensors and electronics. They need the capability of operating at a distance from direct human intervention for long periods of time in the presence of incomplete sensor information and delays in communication.

Our industrial collaborators include: Dassault Systemes, BAE systems, MBDA and Soutek UK.

Recent projects include:

Djedi robotic archaeological expedition

We have developed state of the art robotic systems to explore the Great Pyramid of Giza, Egypt. Our robots have the capability of climbing 70m within a confined space of 20cm by 20cm and deploying snake cameras and drills. On our most recent mission we successfully navigated the southern airshafts of the pyramid and deployed a snake camera and discovered writing that has been hidden for thousands of years.

UK MoD grand challenge

We participated in the UK’s grand challenge to develop autonomous vehicles to survey an area for improvised explosive devices and snipers. We developed (in collaboration with the University of Manchester, BAE systems and MBNA) autonomous air and ground vehicles. The novel patented design of the air vehicle was a configuration from 6 propellers in a Hex-rotor configuration. We developed a ground vehicle based around a six motor drive system to enable independent body orientation and locomotion.

Perch and stare for unmanned air vehicles

We have developed control algorithms to enable unmanned air vehicles to perform perch and stare manoeuvres. The perch and stare approach involves a UAV performing a point landing on a building or wall to observe for an extended period of time and then re-launch.

Biomimetically inspired search and rescue robotics

Despite significant advances in technology and science, natural disasters remain a very real threat to human populations around the world. The challenging task of finding survivors among the rubble of collapsed or damaged buildings is one where mobile robots could make an extremely valuable contribution. The problem, of course, is that this sort of environment is extremely challenging from a locomotion point of view. Traditional wheeled or tracked robots have little chance of successfully navigating through the complex and irregular spaces between debris so alternative strategies must be identified. We look to the natural world for inspiration.

An EPSRC funded project developed a robotic system, biologically inspired by the european mole, to provide search and rescue robots with unique capabilities. The developed system has the capability to burrow through loose debris to gain access deep within damaged buildings to search for survivors. The other search and rescue robot under development is a serpentine (snake-like) robot, although the animal it draws its inspiration from is actually a worm called C. elegans (that also crawls with sinusoidal undulations). Although significant progress has been made in the development of robots with serpentine properties, one of the main outstanding challenges relates to the adaptation of the locomotion waveform to external constraints (obstacles).

Our WormBot uses a novel, computationally efficient distributed control system based on the C. elegans neural network to generate robust undulations. More importantly, the robot exhibits a remarkable ability to adapt its waveform to accommodate external constraints. What is particularly interesting is that this adaptation occurs without any explicit mapping or planning. Instead, obstacles in the environment are overcome based purely on proprioceptive feedback representing the current body shape. The robot does not even need to know the obstacles are there.

For more information contact: Professor Rob Richardson or Dr Jordan Boyle.