Robot sheepdog is ducks' best friend

Industrial Robot

ISSN: 0143-991x

Article publication date: 1 December 1998

76

Keywords

Citation

(1998), "Robot sheepdog is ducks' best friend", Industrial Robot, Vol. 25 No. 6. https://doi.org/10.1108/ir.1998.04925fab.010

Publisher

:

Emerald Group Publishing Limited

Copyright © 1998, MCB UP Limited


Robot sheepdog is ducks' best friend

Robot sheepdog is ducks' best friend

Keywords Agriculture, Robots

A team of three post-graduate students have been working together at Silsoe Research Institute (SRI) to learn how interaction with animals can be made less stressful. Recent research has shown that animals respond well to machines and find them less threatening than humans and other animals. As some animals frequently come into contact with machinery (for example cows during milking) it is important to make the experience as stress free as possible for the animals.

The research has led the team to develop an autonomous robot which can enter an arena, gather a flock of ducks and manoeuvre them safely to a pre-determined goal (see Plate 5). This is a world first ­ no other robot system controls the behaviour of an animal and no methodology existed for designing one. The team did not set out to replace the sheepdog but the sheepdog's gather-and-fetch task was chosen as a good example of interaction with animals because of its familiarity and the interesting relationships between the dog, shepherd and flocking animals. Using ducks, instead of sheep, allowed the trials to take place on a smaller and more convenient scale. Duck flocking behaviour is recognised by shepherds as similar to sheep and ducks are often used in sheepdog training. The project also looked at flock dynamics and how individuals in a flock behave.

Plate 5 The Robot Sheepdog Project (RSP) is a collaboration between SRI and the universities of Bristol, Leeds and Oxford

The Robot Sheepdog Project (RSP) is a collaboration between SRI and the Universities of Bristol, Leeds and Oxford. This multi-disciplinary project covers robot building, machine vision, behavioural modelling and ethology. Instead of constantly using animals in the development process, the team built a minimal generalised model of the underlying flock behaviour and incorporated it into a computer simulation of the arena and the robot. The simulated ducks were called "ducklets" to emphasise the fact that only a subset of duck behaviour was modelled. Experimenting with the simulator led to the design of a robot control program, which moved the robot in just the right way to gather the ducks. The trials using the real robot and ducks were successful.

The robot itself is a vertical cylinder on wheels and is designed to work outdoors on short grass and in real time. It can accelerate to 4m per second (approximately 9mph), easily outpacing the ducks. It is 28cm high and 44cm wide and is covered in a soft plastic cover mounted on rubber springs to ensure duck safety. The robot system comprises a robot vehicle, computer and camera. Images from the camera are analysed by the computer programme to find the positions of the robot and flock. Combining this information with its knowledge of the goal position, the control programme can decide a path for the robot. Commands are then sent by radio to the robot which guides the ducks to the goal. In the tradition of mobile robotics, the robot is called Rover.

This project is part of a major research programme of robotics and engineering and will provide an excellent base for future research into animal-interactive robotics.

The members of the RSP team are:

  • Jane Henderson, University of Bristol and SRI ­ flock behaviour, duck response to visual stimuli.

  • Neil Sumpter, University of Leeds and SRI ­ computer vision and flock modelling.

  • Richard Vaughan, University of Oxford and SRI ­ robot and controller, flock modelling.

For general enquiries and photos please contact Mrs Sarah Campbell, Silsoe Research Institute, Wrest Park, Silsoe, Bedford MK45 4HS, UK. Tel: +44 (0)1525 860000; Fax: +44 (0)1525 860156; E-mail: SRI.PR@bbsrc.ac.uk

Related articles