Diamantas, Sotirios Ch.
Biological and metric maps applied to robot homing
University of Southampton, School of Electronics and Computer Science,
Navigation is at the heart of today’s mobile robotic systems. The use of landmarks for the guidance of autonomous systems is an essential part of the process. Shape, size, and colour are some of the properties of the landmarks. In this work biological and other methods have been developed for navigating a simulated mobile robot back to its home. Optical flow which is not a property of the landmarks but a property of the camera motion, has been used for building topological maps which are used to localise a robot. The advantage of this approach is that there is no need for storing or retrieving images. Any computation is performed on the basis of vectors. Metric maps are generated using a parsimonious localisation and mapping algorithm using a laser range finder.
The purpose of this research is to effectively navigate a robot to its home position using computationally efficient biological and other techniques for navigation. Biology is seen as an alternative solution to the problems robots’ encounter which include algorithmic complexity, performance, and power consumption among others. Biological inspiration provides simple, yet effective methods for the solutions of such problems. The careful examination of such methods has twofold gain. The study of the principles of biological organisms entails making better autonomous systems that will, in turn help us perceive and understand better the underlying mechanisms of biological organisms.
It is therefore essential to have an understanding of how biological and robotic systems work. This work discusses the techniques and strategies found both in animals and robots. It then goes on to describe the implementation of a series of simulation algorithms inspired by biology and other fields that assist in the navigation process, and in particular, homing. Their results are discussed and analysed. The novelty of this research lies in its methods for robot homing that make use of optical flow information to recognise a location as well as methods based on a laser range finder to avoid obstacles, perform path planning, localise a robot, and map the environment.
Actions (login required)