Affiliation:
1. Department of Computer Science, University of Essex, United Kingdom
2. Department of Electronics and Computer Science, University of Santiago de Compostela, Spain
3. Department of Automatic Control and Systems Engineering, University of Sheffield, United Kingdom
Abstract
Developing robust and reliable control code for autonomous mobile robots is difficult, because the interaction between a physical robot and the environment is highly complex, it is subject to noise and variation, and therefore partly unpredictable. This means that to date it is not possible to predict robot behaviour, based on theoretical models. Instead, current methods to develop robot control code still require a substantial trial-and-error component to the software design process. Such iterative refinement could be reduced, we argue, if a more profound theoretical understanding of robot-environment interaction existed. In this paper, we therefore present a modelling method that generates a faithful model of a robot's interaction with its environment, based on data logged while observing a physical robot's behaviour. Because this modelling method — nonlinear modelling using polynomials — is commonly used in the engineering discipline of system identification, we refer to it here as “robot identification”. We show in this paper that using robot identification to obtain a computer model of robot-environment interaction offers several distinct advantages: Very compact representations (one-line programs) of the robot control program are generated The model can be analysed, for example through sensitivity analysis, leading to a better understanding of the essential parameters underlying the robot's behaviour, and The generated, compact robot code can be used for cross-platform robot programming, allowing fast transfer of robot code from one type of robot to another. We demonstrate these points through experiments with a Magellan Pro and a Nomad 200 mobile robot.
Subject
Artificial Intelligence,Computer Science Applications,Software
Cited by
6 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献