Identification and simulation of an automated guided vechile for minimal sensor applications
Aaron Dann · University of Canterbury Research Repository (University of Canterbury) · 1996
The problem of controlling an Automated Guided Vehicles (AGV) with the minimum number of sensors is considered. Sensors add cost and complexity to an AGV both electrically and in terms of increased computational requirements of the controller. Computer simulations are proposed to model the behaviour of the AGV Models of the dynamics of an AGV are proposed and simulated at varying levels of complexity using commercially available numerical software. In order to model the AGV accurately, aspects of the control system and the physical system had to be analysed. Laboratory experiments were designed and performed, and the results were analysed to determine the dynamic properties of sub-systems of the AGV To provide a datum for comparison to the simulations, measurements were made of the performance of an AGV under a variety of control conditions corresponding to the computer models. Comparisons of the simulations and the AGV performance are discussed and suggestions are made for improving the AGV and its control system. The models presented in this thesis demonstrate a good correlation for low performance AGV s in non-rigorous conditions, or well loaded AGV s on good traction surfaces. However they do not accurately represent the AGV at the limits of traction. Two mechanical improvements to the University of Canterbury (UOC) Mk-II AGV are suggested, including the addition of softer compound tyres for use on hard, painted surfaces, and the design of a gear train with lower backlash.