Design of an AGV for improved CDGPS-based control performance

This paper introduces and demonstrates a procedure to optimize the path-following performance of an autonomous vehicle system using Differential Global Positioning System (DGPS) navigation. In traditional applications of this type, the navigation, guidance and control system is designed and built for an existing vehicle. In contrast, in this work we carry out an integrated design of the vehicle with its controller. The integrated design process is oriented toward directly maximizing the path-following performance, rather than individually optimizing secondary vehicle characteristics such as the operating velocity. Using this method, system performance is maximized, and unnecessary and costly additional equipment or over-sized components can be avoided. Detailed covariance analyses and closed-loop simulations are carried out to evaluate the ultimate dynamic performance and to quantify the impact of each vehicle parameter. In particular, the performance sensitivity is analyzed with respect to parameters for which no obvious selection criteria exist, such as the vehicle's sense of motion or the positioning sensor location, which prove to be decisive characteristics. Experimental testing on a prototype rover confirms the simulations' results and validates the selection of the vehicle design parameters. As an example of practical application, the prototype rover is used as an automated lawnmower. The versatile modular design and the flexible parametric control system enable the successful integration of the mower to the vehicle system.