The work presented in this thesis deals with information sharing within mobile robot teams. Main focus is on information related to team positioning and localization. The goal of this research is to find models for how to best deploy a group of robots.
The motion platforms covered are well adapted. although very important for the results. Motion models are crucial when calculating the dead reckoning of each platform. Results in the appended work show that the motion model has heavy impact on the possibility to affect the uncertainty of a robot team as new ground is deployed.
The sensor models discussed may not necessarily be what is commonly used in the research field. Instead good approximations are used, that describe the main features of the sensor type. The focus is to present different sensor types and describe how the characteristics affect the uncertainty of the robot team, not to create new sensor models. A general model for sensors with polar measurement coordinates is derived. This can be used with different sensors with the same characteristics.
A framework for sharing information within the robot team is derived. This is based on earlier work presented as "The Stochastic Map". However, the framework is tailored to suite the purpose of this work. Different estimation methods are also discussed, with main focus is on Kalman filtering techniques.
The resulting part of this work present different ideas of what should be considered when deploying a group of robots. Results show how different dist ribution of sensors will affect the localization uncertainty of the group. It is also shown how different deployment formations affect the uncertainty within the team.
Linköping: Linköpings universitet , 2004. , 52 p.