Time and place
TBA
Presenters
Charlie Street (University of Birmingham), Bruno Lacerda (University of Oxford), Masoumeh Mansouri (University of Birmingham)
Contact email: c.l.street@bham.ac.uk
Expected Background
Familiarity with robot planning and Markov models such as MDPs is desirable but not necessary.
Expected Gained Skills
* Model a multi-robot system under action outcome uncertainty, partial observability, temporal uncertainty, and the effects of robot interactions.
* Identify which multi-robot model and planning approach are most suitable for a given problem.
Brief Description
The demand for multi-robot systems is increasing, due to their performance, flexibility, and fault tolerance. Successful multi-robot deployments have been completed in fulfilment centres, fruit fields, and on roads. In each of these environments, robot behaviour is affected by the stochastic dynamics of the environment and the other robots, and so we require planning solutions that are robust to these sources of uncertainty. In this tutorial, we will explore a spectrum of multi-robot planning approaches which trade off between solution quality and scalability. These approaches differ in how they handle robot interactions during planning. We will also discuss modelling formalisms which capture multi-robot behaviour under different forms of uncertainty, such as temporal uncertainty, which affects robot action durations. This tutorial is accessible to anyone interested in deploying multi-robot systems in dynamic and uncertain environments.