Control research

Control is the process of deliberating and acting in and on the real world. Control of single and multiple agents in uncertain, unstructured and dynamic worlds, is a key research challenge in understanding and building intelligent systems. The control research program is focused on developing deep models of the operation of multiple interacting agents. The main outcomes of the research program is a new understanding of the roles of models in control, planning and coordination of complex intelligent systems.

Research in Control has three main elements:

1. Modelling

Understanding controllability and the possible motions of platforms or actions of systems underlies any ability to usefully plan and execute tasks. We have considerable expertise in modelling the kinematics and dynamics of large and complex platforms (air, land and sub-sea systems).

Research in modelling focuses on:

  • Understanding the effects of model fidelity on system performance, system complexity and sensor requirements.
  • Using platform models to allow learning of terrain and environment structure, and especially traversability of a terrain by an agent of a given modelled capability, and
  • Modelling and describing coupling between platforms, particularly synergistic systems in which one platform deploys or controls other smaller agents.

2. Planning

Planning is control at the level of sequences of operations defining a complete task; this is sometimes called “high-level” control and even “intelligent” control. The partners have a range
of expertise in advanced planning methods, from the world-champion RoboCup team, to the scheduling and operation of robots in applications such as mining or cargo handling. The challenge in planning is to evolve truly intelligent behaviour from lower-level control functions.

Research in planning focuses on:

  • The development of algorithmic models for planning under uncertainty, especially methods based on hybrid systems and modal logic.
  • The development of methods for learning or adapting behaviour through interaction with the environment, and
  • The development of cognitive languages for high-level agent control and for human-machine interaction.

Research in planning aims to develop robust and adaptive high-level control methods.

3. Cooperation

A fundamental, quantitative, understanding of coordination and cooperation between decentralised autonomous agents is a central goal of the control research programme.

Research in cooperation focuses on:

  • The development of mutual information gain as a utility metric for cooperation amongst agents engaged in information gathering, data fusion and mapping.
  • The development of static and dynamic (single and multiple time step) team information structures allowing development of endogenous agent coordination strategies, and
  • The development of piece-wise feed-forward optimal control strategies for groups of cooperating agents, especially as a computational means of implementing team theory methods. Research in cooperation defines a new direction in developing quantitative models of cooperation between multiple agents.