Invited Presentation:

Autonomous Mobile Robots: Large, Medium, Small and Swarms

Ray Jarvis
Intelligent Robotics Research Centre, Monash University, Clayton, Victoria 3168, Australia.


Autonomous mobile robot navigation is essentially about managing mobility in the context of an environment dimly perceived by a set of specialised sensors in fulfilment of a high level task. The task domain may be simply getting from a known point and orientation to a specified point and orientation in an efficient (if not strictly optimal) manner, exploring an initially unknown environment, covering all free space as in a cleaning operation, retrieving target objects at unknown locations, or cooperating with other robotic mechanisms in moving or manipulating objects. The complexity of such activities depends to some extent on how much is known about the environment - whether it is stationary and known initially, whether stationary but initially partially or wholly unknown, whether initially known but time-varying or both initially unknown and time-varying.

The five essential sub-systems needed for autonomous robots navigation are localisation, environmental mapping, path planning, communications and motion control. Each has a wide range of implementation possibilities; however, over and above the requirement to have each of the sub-systems robustly constructed and capable of operating in a timely fashion consistent with the restrictions of the task domain, is the need for sub-system integration towards a working whole which is reliable, safe, fast (enough) and affordable. Integration is not a straightforward process and needs special consideration of instrument interacting, payload and energy considerations, modes of control, computational support and communications requirements within the context of the application domain.

Thus, whilst there are general methodologies underpinning various localisation modes, visual, ultrasonic and tactile preceptors, path planning geometrics and communications and control regimes, selecting and mixing these together to function in supporting operational systems which vary in size, operate in indoor or outdoor settings or have distinct task requirements is a challenging yet fascinating design problem.

This paper presents a number of case studies relating to autonomous mobile robot systems that have been or are being built or used in the Intelligent Robotics Research Centre at Monash University, which span wide size and operational domains. These include an amphibious six wheeled all-terrain vehicle, a number of medium sized indoor vehicles and a swarm of quite small robots which operate on a video control plane.

The rationale behind the developments, the methodologies used in implementations and the future expectation regarding the ``missions'' of these ``creatures'' will be presented.

Conference Home Page