Robotic Soccer consists in its essence of multiple agents that need to collaborate in an adversarial environment to achieve specific objectives. This requirement sets the need to have high-level decision making reasoning in addition to low-level behavioral procedures for the agents. The architecture of our system addresses this combination of high-level and low-level reasoning by viewing the overall system as the combination of the mini-robots, a camera over-looking the playing field connected to a centralized interface computer, and several clients as the minds of the mini-robot players. Figure 1 sketches the building blocks of the architecture.
Figure: Our Robotic Soccer Architecture as a Distributed Deliberation and Reacting System.
In the rest of the paper we will describe in detail the specification of the mini-robots. As opposed to using radio communication as in a variety of other systems including [Sahota et al. 1995], we use infrared communication for its multiple advantages, as identified in [Suzuki et al. 1995].
Each mini-robot has on-board an infrared detector which allows it to receive and decode commands sent by a transmitter at the main host computer. The positioning of the mini-robots and of the ball will be perceived by a vision camera which passes this information to the main host computer. This processor maintains a map of the location of the mini-robots with different levels of confidence in its perception of the actual positions depending on the recency of the information gathered. The main host computer can be seen as the perception/action interface between the mini-robots and the high-level reasoning off-board clients: it perceives the environment, receives commands from the reasoning clients, and transmits these commands to the mini-robots. Commands can be broadcast or sent directly to individual agents. Each robot has a self identification binary code that is used in the infrared communication. Commands include positioning requests and navigation primitives, such as forward, backward, and turning moves at specific speeds. Robots can respond to positioning requests by turning on a high-intensity lighting source.
As we discuss in the future work section, we may provide the mini-robots with transmitting ability for direct communication between agents and some local sensing capabilities. We will consider these alternatives based on the empirical experience that we will be gathering in the near future using the vision camera.
In a nutshell, our architecture implements the overall robotic soccer system as a set of different platforms with different processing features. The mini-robots perform the physical navigation actions and can respond to positioning requests. Off-board computers perceive the environment through a vision camera, perform the high-level decision making and send commands to the mini-robots. Communication is done by infrared radiation.