An adaptive framework for sensor planning in a coordinated multi-agent environment

Abstract

The objective of this research is to develop and automated system for multiple sensor planning based on the coordinated decisions of independent, intelligent agents. The problem domain is such that a single sensor system might not be able to provide adequate information for a given sensor task. Hence, it is necessary to incorporate multiple sensors in order to obtain complete information. The overall goal of the system is to perform feature inspection on one or more target features within a statis modeled environment. In this system, the sensors are mobile, each agent controls the position of a sensor and each agent has the ability to communicate with other agents in the environment.

The system includes a case based reasoning system that enables the agents to learn previous sensor arrangements and apply them to similar scenes. This decreases the amount of communication that is necessary to arrive at a solution. The agents may be trained off-line if necessary but are also quite capable of learning cases online.

The experiments demonstrate the feasibility of the system when using multiple mobile cameras as the sensor suite. Each camera is controlled by an agent and the vision task is the coverage of one or more target objects in a cluttered scene.

The system provides an efficient and reliable method to accomplish the sensor planning necessary to facilitate such tasks as feature inspection and feature detection of stationary targets. The use of agents as autonomous controllers provides a level of re-usability and scalability not normally found in other sensor planning systems. Such a system may be used in environments where the deployment of sensors needs to be an automated process due to potential hazards or where the configuration of the system needs to be changed frequently.