A sensory system which uses more than one sensor to perform its sensory tasks is constantly confronted with two important questions. Firstly, how it should combine the information gathered by the different sensors, so as to obtain the finest possible sensory information? Secondly, how it should distribute the workload associated with the sensory tasks among its sensors so that the best performance in terms of speed and quality is achieved?

This dissertation is devoted to investigating these two important issues. A novel approach for aggregating the sensory data gathered by more than one sensor is proposed. The approach is based on a team consensus formulation which treats the sensors of the system as a team of experts cooperating with each other so as to achieve a common goal. An information theory based weighting scheme which operates recursively on the sensory data until a consensus is reached is proposed. This weighting scheme manifests both the quality of sensory data as well as the cooperation among the different sensors of the system. The approach is computationally tractable, and preliminary analytical as well as comparative studies have demonstrated that the approach can be applied to many practical sensory data fusion and integration applications, e.g., in multi-sensor robotic systems. Furthermore, the dissertation addresses the question of how the workload associated with the sensory tasks is distributed such that an optimal sensing performance is achieved. Two types of sensory tasks, namely, single criterion and multicriteria, are considered.