This paper deals with a sensory network for mobile robotic systems with structured intelligence. A mobile robot requires close linkage of sensing, decision making, and action. To realize this, we propose structured intelligence for robotic systems. In this paper, we focus on the sensing ability for a mobile robot with a fuzzy controller tuned by the delta rule and whose architecture is optimized by a genetic algorithm. We apply the sensory network for controlling attention ranges for external sensors and for adjusting fuzzy controller output from the metalevel. As a simulation example, we apply the proposed method to mobile robot collision avoidance problems. Simulation results show that sensory networks control the attention range for perception and adjust fuzzy controller output based on given environmental conditions. We show the experimental results of mobile robot collision avoidance in work space including several obstacles.