A controller based on fuzzy neural network is designed in the paper. Fuzzy neural networks are introduced into the information fusion of signals from sensors of an AS-R intelligent robot. Characteristic information of unknown environments acquired by ultrasonic sensors, infrared sensors and vision sensors are fused together in order to eliminate uncertainty caused by single sensor. Therefore, precise environment information can be obtained and the fault tolerant capabilities of robots are improved. It is proved that intelligent robots adopting multi-sensor information fusing techniques have better real-time and robust characteristics according to simulation results.