Research And Design Of Cleaning Robot Obstacle Avoidance System Based On Fuzzy Neural Network

Smart cleaning robot which integrates knowledge of artificial intelligence and automatic control, is a product based on the development of automatic control technology and artificial intelligence technology. As a kind of service robot, it includes sensor technology, automatically backfilled technology, automatic obstacle avoidance techniques, path planning technology and power technology. Compared to other technologies, the reasearch of obstacle avoidance is still much room for improvement. In order to improve the effciency of obstacle avoidance, this paper proposed a new strategy based on the fuzzy neural network and gived systematic design for the control system.Environment recognition is the premise of the cleaning robot obstacle avoidance. In this paper, environment recognition system consists of ultrasonic sensors and infrared reflective sensor which distribute in diferent directions. Firstly, the sensor information would be colected and processed by multi-sensor data fusion technology. Then the processed data was transmited to MCU module. Given the complexity of the above data processing, the obstacle avoidance system adoped ARM+DSP processing mode, which used STM32F103 and TMS320F2812 as the control system and the core processing chip. STM32F103 was used to process the interrupt signal, human-computer interaction and drive control module during the cleaning robot obstacle avoidance. TMS320F2812 was adobted for multi-sensor information fusion and fuzzy neural network processing algorithm. Upper and lower computer used CAN bus for real-time communication.This paper gived detailed design for each module including sensor module, power drive module, power module, serial communication module, based on the study of obstacle avoidance algorithm. This work gived the total distributed design of environment sensors and the circuit design for each sensor. The paper studied the differential driving mode and PWM mode, besides it gived detailed design about the H-bridge circuit. Based on the global coordinates and relative coordinates, the paper studied the cleaning robot model. Obstacle model was analyzed and categorized. This work also studied the basic theory of fuzzy neural network algorithm. On this basis, the paper designed the fuzzy control rules. The structure of BP neural network was also designed in detail. Finally, this work verified the specific effecs of the Fuzzy Neural Network obstacle avoidance strategy through simulation and experiments.