智能轨迹控制割草机器人设计——基于FPGA神经网络

为了提高割草机器人自主导航和定位的精确性和智能性,设计了一种新型的基于FPGA神经网络算法的割草机器人。该设计采用FPGA可重构技术,以3层误差反向传播神经网络作为典型的模型来展开;利用成熟的BP算法公式,设计了割草机器人智能控制的模型;利用FPGA技术,设计了割草机器人的硬件系统;最后采用文本输入的设计方法,利用田间试验的方式,对机器人的轨迹规划能力和控制精度进行了验证。试验结果表明:利用FPGA和神经网络模型可以有效地穿越5个障碍物,并可得到满意的轨迹规划结果。将普通的PID控制器和神经网络PID控制器得到的控制结果误差进行了对比,结果表明:神经网络PID控制器得到的割草机器人控制误差明显比传统的PID控制器误差小。该方法为神经网络的硬件实现提供了可靠的理论基础。

Design of Intelligent Control Trajectory FPGA and Neural Network-Based on Robot Mower

In order to improve the accuracy and intelligence of the mowing robot autonomous navigation and localization, design a model of mowing robot based on FPGA ( field programmable gate array) and the neural network algorithm, the design using FPGA reconfigurable technology, to three layer back-propagation neural network as a typical model. Using BP algorithm formula, the design of the Intelligent Robot Mower control model. By using the FPGA technology, design the hardware system of the mowing robot. Finally, the design method of the input text, using field experiment, the robot trajectory planning ability and control precision was verified. The experimental results show that using FPGA and the neu-ral network model can effectively through five obstacles, and can get the satisfactory results of trajectory planning. Com-pared to the ordinary PID controller and neural network PID controller control the error, results can be seen from the, neural network PID controller of the mowing robot control error to be significantly better than the traditional PID controller error small, the method for neural network hardware implementation provides a reliable theoretical basis.