A robot-based intelligent management design for agricultural cyber-physical systems

This work proposes a robot-based intelligent management design that can intelligently manage the opening growth environments of crops. Different from the most existing agricultural robots, the FPGA device is used as the computing architecture of the proposed robot. In this robot design, a binarized neural network (BNN) hardware module provides the real-time and accurate detection of target crops, while cryptographic hardware functions ensure the security of sensor data transfers in an unsupervised environment. Based on a layered and virtualizable design method, the robot can not only control the actuators to maintain the ideal growth environment of crops but also dynamically adapt its cryptographic hardware functions to meet different requirements. Experiments show that, compared to the conventional microprocessor-based architecture, in this proposed robot, the processing time of cryptographic functions can be reduced by 99.46% to 99.62%, while the BNN inference can accelerate by a factor of 13,359.6 on average. To support all the AES, DESand 3DES functions, through system adaptivity, the robot design can reduce 37.4% of slice LUTs and 29.81% of slice registers in a Xilinx Zynq XC7Z020 device, while it can result in a power reduction of 13.6%.

Automated summary

This research introduces a robot-based intelligent management design using FPGA device and BNN hardware module to intelligently manage crop growth environments. Through cryptographic hardware functions and dynamic adjustment, the robot can enhance the efficiency of crop growth environment control and ensure data transmission security.