Abstract: In order to solve the problem of autonomous navigation of robots in greenhouse environment, a navigation system for greenhouse transportation robot based on dual-lidar was developed in this paper, The navigation system consisted of front and rear lidar, compared with the single lidar, the front and rear double lidar could increase the scanning range, reduce the blind area of tracing, improve the efficiency and accuracy of surveying and mapping, and improve the real-time obstacle avoidance ability of the robot. The navigation system of greenhouse transportation robot was composed of a remote monitoring platform and an on-board system. The remote monitoring platform was responsible for selecting the working mode of the on-board system, issuing the the instruction of target points and displaying the location information of the on-board system. As the executor of the instructions, the on-board system was responsible for receiving and executing task instructions ordered by the monitoring platform. Through the real-time communication through wireless network, the remote monitoring platform and the on-board system jointly complete the autonomous navigation task of greenhouse transportation robot. The on-board system hardware mainly consisted of a driving module, a control module, an environmental information perception module, a communication module and a power supply module. The on-board systems software was divided into three layers, user interaction layer, information processing layer and execution layer. The user interaction layer was an open-source Ubuntu-based navigation task scheduler that responsible for adjusting the working mode of the on-board system and issuing target points instructions. The information processing layer was the real-time positioning and map building and fixed-point navigation program based on the Robot Operating System(ROS), which was responsible for collecting the motion information of on-board system and environmental information from lidar, and conducting information fusion. According to the control command and on-board system position and attitude information, map construction, path planning and autonomous navigation were carried out. The executive layer was the mobile platform control program based on the open source real-time operating system of Ubuntu. By collecting the speed information of the encoder, the classical PID algorithm was used to adjust and output the desired Pulse Width Modulation(PWM) wave to control the motor speed, so as to realize the stable and safe movement of the on-board system. The dynamic window algorithm was used to plan the local optimal path to reduce walking time and energy consumption. The test results showed that when the on-board system ran at the speed of 0.2, 0.5 and 0.8 m/s , the average deviation and the standard deviation between the actual navigation path and the target path was less than 13 and 5 cm, respectively; the average value, the root mean square error, the standard deviation of the lateral deviation and longitudinal deviation at each target point was not more than 9, 11.2 and 5 cm, respectively; the average value, the root mean square error, and the standard deviation of the course deviation was less than 10°, 12° and 6°, respectively, which met the navigation accuracy requirements of robot transportation in greenhouse.