Unmanned Roller Based on Visual SLAM

Today, with the rapid development of unmanned driving, the development of hardware facilities for autonomous vehicles has become highly mature. This paper achieves the driverless function of the roller based on the visual slam algorithm. Firstly, the double steel wheel vibration roller is modified, and the roller is equipped with a Zed stereo camera, a Tx2 industrial control computer, and a STM32 drive board. The ROS system is used as the carrier to ultimately achieve unmanned driving of the roller. After the stereo camera collects the image information, it is transmitted to the industrial computer tx2 for processing, and the point cloud map is converted for output. At the same time, the tx2 and stm32 driver boards also transmit data, and a path is planned for the roller through trajectory planning, thereby achieving the autonomous driving function of the roller. The experimental results show that the unmanned roller based on visual slam can effectively locate in real time and realize the automatic driving of the roller.