Abstract

We have developed a system designed to autonomously guide robots to their designated positions with the correct pose for tasks related to the deployment of solar panels. Furthermore, we have created a simulation environment to virtually assess and test the performance of these robots. This proposal outlines a project aimed at creating URDF files for robots, establishing a Gazebo simulation environment, and conducting experiments to evaluate a car moving algorithm within the simulation. The project aims to provide a robust framework for algorithm development and testing in a simulated environment, reducing the need for extensive physical testing.

Introduction

Solar energy is widely acknowledged as a sustainable energy source. In western China, the favorable spatial and resource conditions for solar energy generation present a significant opportunity to establish extensive arrays for solar power generation. Given the considerable size and weight of solar panels, the automation of their deployment necessitates the involvement of robots to ensure efficiency and safety. To enhance the precision and effectiveness of the robotic deployment, an intelligent system is imperative for accurate target localization and motion planning. In response to this need, we have developed a system that enables robots to autonomously plan routes and navigate to designated targets based on sensor input. The complexity of the task is compounded by challenging environmental conditions such as air quality and sand impact. Conducting direct real-world tests under such conditions can result in prohibitively high costs. Therefore, we have implemented a simulation environment to evaluate and enhance the performance of robots in a virtual setting, ensuring both safety and efficiency improvements.

Method

URDF Design

• URDF File Design: Obtain STL files of different parts from CAD drawings
• URDF File Implementation: Utilize STL files to create the URDF files.
  • Define and set up the robot's links and joints.
  • Calculate the inertia matrix and other parameters.
• Add Transmission to joints
• The whole urdf in rviz

Gazebo Design

• Create a flat terrain environment with a desert texture.
• Create rack and solar panel models to facilitate reuse of gazebo simulation world in the future.
• We set the appropriate parameters (density, mass impact volume, etc.) for the rack, solar panel, and our own robots to make the simulation in gazebo conform to the formal situation.
• Based on the parameter information in the map.yaml file, the model is automatically imported into a specific location in the gazebo simulation world.
• Adding sensor plugins such as DGPS and lidar to the robot can provide real-time feedback on the robot's own latitude and longitude and yaw, as well as detect surrounding objects.
• Control the movement of our differential model robot in gazebo simulation environment. We use ros control packages.

Conclusion

In conclusion, this project successfully achieved its objectives of building a comprehensive robotic simulation environment through URDF modeling, Gazebo simulation, and the integration of radar and DGPS technologies. The combination of these elements provided a realistic and dynamic platform for testing and validating the performance of robotic systems in a controlled virtual environment. The URDF modeling process allowed for the accurate representation of the robot's structure, joints, and sensors. Gazebo, as the simulation environment, proved to be a powerful tool, offering a high-fidelity representation of the robot's interactions with its surroundings. The successful integration of radar and DGPS enhanced the robot's perception and localization capabilities within the simulated environment. The results demonstrated the effectiveness of the simulation setup, showcasing the robot's movement and interactions, as well as the accurate outputs from radar and DGPS sensors. The data analysis provided valuable insights into the system's performance, allowing for a deeper understanding of the robot's behavior and the reliability of sensor data.