Li Yunwang, Dai Sumei, Shi Yong, Zhao Lala, Ding Minghua
School of Mechatronic Engineering, China University of Mining and Technology, Xuzhou 221116, China.
Department of Mechanical Engineering, Stevens Institute of Technology, Hoboken, NJ 07030, USA.
Sensors (Basel). 2019 Jul 5;19(13):2976. doi: 10.3390/s19132976.
Computer simulation is an effective means for the research of robot navigation algorithms. In order to implement real-time, three-dimensional, and visual navigation algorithm simulation, a method of algorithm simulation based on secondary development of Unity3D is proposed. With this method, a virtual robot prototype can be created quickly with the imported 3D robot model, virtual joints, and virtual sensors, and then the navigation simulation can be carried out using the virtual prototype with the algorithm script in the virtual environment. Firstly, the scripts of the virtual revolute joint, virtual LiDAR sensors, and terrain environment are written. Secondly, the A* algorithm is improved for navigation in unknown 3D space. Thirdly, taking the Mecanum wheel mobile robot as an example, the 3D robot model is imported into Unity3D, and the virtual joint, sensor, and navigation algorithm scripts are added to the model. Then, the navigation is simulated in static and dynamic environments using a virtual prototype. Finally, the navigation tests of the physical robot are carried out in the physical environment, and the test trajectory is compared with the simulation trajectory. The simulation and test results validate the algorithm simulation method based on the redevelopment of Unity3d, showing that it is feasible, efficient, and flexible.
计算机模拟是研究机器人导航算法的一种有效手段。为了实现实时、三维和可视化的导航算法模拟,提出了一种基于Unity3D二次开发的算法模拟方法。利用该方法,可以通过导入的3D机器人模型、虚拟关节和虚拟传感器快速创建虚拟机器人原型,然后在虚拟环境中使用带有算法脚本的虚拟原型进行导航模拟。首先,编写虚拟旋转关节、虚拟激光雷达传感器和地形环境的脚本。其次,对A*算法进行改进以在未知三维空间中导航。第三,以麦克纳姆轮移动机器人为例,将3D机器人模型导入Unity3D,并向模型添加虚拟关节、传感器和导航算法脚本。然后,使用虚拟原型在静态和动态环境中进行导航模拟。最后,在物理环境中对物理机器人进行导航测试,并将测试轨迹与模拟轨迹进行比较。模拟和测试结果验证了基于Unity3d二次开发的算法模拟方法,表明该方法可行、高效且灵活。