Chapter 3: 3D Scene Modeling and Simulation
During the development of unmanned systems, a large number of semi-physical real-time simulation experiments for the motion control system are required. Traditional simulation interfaces typically display motion trajectories as two-dimensional curves, with motion state shown through text data. As the volume of data increases, this approach becomes non-intuitive and difficult to observe, understand, and analyze, making it hard to uncover the essential characteristics hidden within the data. A realistic image, on the other hand, conveys far more information than curves and text combined. Therefore, applying 3D visualization technology to unmanned system simulation enables system data to be presented in a more natural and understandable visual form, faithfully reproducing the entire operational process of the unmanned system.
The 3D simulation of unmanned systems uses 3D modeling technology to map vehicles and their surrounding environments into a virtual simulation environment. Driven by real-time data from the control computer, it faithfully reproduces the vehicle's motion state, providing a powerful analytical foundation for fault detection, real-time monitoring, training simulation, and other applications. The key to a 3D unmanned system simulation platform is creating a highly realistic simulation environment, with display fidelity and real-time performance being the top priorities.
Achieving both fidelity and real-time performance simultaneously is extremely challenging. For example, CG (Computer Graphics) technology in movies can render characters and scenes that are virtually indistinguishable from the real world, but behind the scenes are large numbers of high-performance computers performing long-term rendering and manual frame-by-frame processing. Similar to cutting-edge game development and VR (Virtual Reality) technology, 3D simulation of unmanned systems aims to generate dynamic scenes that are as realistic as possible (fidelity) in the shortest possible time (real-time). Additionally, 3D unmanned system simulation must be able to output real-time simulated data from various visual sensors (cameras, electro-optical pods, LiDAR, etc.), which demands even higher computer performance and relies on sensor modeling technology.
The RflySim Toolchain primarily uses RflySim3D, built on Unreal Engine, for 3D simulation. To accommodate different simulation needs, the toolchain includes built-in scenes and 3D models, along with corresponding external driver interfaces. To use custom scenes and 3D models, users should also master the basic usage of 3D processing software beyond Unreal Engine. Communication between RflySim3D and other toolchain components is implemented via UDP, with Python and Simulink control interfaces and script examples provided.
Session 2: Experimental Toolchain Configuration & 3D Scene Modeling and Simulation (Part 1)
Session 2: Experimental Toolchain Configuration & 3D Scene Modeling and Simulation (Part 2)
Session 2: Experimental Toolchain Configuration & 3D Scene Modeling and Simulation (Part 3)