Chapter 1: Introduction and System Architecture¶
Intelligent Unmanned Systems (IUS), as an interdisciplinary integration of cutting-edge technologies—including artificial intelligence, machine vision, and swarm intelligence—serve as a concentrated embodiment of contemporary technological advancement. This chapter systematically reviews the foundational theoretical framework and control principles of intelligent unmanned systems, introduces the basic architecture of the RflySim toolchain, and, taking multi-rotor drones as an example, outlines their assembly and debugging procedures, laying a solid theoretical and conceptual foundation for your subsequent in-depth learning.
1.1 Background and Theory¶
With the rapid development of cutting-edge technologies such as artificial intelligence, robotics, embodied intelligence, and autonomous driving, related concepts continue to emerge. Intelligent Unmanned Systems (IUS) build upon traditional unmanned systems by integrating artificial intelligence, endowing them with autonomous capabilities—including perception, reasoning, decision-making, and execution. Based on system scale and organizational complexity, IUS can be categorized into single-unit, formation, and swarm-cooperation systems, operating across underwater, terrestrial, and aerial platforms.
Despite the diversity of intelligent unmanned systems, their underlying concepts and architectures remain highly unified. Scientifically, they can be divided into six core modules: the airframe structure layer, the perception and localization layer, the control and decision-making layer, the actuation and execution layer, the environmental interaction layer, and the swarm collaboration layer. In spatial attitude estimation and control law design, the local navigation coordinate systems (NED / ENU) and the onboard front-right-down (FRD) coordinate system are widely adopted, with vehicle 3D spatial attitude precisely described using Euler angles or quaternions.
Taking multi-rotor drones as an example, a typical physical system comprises an airframe structure, a propulsion system (motors and propellers), a perception system (IMU and GNSS), an autopilot (flight control system), and a data link communication system. After assembly, the system must undergo rigorous steps—including firmware flashing, sensor calibration, power system testing, and field test flights—before it can be deployed for formal R&D experiments.
1.2 Framework and Interfaces¶
The RflySim toolchain supports full-stack development—from low-level control filtering to high-level intelligent perception—and enables smooth transitions (Sim2Real) from pure-software simulation (SITL) to hardware-in-the-loop simulation (HITL) and real hardware deployment.
1.2.1 Overview of the RflySim Toolchain¶
RflySim is a professional, open, and research- and education-oriented simulation and development toolchain for intelligent unmanned systems. Adhering to the core principles of Model-Based Design (MBD) and full hardware-in-the-loop coverage, it provides developers with an integrated development framework supporting multi-rotor, fixed-wing, and unmanned ground vehicle platforms, and natively supports large-scale swarm distributed adversarial simulations involving over 100 nodes.
1.2.2 Core Components and Interfaces¶
The daily operation of RflySim relies on the collaboration of multiple software components: the core simulation engine is CopterSim, a kinematic simulation engine; high-fidelity visual and physical simulation environments are built upon Unreal Engine / RflySim3D; and mission planning and low-level monitoring are handled by the QGroundControl ground station.
For developers, the platform offers not only a firmware-level automatic code generation channel—PX4PSP—based on MATLAB/Simulink for low-level development, but also a rich set of Python / ROS interface libraries (RflySimSDK) for upper-layer AI validation and development.
1.2.3 Recommended Learning Path¶
This course begins with foundational system theory and software operation (Chapters 1–2), guiding you into the construction of high-fidelity 3D environments and mathematical models for various vehicle platforms (Chapters 3–4). After mastering low-level filter design and the core flight control closed-loop (Chapters 5–7), you will advance to high-level practical applications—including multimodal perception, visual mapping, and swarm coordination and game-theoretic adversarial scenarios (Chapters 8–10).
1.3 Showcase of Advanced Cases¶
¶
1.4 Course-Related Video Lectures¶
Lecture recordings for this chapter:
1.5 Chapter Experiment Cases¶
The verification experiments and guided case studies related to this chapter are located in the [Installation Directory]\RflySimAPIs\1.RflySimIntro folder.
1.5.1 Interface Learning Experiments¶
Located in the 1.RflySimIntro\0.ApiExps folder, these experiments cover foundational platform interface tutorials and general introductions to each tool.