Introduction · GitBook

The RflySim platform is an intelligent unmanned swarm system development and testing platform initiated by the Reliable Flight Control Group (Rfly) in 2016. It is led by Quan Quan (full professor from Beihang University) and developed by Xunhua Dai (Associate Professor from Central South University). The advanced functionalities are further developed and long-term operation and maintenance are overseen by FeatsLab under Beijing Droneyee Intelligent Technology Co., Ltd.. It is specially developed for the frontier research fields of unmanned platform control system development, large-scale cluster collaboration, artificial intelligence vision and so on. The platform adopts a model-based design (MBD) concept, is based on Pixhawk/PX4, MATLAB/Simulink, ROS, shelf intelligent hardware, and the like, and can carry out (but not limited to) simulation and real flight/motion of unmanned agent control, simulation and real flight/motion of unmanned agent clusters, As well as simulation and real flight/motion based on unmanned agent vision. In order to solve the above problems, unmanned system modeling, controller design, Software-In-The-Loop (SIL) and Hardware-In-The-Loop (HIL) can be carried out. Through the automatic code generation technology of MATLAB/Simulink, the controller can be conveniently and automatically downloaded to the hardware for HIL simulation and actual flight test to realize Sim2Real. ○ See link ○ for details

The RflySim platform is currently available in three versions: Free version, full version, enterprise customized version, consult [email protected]).

For more differences between versions, see: ○ Version differences ○

The advanced and customized versions of RflySim platform have the following features:

Unity。 The whole research framework is extended to all unmanned control systems, forming a standard automatic development, test and evaluation framework system;

High reduction degree of UAV physical characteristics simulation。 Its developers are all UAV research teams with rich experience in UAV research;

Ease of use。 One-key installation, one-key code generation, one-key firmware deployment, one-key software and hardware in the loop simulation and fast flight are carried out under the Windows platform, which is very convenient and easy to use. Users do not need to understand the underlying knowledge of flight control source code, Linux programming, C/C + + programming, network communication, aircraft assembly, etc. They only need to have basic Simulink (or Python) knowledge to quickly verify their algorithms and apply them to the real machine, which is helpful to focus more on the development and testing of algorithms.

Fully distributed architecture。 All application software can be opened on the same or multiple computers, and each application can send and receive messages to and from each other through the UDP network, so the distributed architecture is very suitable for large-scale UAV cluster simulation test with vision;

Support simulation of multiple models。 Support car, fixed-wing, vertical take-off and landing aircraft (VTOL) and other models. Users can build a rack model in Simulink according to the standard interface, and then automatically generate DLL files for HIL simulation. Further, the experimental platform can be extended to any unmanned system;

Support large-scale UAV cluster SIL/HIL simulation。 Under the same LAN, developers can use CopterSim to connect multiple Pixhawks for hardware or software in-loop simulation. At the same time, Simulink or C + + program can be used to control the aircraft, and the control commands will be sent to Pixhawk by Mavlink protocol through serial port (data transmission) or network (WIFI);

Provide highly realistic 3D view。 Provide source code and tutorials to help developers build highly realistic 3D scenes in Unreal Engine (UE) for indoor and outdoor environment simulation or development based on visual algorithms; Scenes support physical collision engine, global terrain and map, OSGB + Cesium oblique photography visual map import, custom GPS coordinates, arbitrary multi-window switching observation, RGB, depth, grayscale, IMU, lidar and other sensor data output, support shared memory or UDP image direct transmission to designated IP address. And can be use for hardware-in-the-loop SLAM simulation of an airborne computer.

Support for vision-based control。 The 3D scene platform based on UE also supports the view switching function, so that the image data of multiple views can be obtained conveniently. It also supports real-time acquisition and processing of image data in Simulink, Python, C/C + + and other code platforms through shared memory, and the processed visual data can be returned to CopterSim or Simulink control through UDP to form a hardware-in-the-loop simulation closed loop with vision.

Support multiple fault injection。 The fault types that can be realized include model fault, communication fault and environment fault.