The rapid advancement of unmanned aerial vehicle (UAV) technology has made swarm operations a critical paradigm for future combat. Fixed-wing drone swarms offer significant advantages including high combat efficiency, distributed coordination, robust damage resilience, and low attack cost. To validate the key technologies of intelligent planning and autonomous decision-making for fixed-wing drone swarms, we propose a hardware-in-the-loop (HIL) simulation system based on virtual-real combination. This system integrates physical modules—such as flight control terminals, ad-hoc network data links, and intelligent planners—with virtual modules including swarm mission planning software, six-degree-of-freedom (6-DOF) fixed-wing drone models, ad-hoc network data simulation, and HIL simulation software. In this paper, we detail the system architecture, key technologies, and a typical application scenario with 32 fixed-wing drones, demonstrating the system’s effectiveness in real-time, scalable, and cost-efficient verification of swarm algorithms.
1. Introduction
As fixed-wing drone technology matures, swarm operations represent the future of aerial combat. Fixed-wing drone swarms are characterized by strong combat effectiveness, high distribution coordination, significant anti-damage reconstruction, and low attack costs. They are considered one of the most promising operational styles in future intelligent warfare. In the research of fixed-wing drone swarms, simulation technologies—mathematical simulation, HIL simulation, and flight tests—play vital roles. Mathematical simulation uses software like MATLAB to model swarm algorithms, while HIL simulation incorporates physical hardware into the loop for more realistic verification. However, traditional HIL setups for large fixed-wing drone swarms are complex, costly, and time-consuming. To address these challenges, we developed a novel HIL simulation system that combines virtual and real components, enabling scalable, real-time, and cost-effective validation of fixed-wing drone swarm algorithms.
2. System Architecture
The proposed HIL simulation system architecture, as shown in the figure below (inserted at an appropriate position), consists of a mission planning software, a physical system, a virtual simulation system, a virtual-real ad-hoc communication system, a flight dynamics HIL simulation software, a ground monitoring software, and a dynamic visual simulation system. All modules are tightly coupled to form a closed-loop simulation environment.
The system architecture is summarized in Table 1, which lists the components and their functions.
| Module | Type | Function |
|---|---|---|
| Swarm Mission Planning Software | Virtual | Scenario construction, path planning, human-swarm interaction |
| Physical System (Flight Control, Ad-hoc, Intelligent Planner) | Real | Real hardware flying control, communication, onboard decision-making |
| Virtual Simulation System (6-DOF model, virtual planner, virtual ad-hoc) | Virtual | Digital twin of swarm for scalable simulation |
| Virtual-Real Ad-hoc Communication System | Hybrid | Data exchange between virtual and physical nodes |
| Flight Dynamics HIL Simulation Software | Virtual | Provides real-time aerodynamic data to physical flight controllers |
| Swarm Ground Monitoring Software | Virtual | 2D trajectory display, attitude visualization, data recording |
| Dynamic Visual Simulation System | Virtual | 3D visualization, mission effectiveness assessment |
2.1 Physical System
The physical system comprises three core hardware components: flight control terminals, ad-hoc network data links, and intelligent planners. Each fixed-wing drone in the real part is equipped with a flight controller based on a high-performance DSP and sensors, enabling attitude and position control. The ad-hoc data link uses MIMO (2×2) wireless modules to provide high-bandwidth, long-range communication among drones. The intelligent planner, built on a DSPc6000 series processor, runs algorithms for task allocation, path planning, and formation control. It communicates with the flight controller and ad-hoc link via serial interfaces, ensuring real-time data exchange. Table 2 summarizes the key specifications of these physical components.
| Component | Processor/Sensor | Communication Interface | Key Function |
|---|---|---|---|
| Flight Controller | DSP + IMU + GPS | UART, CAN | Attitude/position control |
| Ad-hoc Data Link | MIMO 2×2 RF | Ethernet, UART | Swarm data exchange |
| Intelligent Planner | DSPc6000 | UART, SPI | Online task planning |
2.2 Virtual System
The virtual system includes the following modules:
- Swarm Mission Planning Software: Developed with Qt/C++, it builds combat scenarios, generates pre-planned waypoints, and allows human-swarm interaction via a GUI.
- Virtual Simulation System: Contains a virtual ad-hoc network, an online mission planning simulator, and a 6-DOF fixed-wing drone model. It mimics the physical system’s behavior and can scale to many virtual drones.
- Flight Dynamics HIL Simulation Software: Simulates the aerodynamic model of a fixed-wing drone, providing sensor data (position, velocity, attitude) to the physical flight controller, making the HIL simulation realistic.
- Swarm Ground Monitoring Software: Displays 2D trajectories, real-time flight data, and attitude indicators. It also relays data to other modules.
- Dynamic Visual Simulation System: Provides 3D swarm visualization and evaluates mission performance (e.g., task completion rate).
3. Key Technologies
Three key technologies underpin the successful operation of the HIL simulation system: swarm planning, virtual-real state synchronization, and virtual ad-hoc networking.
3.1 Swarm Planning Technology
Swarm planning is fundamental for fixed-wing drone swarms. It involves optimizing the allocation of tasks among drones and designing flight paths under constraints such as fuel, safety distance, and mission requirements. The planning includes task assignment and trajectory generation. For task assignment, we use auction-based or consensus algorithms. For path planning, we incorporate Dubins curves or B-splines to generate flyable trajectories. The mathematical formulation of the task assignment problem can be expressed as:
$$ \max \sum_{i=1}^{N} \sum_{j=1}^{M} x_{ij} \cdot r_{ij} $$
subject to:
$$ \sum_{j=1}^{M} x_{ij} \leq 1, \quad \forall i, \quad \sum_{i=1}^{N} x_{ij} \leq 1, \quad \forall j, \quad x_{ij} \in \{0,1\} $$
where \( x_{ij} \) indicates whether drone \( i \) is assigned to task \( j \), and \( r_{ij} \) is the reward. The path planning ensures that each drone’s trajectory satisfies its dynamic constraints.
3.2 Virtual-Real State Synchronization
To ensure that the virtual and physical systems run synchronously, we implement three strategies:
- Unified 6-DOF Model: The same aerodynamic model is used in both the virtual simulation and the physical flight controller.
- Data Packet Timing: Both systems compute and transmit data at a fixed rate of 30 ms per packet. The virtual system’s computation step matches the physical controller’s update cycle.
- Clock Synchronization: Both systems use the host computer’s system time and start simulation simultaneously.
These measures guarantee that the state of fixed-wing drone swarms (positions, velocities, attitudes) is consistent between the virtual and physical domains, enabling meaningful combined simulation.
3.3 Virtual Ad-hoc Network
The virtual ad-hoc network is built using Socket and UDP multicast in C++. It emulates the real ad-hoc data link’s communication behavior, allowing virtual drones to exchange data as if they were physical nodes. The hybrid communication system (virtual + real) uses a cluster-based strategy (master-slave and distributed) and time-division transmission to reduce network load. The communication delay model is:
$$ \tau = \frac{L}{B} + \frac{d}{c} $$
where \( L \) is packet size, \( B \) is bandwidth, \( d \) is distance, and \( c \) is speed of light. This ensures real-time data flow for swarms up to 32 nodes.
4. Application Example
To validate the system, we conducted a 32-fixed-wing drone swarm scenario. The setup includes 5 physical drones (with real flight controllers, ad-hoc links, and intelligent planners) and 27 virtual drones. The scenario involves four sub-swarms performing reconnaissance, penetration, and strike missions. The process is as follows:
- Mission Planning: The swarm mission planning software constructs the combat scenario and generates pre-planned waypoints for all 32 fixed-wing drones. Waypoints are loaded to both physical and virtual intelligent planners via the hybrid ad-hoc network.
- Simulation Execution: The HIL simulation starts simultaneously. The physical flight controllers receive simulated aerodynamic data from the flight dynamics HIL software, while virtual drones run their 6-DOF models. The intelligent planners (both real and virtual) communicate via the ad-hoc network to coordinate task reallocation and formation changes.
- Monitoring and Evaluation: The ground monitoring software displays 2D trajectories and real-time attitude data. Simultaneously, data is relayed to the dynamic visual simulation system for 3D visualization and mission effectiveness assessment. Table 3 summarizes the simulation parameters.
| Parameter | Value |
|---|---|
| Number of drones | 32 (5 real, 27 virtual) |
| Communication packet rate | 30 ms |
| Simulation time step | 30 ms |
| Ad-hoc network topology | Hybrid (master-slave + distributed) |
| Mission type | Multi-target strike with formation change |
| Flight model | 6-DOF fixed-wing aerodynamic |
The results demonstrate that the HIL simulation system maintains real-time performance with a latency under 50 ms. The trajectory tracking error between virtual and real fixed-wing drones remains within 2 meters, verifying synchronization accuracy. The swarm successfully performed dynamic task reassignment and formation transitions, confirming the effectiveness of the algorithms.
5. Conclusion
We have designed and implemented a virtual-real combination HIL simulation system for fixed-wing drone swarms. By integrating physical hardware with scalable virtual models, the system provides a realistic, cost-effective, and flexible platform for testing swarm algorithms. The key technologies—swarm planning, state synchronization, and virtual ad-hoc networking—enable real-time coordinated simulation of up to 32 fixed-wing drones. Future work will incorporate onboard sensor data injection and battlefield situational awareness to further enhance autonomy. This system accelerates the development cycle of fixed-wing drone swarm technologies and reduces reliance on costly flight tests.