Abstract: In this paper, an improved leader-follower configuration control algorithm is proposed to solve the formation problem in Vicon 3D motion capture system. Firstly, an appropriate velocity filter with weight is designed and the acceleration gain matrix is reconstructed to reduce noise and oscillations caused by velocity and acceleration computed by discrete position. Secondly, using Sliding Model Control (SMC) for formation, damping parameters are added to the signal of tracking control for each single quadrotor. Thirdly, formation flight with two quadrotors is carried out. Finally, the experiment results show the feasible and effectiveness.
Introduction
Cooperative formation flight—where multiple drones work together—is essential for tasks like search and rescue or military reconnaissance. However, controlling these formations is technically difficult, especially in high-precision indoor environments. This paper addresses a specific "noise" problem encountered when using motion capture systems like Vicon. While Vicon provides incredibly accurate position data (within 1mm), it provides this data in discrete steps. When a controller calculates velocity or acceleration by taking simple derivatives of these steps, it introduces significant high-frequency noise. This leads to "chattering"—rapid, jerky oscillations in the drone's movement—which reduces flight stability and can lead to system failure.
System Architecture
The researchers utilized a sophisticated hardware-in-the-loop environment to test their algorithm. The system integrates Vicon T40s cameras for positioning, Pixhawk flight controllers for onboard processing, and XBee wireless modules for real-time telemetry between the quadrotors and a ground station PC.
The Vicon system captures data at 200Hz, providing the sub-millimeter positioning required for tight formation flight. The ground station serves as the central hub, processing the Vicon data and transmitting control commands back to the UAVs.
Control Methodology
To tackle the chattering issue, the researchers proposed a multi-layered control strategy focused on signal smoothing and robust formation logic.
1. Weighted Velocity Filtering
Instead of calculating velocity based on just the current and previous position, they designed a filter that weights the current velocity against the two previous sampling intervals. This smooths out the spikes caused by discrete position jumps, providing a cleaner signal for the flight controller.
2. Formation Control via SMC
The system employs a leader-follower architecture where the "Follower" drone maintains a specific relative distance and angle to the "Leader." This is managed through a Sliding Mode Control (SMC) framework, chosen for its robustness against external disturbances.
3. Damping and Gain Reconstruction
To further suppress noise, the researchers reconstructed the acceleration gain matrix, effectively preventing the controller from overreacting to noisy acceleration estimates. Additionally, damping parameters were added to the PD (Proportional-Derivative) tracking control for each quadrotor, allowing the follower to track the leader's trajectory gracefully rather than mirroring every jitter.
Experimental Results
The algorithm was evaluated through physical flight experiments in a 5m × 5m × 2.7m indoor arena. The system successfully commanded two quadrotors to fly in a circular formation with a radius of approximately 1.0 meter.
Qualitative analysis of the flight telemetry showed highly smooth trajectories for both the leader and the follower. The velocity filter and damping parameters effectively eliminated the chattering seen in standard SMC implementations. The system maintained the rigorous positioning requirements of the Vicon environment while ensuring the follower accurately tracked the reference trajectory generated by the leader's movement.
Conclusion and Future Work
This research demonstrates that high-precision sensors require sophisticated filtering to be useful for real-time control. By combining weighted filtering with a robust SMC framework, the authors achieved stable, smooth formation flight. Future work will focus on moving this system to outdoor environments where Vicon is unavailable and implementing dynamic formation reconstruction to allow the swarm to navigate around obstacles.
Summary updated