Viewpoint-Driven Formation Control of Airships for Cooperative Target Tracking

Eric Price, Michael J. Black, Aamir Ahmad

TL;DR

This paper presents a novel Model Predictive Control (MPC) approach for controlling formations of airships to cooperatively track moving subjects. It addresses the unique non-holonomic constraints of airships, unlike previous methods designed for multi-rotors, and demonstrates its effectiveness through simulations and a real-world experiment.

What Problem Does This Paper Address?

Accurately tracking and capturing the motion of subjects (like animals in their natural habitat) from multiple aerial viewpoints is crucial for applications like motion capture. While multi-rotor drones have been used for this, they are noisy, have limited flight times, and can be unsafe. Airships offer a quieter, longer-endurance, and safer alternative. However, airships have complex, non-holonomic dynamics (their movement and orientation are tightly coupled, and they are affected by wind) that make existing multi-rotor formation control strategies inapplicable. This paper tackles the challenge of developing a formation control system for airships that can maintain optimal tracking conditions (subject centered in camera, safe distance, consistent angular separation) despite these complexities.

Methods

The core of the paper is a Model Predictive Control (MPC) framework adapted for airship formations. Key methodological contributions include:

  • Analytical Solutions for 2D and 3D Orbits: The authors derive analytical solutions for airship orbits around a subject, first in a simplified 2D case and then extending to a more realistic 3D model incorporating realistic airship physics (lateral angle of attack, roll, pitch, and altitude dynamics). This analysis helps understand the fundamental constraints and relationships between airship speed, subject velocity, and formation requirements.
  • Exploiting Periodic Relationships: A crucial insight is the exploitation of a periodic relationship between an airship's airspeed and its distance to the subject. This relationship is key to achieving stable orbits that satisfy the optimal MoCap conditions.
  • Numerical Optimization and MPC: Based on the analytical insights and a cost function that prioritizes subject centering and inter-airship angular separation, a numerical solver is developed. This solver is then used to implement a fixed-time MPC controller that generates control commands (acceleration and yaw rate) for the airships.
  • Realistic Airship Modeling: The paper incorporates detailed modeling of airship dynamics, including aerodynamic forces (lift and drag), lateral angle of attack, roll, pitch, and altitude control, which are essential for accurate control.

Datasets

Not discussed in the paper. The experiments rely on simulations and a real-world flight with a fixed subject position.

Key Insights and Ideas

  • Airships are Viable for Cooperative Tracking: The research demonstrates that airships, despite their complex dynamics, can be effectively controlled in formations for cooperative target tracking, offering a compelling alternative to noisy multi-rotors.
  • Wind and Subject Motion are Coupled Challenges: The paper highlights how both ambient wind and subject motion contribute to the relative velocity between airships and the subject, and the proposed controller effectively handles these combined effects.
  • Periodic Airspeed-Distance Relationship is Key: The discovery and exploitation of a periodic relationship between airship airspeed and its distance to the subject is a novel approach that enables stable, optimal tracking orbits.
  • 3D Dynamics Significantly Impact Orbits: The inclusion of realistic 3D airship dynamics (especially pitch angle) leads to less eccentric orbits compared to simplified 2D models, suggesting that the 3D model is crucial for practical applications.
  • MPC Offers Robustness: The MPC approach proves capable of handling non-optimal starting conditions and converging to optimal formation trajectories, demonstrating its robustness.

Key Results

  • Simulation Performance: Experiments show that as the number of airships in the formation increases (from 1 to 6), tracking accuracy and uncertainty decrease significantly.
  • Wind Robustness: The controller maintains good performance up to moderate wind speeds (around 1.4 m/s). Beyond this, tracking accuracy degrades, and subject visibility can be compromised, though the system still attempts to keep the subject in view.
  • Subject Motion Handling: Even with a moving subject exhibiting sharp turns (simulating worst-case scenarios), the MPC controller largely maintains the subject within the camera's Field of View (FOV) for a high percentage of frames (96-98%).
  • Real-World Validation: A real-world experiment with a single airship under significant wind conditions (1-2 m/s with gusts) demonstrated that the controller's performance aligns with simulation results, validating its practical applicability.

Limitations and Future Work

  • Analytical Boundary for 3D is Complex: While analytical solutions were derived, the exact boundary for subject velocity in the 3D case, considering all parameters, is not analytically determined and depends on multiple factors.
  • Assumptions on Wind and Subject State: The MPC implementation assumes constant wind and subject states over the planning horizon, which might not hold in highly dynamic environments.
  • Collision and Obstacle Avoidance: The current work focuses on formation control and does not explicitly address collision avoidance between airships or with static obstacles.
  • Generalization to Other Vehicles: The authors plan to investigate generalization to other non-holonomic vehicles, such as fixed-wing aircraft.
  • Real-time Obstacle Avoidance: Future work will explore integrating collision and obstacle avoidance into the airship formation control framework.

Summary updated

Abstract

For tracking and motion capture (MoCap) of animals in their natural habitat, a formation of safe and silent aerial platforms, such as airships with on-board cameras, is well suited. In our prior work we derived formation properties for optimal MoCap, which include maintaining constant angular separation between observers w.r.t. the subject, threshold distance to it and keeping it centered in the camera view. Unlike multi-rotors, airships have non-holonomic constrains and are affected by ambient wind. Their orientation and flight direction are also tightly coupled. Therefore a control scheme for multicopters that assumes independence of motion direction and orientation is not applicable. In this letter, we address this problem by first exploiting a periodic relationship between the airspeed of an airship and its distance to the subject. We use it to derive analytical and numeric solutions that satisfy the formation properties for optimal MoCap. Based on this, we develop an MPC-based formation controller. We perform theoretical analysis of our solution, boundary conditions of its applicability, extensive simulation experiments and a real world demonstration of our control method with an unmanned airship.
BibTeX
@article{price2023viewpoint,
  author = {Eric Price and Michael J. Black and Aamir Ahmad},
  title = {{Viewpoint-Driven Formation Control of Airships for Cooperative Target Tracking}},
  journal = {IEEE Robotics Autom. Lett.},
  year = {2023},
  url = {https://openreview.net/forum?id=wZVF8C_8DJ},
}
Figure