MARS2: Agile and Robust Flight Control of Modular Aerial Robot Systems

Rui Huang, Zhiqian Cai, Siyu Tang, Jialin Zhang, Zhenyu Zhang, Lin Zhao†
National University of Singapore
Workshop paper Code - Coming soon Full paper - Coming soon (arXiv)

Learning from model weights


Illustration of the proposed MARS-Dragonfly strategy. (a) Dragonfly-inspired MARS design. (b) Equivalent model mapping control from the virtual quadrotor to MARS. (c) Novel transportation modes enabled by MARS. (d) Versatile trajectory planning and tracking, enabling MARS to operate like a quadrotor and eliminating the need for customized algorithms.

Abstract

Modular Aerial Robot Systems (MARS), composed of multiple self-reconfigurable drone units, offer high adaptability to diverse mission scenarios and fault conditions. Existing docking mechanism designs for MARS are prone to significant oscillations during docking and separation. Moreover, their control systems are conservative and sensitive to disturbances, which only allow quasi-static hovering flight and waypoint-based flight. To address these disadvantages, we first develop a highly integrated mechanical system that significantly simplifies passive docking, detection-free passive locking, and active separation by using a single servo motor. Our experimental results demonstrate substantial improvements in smooth and stable docking and separation performance. To fully unleash the redundant control actuation of MARS, we further propose an equivalent modeling framework that abstracts arbitrary MARS configurations into a unified virtual quadrotor representation. Building on this abstraction, we design and implement a model predictive control algorithm for MARS of arbitrary configurations. Moreover, we develop efficient algorithms to map the virtual thrust and torque control commands to those of individual units.

Simulation in gazebo

Trajectory Tracking

(a) 2 x 1
(b) 1 x 1 x 2
(c) L-shaped transportation with 2 units
(d) 3 x 1
(e) 2 x 1 transportation
(f) L-shape

Agile flight

(a) Pitch: 45 deg
(b) Pitch: 60 deg
(c) Roll: 45 deg

Docking Mechanism Experiments

(a) Passive detection-free locking
(b) Active repulsion-based separation
(c) Pendulum test evaluates docked connection strength

Real-world Experiments

Mid-air docking

(a) Smooth docking with minimal disturbance (Ours)
(b) Large post-docking position oscillation [1]
(c) Large post-docking position oscillation [2]

Active repulsion-based separation

(a) Minimizing attitude change and enhancing safety and robustness (Ours)
(b) Significant attitude and position oscillations during separation [3]

Trajectory tracking

Circular trajectory: stable continuous flight at 1 m/s
Agile trajectory: robust performance under agile, high-speed flight

Collision-free trajectory tracking

MARS accurately follows the collision-free trajectories from our previous work [4]

Agile trajectory tracking

(a) Pitch: 20 deg
(b) Pitch: 30 deg
(c) Pitch: 40 deg
(d) Roll: 20 deg

Object transportation

(a) Cooperative transportation with a 40 cm basket
(b) Transportation with a sloshing load (basketball)

Full docking–separation process

(a) Docking
(b) Separation

Poster

BibTeX

If you find our work useful, please consider citing:

@article{xxx,
          title={MARS},
          author={},
          journal={arXiv preprint arXiv:},
          year={2025}
        }

References

[1] Saldana, David, Bruno Gabrich, Guanrui Li, Mark Yim, and Vijay Kumar. "Modquad: The flying modular structure that self-assembles in midair." In 2018 IEEE international conference on robotics and automation (ICRA), pp. 691-698. IEEE, 2018.
[2] Li, Guanrui, Bruno Gabrich, David Saldana, Jnaneshwar Das, Vijay Kumar, and Mark Yim. "ModQuad-Vi: A vision-based self-assembling modular quadrotor." In 2019 International Conference on Robotics and Automation (ICRA), pp. 346-352. IEEE, 2019.
[3] Saldana, David, Parakh M. Gupta, and Vijay Kumar. "Design and control of aerial modules for inflight self-disassembly." IEEE Robotics and Automation Letters 4, no. 4 (2019): 3410-3417.
[4] Rui Huang, Zhenyu Zhang, Siyu Tang, Zhiqian Cai, and Lin Zhao. "Robust Fault-Tolerant Control and Agile Trajectory Planning for Modular Aerial Robotic Systems." In 2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). arXiv preprint arXiv:2503.09351 (2025).

Our related works

[1] Rui Huang, Siyu Tang, Zhiqian Cai, and Lin Zhao. "Robust Self-Reconfiguration for Fault-Tolerant Control of Modular Aerial Robot Systems." In 2025 International Conference on Robotics and Automation (ICRA). arXiv preprint
[2] Rui Huang, Zhenyu Zhang, Siyu Tang, Zhiqian Cai, and Lin Zhao. "MARS-FTCP: Robust Fault-Tolerant Control and Agile Trajectory Planning for Modular Aerial Robot Systems." In 2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). arXiv preprint