UAS Simulator for Modeling, Analysis and Control in Free Flight and Physical Interaction

Azarakhsh Keipour, Mohammadreza Mousaei, Junyi Geng, Dongwei Bai, Sebastian Scherer

TL;DR

ARCAD is a modular, MATLAB-based simulator designed to fast-track the development of Unmanned Aerial Systems (UAS), from standard quadrotors to complex hybrid VTOLs. It distinguishes itself by offering specialized tools for low-level control design and the ability to simulate physical interactions—like a drone writing on a wall—with minimal coding overhead.

What Problem Does This Paper Address?

In the world of drone research, developers often face a "Goldilocks" problem with simulators. General robotics simulators like Gazebo are powerful but can be cumbersome and time-consuming to extend for new aircraft architectures. On the other hand, simulators like AirSim prioritize photorealistic environments for high-level AI tasks but lack the "under-the-hood" flexibility needed for low-level control engineering. Finally, high-fidelity flight sims like XPlane are built for large aircraft and don't translate well to small, experimental UAS.

The authors introduce ARCAD (AirLab Rapid Controller and Aircraft Design) to fill this gap. It provides a "sandbox" where researchers can quickly define a new drone shape, automatically generate its control logic, and test how it handles both free flight and physical contact with its environment.

Methods

The core of ARCAD is its hierarchical modular design. This allows a researcher to swap out a specific component—say, a position controller—without having to rewrite the entire simulation loop.

Key technical features include:

  • Automatic Control Allocation: Users define rotor placements and angles; the simulator automatically generates the geometry and the control allocation matrix (the math that tells each motor how hard to spin to achieve a desired movement).
  • Wrench-Set Computation: The simulator can calculate "force and moment polytopes" in real-time. This helps engineers visualize the physical limits of their design—essentially showing the "envelope" of forces the drone can exert in any direction.
  • Hybrid Force-Position Control: Unlike standard simulators that only care about where the drone is, ARCAD includes modules for managing contact forces, enabling simulations of tasks like inspection or manipulation.
  • MATLAB Integration: By building on MATLAB, the authors leverage a language familiar to control engineers, making it easier to perform step-response analyses and flight optimizations.

Datasets

Not discussed in the paper. As this paper introduces a simulation framework rather than a machine learning model, it does not rely on external training datasets. Instead, it serves as a tool to generate synthetic data for aircraft analysis.

Key Insights and Ideas

  • "Flying Tools" vs. "Flying Cameras": Most simulators treat drones as moving sensors. ARCAD treats them as active agents capable of physical work. The inclusion of "wrench-set" analysis is a major "aha moment" for designers of fully-actuated drones, as it allows them to optimize rotor tilts for maximum pushing force.
  • Rapid Iteration over High Fidelity: The authors prioritize the ability to change a design and see the results in seconds. This is a strategic trade-off; while it lacks the visual polish of Unreal Engine-based sims, it provides much deeper insights into the aircraft's underlying dynamics.
  • Multimodal Support: The simulator isn't just for quadrotors. It natively supports fixed-wing aircraft and hybrid VTOLs (Vertical Take-Off and Landing), which are notoriously difficult to model because they transition between two different flight regimes.

Key Results

The paper demonstrates the simulator's utility through several complex use cases:

  • Physical Interaction: The authors successfully simulated a hexarotor with tilted arms performing a "writing" task on a wall, maintaining a precise 5 Newton force while moving.
  • Architecture Diversity: The tool was used to model and visualize a wide array of designs, including octorotors, multilink aircraft, and hybrid VTOLs with variable-pitch rotors.
  • Automated Analysis: The system generates "Attitude Response Analysis" plots automatically, providing instant feedback on metrics like rise time, delay, and settling time for new controller designs.

Limitations and Future Work

While ARCAD is a powerful tool for control engineers, the authors acknowledge a few areas for growth:

  • Connectivity: Currently, it lacks native integration with industry-standard protocols like ROS (Robot Operating System) and MAVLINK. Adding these would allow the simulator to talk directly to real flight hardware.
  • Library Expansion: The authors plan to provide more pre-written controllers to allow researchers to benchmark different control methods (e.g., PID vs. MPC) more easily.
  • Visuals: While functional, the 3D visualization is basic. Future iterations may include more textures and environment details, though likely without chasing the photorealism of gaming-engine simulators.

Source Code: The project is open-source and available on GitHub at keipour/aircraft-simulator-matlab.

Summary updated

Abstract

View Video Presentation: https://doi.org/10.2514/6.2023-1279.vid This paper presents the ARCAD simulator for the rapid development of Unmanned Aerial Systems (UAS), including underactuated and fully-actuated multirotors, fixed-wing aircraft, and Vertical Take-Off and Landing (VTOL) hybrid vehicles. The simulator is designed to accelerate these aircraft's modeling and control design. It provides various analyses of the design and operation, such as wrench-set computation, controller response, and flight optimization. In addition to simulating free flight, it can simulate the physical interaction of the aircraft with its environment. The simulator is written in MATLAB to allow rapid prototyping and is capable of generating graphical visualization of the aircraft and the environment in addition to generating the desired plots. It has been used to develop several real-world multirotor and VTOL applications. The source code is available at https://github.com/keipour/aircraft-simulator-matlab.
BibTeX
@inproceedings{keipour2023uas,
  author = {Azarakhsh Keipour and Mohammadreza Mousaei and Junyi Geng and Dongwei Bai and Sebastian Scherer},
  title = {{UAS Simulator for Modeling, Analysis and Control in Free Flight and Physical Interaction}},
  booktitle = {AIAA},
  year = {2023},
  doi = {10.2514/6.2023-1279},
}

Loading related content...

Figure