Researchers built an aerial robot that uses two independently controlled quadrotors and a long central frame to produce stronger interaction torque than a conventional single-drone manipulator. Called FAM-DQ, the system also controls the tool’s position and orientation in all six degrees of freedom, pointing toward drone-based maintenance tasks such as screw driving, valve turning, and bolt work.
What Did the Researchers Build?
FAM-DQ is a dual-quadrotor aerial manipulator designed for physical contact rather than simply flying, filming, or carrying sensors. The prototype combines a central frame, a steerable end-effector, and two propulsion modules mounted at opposite ends of the frame through passive joints.
The central frame uses a stiff carbon-fiber tube. This tube provides structural support, but it also acts as a long lever: thrust from the two propulsion modules creates larger torques when applied farther from the system’s center. That arrangement is intended to help the robot resist contact forces during tasks such as tightening a screw.
A gear-driven servo rotates the end-effector around the main frame. This allows the tool to point in different directions instead of remaining fixed relative to the drone body. The result is a platform designed to control both where the tool is located and how it is oriented.
Each propulsion module contains four 1204 brushless motors with 3015 propellers, its own flight controller, electronic speed controllers, and an onboard battery. The modules exchange commands and status information with a central controller through a high-speed wireless connection.
Unlike a conventional aerial manipulator, which usually places a robotic arm beneath an underactuated multirotor, FAM-DQ distributes propulsion around the central frame. This changes the way forces and torques reach the tool and gives the platform a route toward omnidirectional manipulation.
What Were the Key Results?
The experiments demonstrated four important capabilities: spatial trajectory tracking, attitude tracking, static torque generation, and screw driving. Indoor motion capture supplied real-time position feedback while the robot performed the tests.
In the trajectory experiments, FAM-DQ followed circular spatial motion while maintaining a separately commanded attitude. In the attitude tests, the platform changed its orientation while holding its position. These tests target the main weakness of conventional underactuated aerial manipulators, where changing the vehicle’s orientation is often tied closely to producing horizontal motion.
The static torque experiment examined the platform’s ability to create rotational force for contact tasks. The extended central frame increases the lever arm between the propulsion forces and the system center, allowing thrust to generate useful torque without depending entirely on a separate robotic arm.
The final demonstration attached a screwdriver bit to the end-effector and used it to drive a socket-head cap screw mounted on a fixed workpiece. This is a meaningful step beyond free-flight positioning because the tool must make contact, maintain alignment, and resist reaction forces.
The supplied paper text reports successful qualitative validation but does not provide numerical peak torque, tracking error, or screw-driving success-rate values. As a result, the strongest comparison is functional: FAM-DQ demonstrates decoupled position and attitude control plus physical interaction, while the paper leaves detailed numerical benchmarking against other aerial manipulators for future evaluation.
How Does FAM-DQ Work?
FAM-DQ achieves full actuation by combining the force from two independently controlled quadrotor modules with the geometry of the central frame. “Fully actuated” means the platform can command motion across all six pose components: movement along three axes and rotation around three axes. A conventional multirotor normally cannot generate these commands independently because its thrust direction is tightly linked to its body attitude.
The two propulsion modules do not have identical joint arrangements. Propulsion module 2 uses two axial passive joints, allowing its thrust vector to point in any direction within the central frame’s operating geometry. Propulsion module 1 uses one axial passive joint, so its thrust direction is constrained to a two-dimensional plane. The combination gives the full system more control authority than either module could provide alone.
The long central frame is important mechanically. If a thrust force acts at a distance from the center of mass, it produces a torque around that center. Increasing the distance increases the torque available from the same thrust, which is why the frame serves as a physical lever. This design also spreads the propulsion units apart, creating a larger effective moment arm for interaction.
The control system has three layers. A global controller first calculates the total thrust and torque required by the platform. A control-allocation stage then divides those demands into individual attitude and thrust commands for the two propulsion modules. Finally, local geometric controllers on each module track their assigned commands.
This hierarchy separates system-level task control from module-level flight stabilization. The global layer handles the desired tool motion, while the local layers keep each quadrotor oriented and producing the requested thrust. The approach is described as lightweight, which matters for an aerial platform with limited onboard computing, battery capacity, and payload.
The dynamics model treats the complete vehicle and the individual propulsion modules as linked but distinct systems. That separation accounts for the passive joints, the thrust-generated torques, and the coupling created by the long frame. The controller then uses this model to regulate position and attitude independently.
Why Does FAM-DQ Matter for Robotics?
FAM-DQ targets a difficult gap in aerial robotics: reaching locations that are hard or dangerous for ground robots while still applying useful physical force. Inspection drones can already survey infrastructure, but a robot that can press, turn, tighten, or loosen components could reduce the need for human access to elevated, confined, or hazardous areas.
Potential applications include industrial maintenance, valve operation, screw driving, bolt tightening, and contact-based inspection. These tasks require more than accurate flight. The tool must stay aligned while forces from the environment push back against the vehicle, and the robot must avoid translating unexpectedly when changing its orientation.
The design also offers a different path from simply adding a larger robotic arm to a multirotor. A long arm increases reach but can add mass, inertia, vibration, and control complexity. FAM-DQ instead uses distributed propulsion and a structural lever to generate interaction torque directly through the vehicle frame.
That does not make FAM-DQ a replacement for every industrial system. Ground-based robots remain better suited to sustained high-force work, while used industrial robots provide mature payload and repeatability options inside factories. However, aerial manipulators could complement them where fixed bases, cranes, or access platforms are impractical.
For operations managers evaluating automation options, the key question is not only whether a robot can reach a worksite, but whether it can maintain a stable tool pose under contact. FAM-DQ directly addresses that requirement and could influence the design of future used cobots for sale and aerial-ground robot teams.
What Are the Limitations and Open Questions?
The prototype was tested indoors with motion-capture feedback, so its performance in outdoor wind, GPS-denied spaces, or cluttered industrial environments remains unverified. Wireless communication between the propulsion modules and central controller also introduces a dependency that could become more important when latency, interference, or packet loss increases.
The reported experiments establish feasibility but do not provide the numerical benchmarks needed for a purchasing or deployment decision. Peak sustainable torque, energy consumption, payload capacity, tracking error, endurance, and comparison with a conventional aerial manipulator are not stated in the supplied text.
The passive joints simplify some aspects of force generation but can also introduce mechanical compliance and unmodeled motion. Future work must improve torque output and energy efficiency while handling dynamic contact, moving targets, tool changes, and unexpected impacts.
Frequently Asked Questions
What is FAM-DQ?
FAM-DQ is an aerial manipulator built from two quadrotor propulsion modules connected by a central frame. It is designed to control a tool in six degrees of freedom while producing torque for physical interaction.
Why use two quadrotors instead of one?
Two modules distribute thrust and create a larger lever arm around the central frame. This gives the platform more control authority and greater potential interaction torque than a single underactuated vehicle.
What tasks did the prototype perform?
The prototype tracked spatial and attitude trajectories, measured static torque, and drove a screw with a screwdriver mounted on its end-effector. The tests were performed indoors using motion-capture position feedback.
Is FAM-DQ ready for industrial deployment?
The experiments show a credible laboratory demonstration, not a production-ready system. Outdoor robustness, endurance, quantitative torque performance, safety, and operation without motion capture still require validation.
What Is the Bottom Line?
FAM-DQ shows how two coordinated quadrotors and a lever-like central frame can give an aerial robot both omnidirectional tool control and useful contact torque. Its screw-driving demonstration moves aerial robotics closer to practical maintenance, while its missing deployment benchmarks leave important engineering questions open.