Converting electrical drive into predictable mechanical force is a universal hurdle in robotics. That gets harder on weight-constrained, high-vibration platforms like drones, where every gram and every bit of play works against flight performance. Linear actuators offer a direct, compact solution to that problem.
Researchers at Ontario Tech University took on this challenge directly with an autonomous flying excavator, a drone platform designed to identify, dig, and clear material entirely while airborne. The arm and bucket on the final prototype were driven by Actuonix L12 series micro linear actuators, and the team worked through three separate airframe designs before landing on a configuration that could hold its balance during excavation.
Technical Specifications: For more info about the specific product line used in this application, you check out our L12 Actuator Product Page
Designing a Drone Airframe Around a Robotic Arm
Adding a robotic arm to a flying platform changes the vehicle's center of mass the moment the arm extends or the bucket fills with material. A drone that flies level when empty can become unstable the instant it starts to dig, and that instability was the central problem the research team had to solve across three separate airframe designs.
Overall dimensions of the optimized airframe, showing the shortened arm and landing gear used in the final design.
- Design 1, F450 quadcopter frame: The team's first attempt combined a basic F450 quadcopter frame with the boom and bucket manipulator. The added arm shifted the platform's center of mass significantly off center, and the resulting instability caused the platform to crash during flight testing.
- Design 2, Tarot octocopter frame: To handle the added weight and lifting demands, the team scaled up to a Tarot octocopter frame with eight motors. This version successfully stabilized during excavation, but the platform's overall weight, close to 20 kg, made it difficult to hold a fixed position. Continuous drift caused the ground sensor to lose lock on the excavation site, and the platform crashed a second time.
- Design 3, optimized octocopter frame: In the final design, the team shortened the manipulator from 410 mm to 374 mm and replaced the boom with a bracket. This allowed the arm to balance using only two actuators instead of the earlier linkage, and the team also significantly shortened the landing gear legs to reduce the platform's profile and prevent it from tumbling during excavation.
The L12 was a fitting choice for this kind of weight-sensitive redesign. The team specifically used the 100mm stroke, 210:1 gear ratio configuration, weighing 56g per unit, or 112g for the pair driving the arm and bucket. That's a small fraction of the completed platform's total mass of 3.357 kg, a figure the researchers calculated directly from the weight of every individual component on board, down to the drone frame, motors, battery, and depth camera.
How the Camera, Kinematics, and Actuators Work Together
With the airframe stabilized, the platform needed a way to translate what its sensors saw into precise physical movement of the arm and bucket. This is where the two L12 actuators and the onboard vision system had to work in sync, with each actuator handling a distinct part of the excavation task.
The assembled prototype, labeled with the flight controller, companion computer, GPS, and depth camera positioned alongside the arm and bucket actuators.
Boom Actuator: Reach and Depth
An Intel D415 depth camera, mounted on the front of the platform, captured point cloud data of the excavation soil ahead of the drone. That data was cropped and filtered down to the region of interest, then run through an inverse kinematics calculation to determine the boom actuator's required stroke length to reach the target depth and position.
Bucket Actuator: Collection and Dumping
The second actuator controlled the bucket angle directly, curling to collect material during digging and rotating to release it during the dumping phase. The system tracked a dumping state so that camera data was ignored while the bucket was in that position, since the ground profile does not change during a dump.
Both actuators were fed their target stroke lengths through an Arduino, which received the calculated values from the onboard Jetson Nano and triggered each actuator to reach its commanded position. The paper credits the actuators with integrated PID controllers, which regulate the bucket tip's position based on the kinematics of the manipulator arm.
Vibration and EMI Management in a Compact Payload
Moving from a CAD model to physical hardware introduced two problems that don't show up in simulation: mechanical vibration from the drone's motors, and electromagnetic interference from its high-current power lines.
- Structural fasteners: Nylock nuts secured the drone's structural components to prevent them from loosening under the vibration generated during flight.
- EMI and magnetic separation: The wiring harness was routed to keep the high-current lines feeding the electronic speed controllers separated from the compass and GPS modules, preventing signal corruption and navigational drift.
- Power delivery: Battery voltage was passed through a step-down regulator so each onboard component received power proportional to its rating.
The full ground excavation flight, from initial approach through digging, dumping, and safe landing.
The completed platform was tested in two scenarios. In the first, the drone flew autonomously to a target site, excavated soil from the recognized ground, dumped it at a separate location, and landed safely. The researchers measured a root mean square error of just 0.00894 m between the desired and actual flight path, with the platform maintaining an average lean angle of 20 degrees against a maximum permitted 30 degrees. In the second, the drone attempted to excavate a raised target roughly 0.75 m off the ground while remaining fully airborne, with a measured RMSE of 0.02 m between the desired and actual path and an average lean angle of 15 degrees. It successfully reached the target using GPS and RTK positioning, though the manipulator lost stability during digging even as the platform itself stayed balanced and landed safely afterward.
Taken together, the two tests show a real difference between excavation from a stable landed position and excavation while hovering, and the researchers themselves flagged airborne stability as an area for further work. What the project does demonstrate clearly is that a pair of off-the-shelf micro linear actuators, paired with camera-driven kinematics, is enough to drive a functional excavation arm on a flying platform without the weight penalty of a custom gear-driven system. That's a meaningful result for a first-generation research prototype, and it points toward a growing set of applications, inspection, sampling, and remote construction tasks, where a compact linear actuator can do the job a much heavier mechanism used to require.
Technical FAQ: Linear Actuators in Aerial Robotic Arms
Why does the final design only use two actuators?
Shortening the manipulator arm from 410 mm to 374 mm reduced the mechanical demands on the linkage enough that the team could balance the platform using one actuator for the boom and one for the bucket. Keeping the actuator count low kept weight close to the drone's center of mass, which was the recurring problem across the earlier airframe designs.
How does the drone know where to dig?
Positioning was driven by an onboard depth camera rather than the actuators themselves. The camera captured point cloud data of the ground ahead of the platform, and an inverse kinematics calculation converted that data into the stroke lengths each actuator needed to reach the target position. The actuators execute the movement, but the camera and onboard computer decide where that movement needs to go.
Why choose the L12 series for a flying application?
Weight was the deciding factor throughout this project's development, with two of the three airframe designs failing in part due to excess mass. The team used the 100mm stroke, 210:1 gear ratio L12 configuration, at 56g per actuator, or 112g total for the pair driving the arm and bucket, against a total platform mass of 3.357 kg. The researchers calculated the airborne excavation task alone required about 10N of digging force, determined through their own trial-and-error testing, well within what this L12 configuration is rated to deliver.
Was the flying excavator fully successful?
The ground-based excavation cycle, approach, dig, dump, and landing, completed successfully with strong positional accuracy. The airborne excavation test reached its target and landed safely, but the manipulator lost stability while digging, and the researchers noted this as an open challenge for future work. As with most first-generation research prototypes, the results are a mix of clear proof of concept and areas still being refined.
Source: Zaman, A.; Seo, J. Design and Control of Autonomous Flying Excavator. Machines 2024, 12, 23. https://doi.org/10.3390/machines12010023
Moving From Research Prototype to Real Application
This project is a strong example of how a compact, lightweight actuator can replace a much heavier custom mechanism in a weight-sensitive design. Whether you're working on an aerial platform, a mobile robot, or another compact automated system, the same power-to-weight tradeoffs that shaped this excavator's design apply broadly to any project where every gram counts.
If you need assistance selecting a product that meets your application requirements, please contact our sales team for assistance.
