Aerial robotics · Mechatronics · System integration

Pressure-Washing Drone.

An ongoing aerial cleaning platform integrating a 5 kg MTOW multirotor, pressure-washing hardware, tethered fluid delivery, Raspberry Pi computing, Pixhawk flight control, camera and LiDAR sensing, and remotely controlled washer actuation.

Role Mechatronics and multisensor integration

Platform X650 multirotor · Pixhawk 2.4.8

Research Creative Machines Lab · Columbia

Status Prototype testing and sensing development

Pressure-washing drone system concept with onboard washer and external water supply — System visualization showing the intended aerial cleaning configuration, including the drone-mounted Ryobi washer and external water supply.

Drone propulsion and electronics integration during bench testing — Propulsion and electronics integration during motor bring-up and controlled bench testing.

Assembled multirotor prototype — Integrated multirotor prepared for flight-system calibration and controlled testing.

01 / FLY

Flight Platform

Multirotor propulsion, Pixhawk flight control, radio control, ESCs, and vehicle power.

→

02 / CLEAN

Washer Payload

Ryobi washer, trigger actuation, fluid delivery, siphon line, and transfer hardware.

→

03 / COMPUTE

Raspberry Pi

MAVLink telemetry, sensor logging, perception, and future autonomy.

→

04 / SENSE

Camera + LiDAR

Wall-distance estimation and facade-relative perception.

01 / System

A flying robot built around propulsion, fluid delivery, sensing, and control.

Cleaning a building facade requires more than mounting a pressure washer beneath a drone. The platform must remain within a strict mass budget, deliver water and power safely, maintain stable flight near a wall, activate the washer remotely, and eventually estimate its position relative to the facade.

The system therefore combines flight hardware, a cleaning payload, onboard computing, communication interfaces, and environmental sensing into one tightly constrained mechatronic platform.

Flight X650 multirotor + Pixhawk

Cleaning Ryobi 18 V washer

Compute Raspberry Pi 4

Perception Camera + LiDAR

02 / My Role

Mechatronics integration across flight, payload, sensing, and testing.

My work focused on bringing the physical and computational subsystems together into a testable robotic platform.

Integrate

Connect flight, compute, payload, and sensing.

Integrated the Pixhawk flight controller, Raspberry Pi, camera and LiDAR sensing, propulsion components, payload electronics, and washer actuation hardware.

Evaluate

Compare payload architectures.

Evaluated onboard and tethered pressure-washing architectures against the platform's 5 kg maximum takeoff mass.

Characterize

Measure the washer power path.

Characterized the Ryobi washer startup demand, including the approximately 25 A surge and the limitations of the available bench supply.

Debug

Isolate failures before flight testing.

Debugged camera and LiDAR alignment, reflective-surface sensing, serial communication, electrical interfaces, and integrated subsystem behavior.

03 / Payload Architectures

Compare mass, control, and tethering tradeoffs before flight.

Two pressure-washing configurations were evaluated. One places the cordless Ryobi washer onboard the aircraft and draws water through a lighter siphon line. The second keeps the pressure washer on the ground and sends pressurized water to the aircraft through a heavier tether.

Configuration A

Onboard washer

4.52 kg

Places the Ryobi washer on the aircraft and uses a siphon line from the ground. This reduces high-pressure hose mass but adds washer and battery mass directly to the drone.

Configuration B

Ground washer

4.05 kg

Keeps the pressure washer on the ground and uses a high-pressure tether. The aircraft remains lighter, but hose mass and tether forces increase as operating height grows.

04 / Power and Actuation

Manage startup current without losing remote control.

The 18 V Ryobi washer produced an approximately 25 A startup demand, exceeding the available 12.55 A bench-supply limit.

This required separating the washer's high-current power path from its low-power control interface and selecting switching hardware with sufficient current margin.

Measure

Characterize startup current.

Measured washer startup behavior rather than sizing the electrical interface from nominal current alone.

Switch

Separate control from high-current power.

Developed a relay-based control interface so the washer could be commanded without routing its high-current path through the flight computer.

Actuate

Remotely operate the trigger.

Used a servo-operated mechanism to pull the washer trigger while keeping flight and cleaning commands independent.

05 / Fluid and Mass Analysis

Every additional foot of hose changes the flight problem.

Fluid-line selection considered pressure loss, water mass, hose mass, operating height, and the resulting load on the aircraft.

A 1/4-inch fluid line provides a lightweight option for the onboard washer architecture, while the ground-washer configuration requires substantially heavier high-pressure hose.

Lightweight line 1/4 inch

Estimated loss ≈ 0.576 PSI / ft

Water mass ≈ 16.33 g / ft

High-pressure hose ≈ 109 g / ft

06 / Prototype Testing

Validate subsystems before combining them in flight.

Testing progressed from individual electrical and mechanical interfaces to propulsion bring-up and controlled indoor flight.

The first indoor lift test confirmed sufficient thrust for the assembled platform. Payload flight, outdoor operation, and autonomous cleaning remain under development.

Bench

Electrical integration.

Checked power distribution, receiver behavior, actuator interfaces, flight-controller communication, and washer control.

Propulsion

Motor bring-up.

Verified ESC calibration, motor response, command direction, and propulsion-system behavior.

Flight

Indoor lift test.

Completed an initial controlled flight to verify adequate lift before advancing to payload testing.

07 / Raspberry Pi and Sensing

Build a perception layer above the flight controller.

A Raspberry Pi 4 provides the onboard computing layer for camera and LiDAR processing.

MAVLink communication with the Pixhawk was brought up over USB serial at 115200 baud, with heartbeat and attitude data used to verify the communication path and telemetry flow.

Flight control Pixhawk 2.4.8

Companion compute Raspberry Pi 4

Communication MAVLink · 115200 baud

Perception Camera + LiDAR

08 / Current Work

Move from integrated hardware toward sensor-aware flight.

The project is ongoing. Current work focuses on reliable sensing, synchronized data collection, calibration, and establishing the state estimates needed for controlled operation near a building facade.

Now

Sensor bring-up and calibration.

Camera and LiDAR bring-up, extrinsic alignment, reflective-surface sensing investigation, and reliable timestamped data collection.

Facade-relative state estimation.

Fuse camera, LiDAR, and flight-state data for wall-distance estimation and closed-loop positioning.

Later

Autonomous cleaning behavior.

Develop recoil and tether-disturbance compensation, standoff control, and cleaning-coverage perception.

09 / Technologies

Hardware, communication, sensing, and system integration.

Back to all projects ↗