project specification

PaintCopter

A system for autonomous spray painting using a UAV, suitable for industrial applications. The PaintCopter is a quadrotor that has been custom fitted with an arm plus a spray gun on a pan-tilt mechanism. To enable long deployment times for industrial painting tasks, power and paint are delivered by lines from an external uni

Specifications

Weighttotal 4200g
arm 140 g
UAVDJI Matrice 100
Servo (PTU)2 Dynamixel 3 AX-12A
Computers1 Intel UP board
1 Jetson TX2 board
Sensor rig1 S360 stereo camera
1 ADIS IMU
1 Intel RealSense R200 depth camera

Overview

The PaintCopter is a quadrotor that has been custom fitted with an arm plus a spray gun on a pan-tilt mechanism. To enable long deployment times for industrial painting tasks, power and paint are delivered by lines from an external unit.

This project enables sophisticated  operations that subsumes the basic tasks, including painting on 3D structures, and  painting of a desired texture appearance.  

The system operation consists of an offline component to capture a 3D model of the target surface, an offline component to design the painted surface appearance, and generate the associated robotic painting commands, and a live system that carries out the spray painting. 

The UAV hardware consists of a DJI Matrice 100 that is augmented with a custom arm and a spray gun mounted on a pan-tilt unit  (PTU).  The arm consists of three carbon fiber tubes in a triangular configuration, with 3D printed aluminum mounts. The PTU has two Dynamixel 3AX-12A servo motors to allow yaw ±90◦ and pitch ±45◦ control of the spray gun. An additional servo actuates the spray release, with controllable aperture between closed and open positions. A spring is placed at the end of the arm beyond the spray gun, to prevent damage in case of accidental contact with the  painting  surface.

References

Describes related work. Describes design and hardware, covers software architecture, control of the UAV and spray gun, and describes the full painting pipeline as a multi-stage mission. Provides experimental results.

A. Vempati, M. Kamel, N. Stilinovic, et al. - June 2018.

Wevolver 2022