Surface Profiling of Drywalls for

Automated Sanding

Dony Alex Dr. Mohamed Al-Hussein Dr. Saeed Behzadipour

Hole School of Construction Engineering

Dept. of Civil & Environmental Engineering

University of Alberta

Canada

June 29th, 2011

Outline

●Introduction

●Surface Profiling Techniques

●Proposed Methodology

●Algorithms

●Case Study

●Conclusion

Current Challenges in Construction

●Declining Productivity

●Hazardous Working Conditions

●Shortage of Skilled Workers

●10 -15% increase in overall productivity

●Reduce exposure to hazardous environment

●Better quality

●Higher standards of work

How Automation Helps?

Challenges in Construction

sources: Euroconstruct, Eurostat, ACEA)

The Sanding Process

●Ensures smooth surface

●Exposure to dust - at least 10

times the Permissible Exposure

Limits (PEL)

●Integration of robotics - a relatively

new concept

Tape Coat Drywall

Tape

Block

Coat

Skim

Coat

Sanding

Image: http://home.howstuffworks.com/drywall4.htm

Surface Profiling

●Process of identifying the surface

geometry.

●Identifies irregularities

●Currently performed manually.

Research Objectives

Surface Profiling Techniques

Classification

●Contact Based

●Non-contact Based

Sensor Accuracy Cost Complexity

Stylus High Medium High

Ultrasonic High High Low

Shadow

Profilometry

Medium Low Low

Capacitance Medium Low High

Shadow Profilometry

●Technique of tracing a surface profile using shadows

●When a plane of light is made to intersect with an irregular surface at

an angle, the resultant intersection line follows the topography of the

surface.

Shadow scanner for evaluating surface smoothness in wood industry (Sandak and Tanaka, 2005).

Proposed Methodology

Proposed Methodology – Test Setup

Test Setup

oVirtual test environment setup in 3DS Max

oComprises of surface (1) , curtain (2), light source (3) and

camera (4)

oCaptures image of the shadow profile over the surface cross

section

oCaptures location at which shadow edge is formed

Base Simulation model

1. Flat surface

2. Depression in surface

3. Elevated surface

4. Nail Hole

Proposed Methodology – Image Processing

Case Study

H

D1

D1 = Distance at which Shadow edge is formed.

H = Height of curtain from the surface.

H= 6cm

Angle of Incidence (α)= 450

D

H

Image Resolution = 848 x 480

pixels

Image Pre Processing

+

Edge Detection

Accuracy of the reconstructed profile

Maximum Error = 0.1 cm

Average Error = 0.01 cm

Case Study 2 – Nail Hole

Maximum Error = 0.3 cm

Average Error = 0.02 cm

Accuracy of the reconstructed profile

Case 3 – Curved Elevation

Maximum Error = 0.1 cm

Average Error = 0.04 cm

Accuracy of the reconstructed profile

Conclusions - Contributions

●Introduction of shadow profilometry as a method of profiling

the drywall surface

●Successful simulation of shadow profilometry for surface

profiling

●Successful 3D reconstruction of the surface

Conclusion – Limitations and Future Scope

● The research treats the drywall as a single

surface

●Accuracy greatly depends on sharpness of the

shadow

●Experimental implementation and validation

●Robot task planning based on the surface profile

●Integration of sensor into a robotic arm

Limitations

Future Scope