EVEN, Philippe
SEMI-AUTOMATED EDGE SEGMENT SPECIFICATION
FOR AN INTERACTIVE MODELLING SYSTEM OF ROBOT ENVIRONMENTS
Philippe EVEN, Anne MALAVAUD
Atomic Energy Commission, Teleoperation and Robotics Department,
(CEA/STR), PO Box 6, F-92265 Fontenay-Aux-Roses, France,
Philippe.Even@cea.fr, Anne.Malavaud@cea.fr
KEY WORDS: Modelling, Semi-automation, User interfaces, On-line, Robots.
ABSTRACT
Interactive methods are well suited to telerobotics purposes. Based on the super-imposition of the model on video images,
they provide a friendly way to acquire or update the environment model from an on-board CCD camera. It is a flexible
way to cope with a priori model inaccuracies and any possible mission incidents. Modelling time is a key requirement
which can be fulfilled through the integration of semi-automatic assistances. This paper deals with the semi-automation
of edge segment specification tasks. Edge segments combinations define 3-D lines or planes, which are used to orient the
modelled objects. Their accurate specification is a tiring and time consuming task, which can not be fully automated. In
the semi-automatic mode we set up, the operator quickly draws a line over the image. This line is automatically attracted
towards the nearest extracted edge contour. The automatic attraction function is based on the Hough transform. This semi-
automatic assistance has been integrated into the Pyramide interactive 3-D modelling system, which has been developed
at CEA/STR. Evaluations on realistic sites showed its high flexibility and efficiency.
1 INTRODUCTION
Within the last decade, CEA/STR has developed an interactive system, called Pyramide, to acquire a 3-D model of a
teleoperated robot environment using video images provided by an embedded CCD camera (Even and Marcé, 1988). Its
principle consists in superimposing solid primitives on the images, and interactively adjusting their position and para-
meters until they match the objects contours. It has been designed for on-line applications to provide computer graphics
assistances during robotics missions. Fast acquisition and user-friendly operating mode are important requirements. In
order to fulfil these requirements, several assistances based on image processing techniques or structural and functional
knowledge have been integrated. A major one is a semi-automated edge segment specification assistance, which brought
noticeable improvements.
Pyramide provides a generic modelling module aiming at modelling any structured environment as an assembly of basic
volumes (block, cylinder, cone, ...). 3-D features (plane, line, point, orientation) are determined from image features (edge
segments or points) or already defined objects (planes, edges, vertices), and displayed as manipulation frames. They are
then used to position the selected primitives in the 3-D space. For instance, the selection of the same contour on two
oriented images defines a 3-D line used to constrain the position of one of the object edges. In order to speed up that
process, the structural or functional knowledge on some particular environment is exploited inside specialised modules
featuring dedicated primitives, constraints and modelling methods. In the case of a piping module, a straight pipe is
modelled by specifying its contours in two images or more. So edge segment specification is often required as well in the
generic module as in specialised ones.
Careful matching of 2-D segments on image edges in manual mode is long and visually tiring. On the over hand, its full
automation is often doomed to failure, because of poor visibility, bad scene illumination, numerous reflects on stainless
steel surfaces, frequent occlusions, ... Therefore a semi-automated mode has been implemented. Each time the operator
draws a segment over an image, a local edge detection function is automatically executed, and attracts the segment on the
nearest edge found in the image. Hence only a coarse specification is required from the operator. Of course the segment
may be attracted towards some disturbing edge close to the desired one. The operator then selects one of the segment
ends and drags it towards the relevant edge. New edge points are thus extracted on a different image area. This is the only
way to act on the automatic function. In case of repeated failures, the operator can come back to the manually specified
segment with a simple undoing action. When it happens too often, the automatic function can be disabled.
The edge detection function consists in extracting edge points inside a rectangular image area around the selected segment.
For each point, possible edges are accumulated inside a Hough table. A vote is then performed to provide the best
candidate. The sides of the detection area are parallel to the image borders. The distance between the initial segment and
222 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B5. Amsterdam 2000.
the @
shot
initi
desi
This
size
time
But
spec
mod
perf
The
sele
date
to b
CE,
met
cali
alsc