14-06-2012, 04:40 PM
Real-Time Obstacle Avoidance by an Autonomous
Mobile Robot using an Active Vision Sensor
[attachment=24473]
Abstract
This article presents a real-time vision-based obstacle detection and
avoidance method in an indoor environment for an autonomous mobile robot.
In this study, we propose a scenario in which 3-Dimensional obstacles mainly
building our real world environment, are detected by scanning a vertically emitted
laser slit paired with an active stereo vision system.
Introduction
For an autonomous mobile robot performing a navigation-based task in a vague
environment, to detect and to avoid encountered obstacles is an important issue and
a key function for the robot body safety as well as for the task continuity. Obstacle
detection and avoidance in a real world environment - that appears so easy to humans
- is a rather difficult task for autonomous mobile robots and is still a well-researched
topic in robotics.
Obstacle detection
Principles
In this study, we use a laser range sensor to detect obstacles. A laser range sensor
consists of coupling a vision sensor with an -usually- horizontally emitted laser slit.
It enables an intelligent robot to get the distance to any obstacle present in its 2-
Dimensional vicinity, thanks to a simple and speedy image processing[12]. However,
emitting horizontally the laser slit does not procure the height information of obstacles
such as tables, chairs and many others present in an indoor environment. In this work,
in order to obtain the height information of any object obstructing the robot path,
regardless of its shape, we emit our laser slit vertically as illustrated in figure 1.
Image processing
Previously we showed that any possible obstacle standing in the robot’s front environment
would show upas a vertical segment in the left region of the image captured
by our sensor(refer to figure 2(b)). Here, we outline that the use of an optical filter
to filtrate only the laser slit and to cut all visible lights enables our sensor to capture
nearly binarized images, which consist of an uniformly black textured background with
a white textured vertical segment, in case of obstacle presence. Doing so simplifies
greatly the image processing to be conducted in this study in matter of difficulty, and
mostly helps our seek for a real-time obstacle detection.
Orienting dynamically the active vision sensor on the robot’s path.
Once the mobile robot equipped with such sensor system, is navigating a defined
path, particularly a circular one, the equipped sensor’s orientation on the path becomes
a matter of great importance here, since the robot must anticipatorily detect
and avoid obstacles on the path. In this study, such navigation based-problem is solved
by controlling the field scanned by our sensor.
Mobile Robot using an Active Vision Sensor
[attachment=24473]
Abstract
This article presents a real-time vision-based obstacle detection and
avoidance method in an indoor environment for an autonomous mobile robot.
In this study, we propose a scenario in which 3-Dimensional obstacles mainly
building our real world environment, are detected by scanning a vertically emitted
laser slit paired with an active stereo vision system.
Introduction
For an autonomous mobile robot performing a navigation-based task in a vague
environment, to detect and to avoid encountered obstacles is an important issue and
a key function for the robot body safety as well as for the task continuity. Obstacle
detection and avoidance in a real world environment - that appears so easy to humans
- is a rather difficult task for autonomous mobile robots and is still a well-researched
topic in robotics.
Obstacle detection
Principles
In this study, we use a laser range sensor to detect obstacles. A laser range sensor
consists of coupling a vision sensor with an -usually- horizontally emitted laser slit.
It enables an intelligent robot to get the distance to any obstacle present in its 2-
Dimensional vicinity, thanks to a simple and speedy image processing[12]. However,
emitting horizontally the laser slit does not procure the height information of obstacles
such as tables, chairs and many others present in an indoor environment. In this work,
in order to obtain the height information of any object obstructing the robot path,
regardless of its shape, we emit our laser slit vertically as illustrated in figure 1.
Image processing
Previously we showed that any possible obstacle standing in the robot’s front environment
would show upas a vertical segment in the left region of the image captured
by our sensor(refer to figure 2(b)). Here, we outline that the use of an optical filter
to filtrate only the laser slit and to cut all visible lights enables our sensor to capture
nearly binarized images, which consist of an uniformly black textured background with
a white textured vertical segment, in case of obstacle presence. Doing so simplifies
greatly the image processing to be conducted in this study in matter of difficulty, and
mostly helps our seek for a real-time obstacle detection.
Orienting dynamically the active vision sensor on the robot’s path.
Once the mobile robot equipped with such sensor system, is navigating a defined
path, particularly a circular one, the equipped sensor’s orientation on the path becomes
a matter of great importance here, since the robot must anticipatorily detect
and avoid obstacles on the path. In this study, such navigation based-problem is solved
by controlling the field scanned by our sensor.