Can anyone give me an idea about moving a robots by using image processing concept? It should reach the destination but it won't be affected by the obstacles?
This is a good question.
A complete answer to this question is given in
N. Keeratipranon, Robot Navigation in Sensor Space, Ph.D. thesis, Queensland University of Technology, 2009:
http://eprints.qut.edu.au/29316/1/Narondech_Keeratipranon_Thesis.pdf
See Section 1.1, starting on page 1, for overview of approaches to map-based and map-free robot navigation. An intensive review of all known approaches to robot navigation is given in Chapter 2, especially Section 2.2, starting on page 20. A very interesting part of this thesis is Section 2.6, starting on page 52 (Dimensionality reduction algorithms). Dimensionality reduction comes into play based on the need for more concise, lower dimensional representations of raw, complex high-dimensional signals. This means that manifold learning needs to be considered to make robot navigation efficient with fast responses to various situations encountered by a wandering (exploring) mobile robot. The introduction to approaches to manifold learning is both interesting and well-done.
The focus of this thesis is on navigation in a sensor space, starting on page 61, Chapter 3. This is followed by a complete overview of map learning in a sensor space, starting on page 135.
Navigating by image processing should done in adaptive phase due to the factor of time.
Here is a paper that will give you few hints
http://www.fujipress.jp/finder/xslt.php?mode=present&inputfile=JACII001300030020.xml
The paper introduces adaptive controller with learning and memory for mobile robots in complex, highly changeable environments. The model’s novelty lies in 1) simplifying complex environments into simple patterns and 2) training, storing, and recalling individual patterns independently through a dynamic memory.
http://www.fujipress.jp/finder/xslt.php?mode=present&inputfile=JACII001300030020.xml
There are a couple of robot software frameworks and operating systems for mobile robots which come with modules for navigation and collision avoidance, e.g. ROS.
You may also look at the robot simulation tools like VREP or MORSE, which also have path planning algorithms implemented
http://www.ros.org
http://www.coppeliarobotics.com/index.html
https://www.openrobots.org/wiki/morse/
I think using image processing for path planing is SLAM. I suggest you to try different types of SLAM. The below links are good start for them.
https://www.openslam.org/
https://www.google.com/url?sa=t&rct=j&q=&esrc=s&source=web&cd=2&cad=rja&uact=8&ved=0CCgQFjAB&url=http%3A%2F%2Focw.mit.edu%2Fcourses%2Faeronautics-and-astronautics%2F16-412j-cognitive-robotics-spring-2005%2Fprojects%2F1aslam_blas_repo.pdf&ei=VrndVJqoJYXvUMOtgeAJ&usg=AFQjCNFjGNTRFcyp6cLcD-EB8cdTsbSbYQ&sig2=5gFKRH-fK50iCX2p4JF2Og&bvm=bv.85970519,d.d24
Hi, This is a good question you should go with SLAM based navigation and some basic working knowledge of mobile navigations.
http://rkala.in/publications.php
- Badges
- Science topic
- Similar topics
- Social Psychology
- Attitude