Difference between revisions of "Cognitive SLAM"

From AIRWiki
Jump to: navigation, search
 
(3 intermediate revisions by the same user not shown)
Line 1: Line 1:
 
{{ProjectProposal
 
{{ProjectProposal
 
|title=Cognitive SLAM
 
|title=Cognitive SLAM
|description=
+
|description= We have developed a system that is able to detect, recognize and track objects in an image taken from a low cost robot equipped with a IMU and a low cost camera. The system is capable to detect and recognize objects using a user defined fuzzy tree classifier. However the system performance is heavily dependent on high level feature extraction, such as geometric features. The problem is non trivial due to noisy low cost camera and changes in the light conditions. The aim of this project is to improve the feature extraction and description process, both in performance and quality, possible adding a more complete description or others type of features. The long term aim of the research is to have an autonomuos robot capable to create a semantic map of the envirorment, localize himself , make inference on the map, navigate into the envirorment using the objects as landmarks.
 +
No special skills are required, except basic c and object oriented programming.
 
|tutor=AndreaBonarini; DavideTateo
 
|tutor=AndreaBonarini; DavideTateo
 +
|image=C_SLAM_Recognition2.png
 
|start=1/1/2015
 
|start=1/1/2015
 
|studmin=1
 
|studmin=1
 +
|studmax=2
 
|cfumin=5
 
|cfumin=5
 
|cfumax=20
 
|cfumax=20
|resarea=Computer Vision and Image Analysis
+
|resarea=Robotics
 
|restopic=SLAM; Feature Extraction
 
|restopic=SLAM; Feature Extraction
 
|level=Ms
 
|level=Ms

Latest revision as of 01:10, 18 December 2014

Title: Cognitive SLAM
C SLAM Recognition2.png

Image:C_SLAM_Recognition2.png

Description: We have developed a system that is able to detect, recognize and track objects in an image taken from a low cost robot equipped with a IMU and a low cost camera. The system is capable to detect and recognize objects using a user defined fuzzy tree classifier. However the system performance is heavily dependent on high level feature extraction, such as geometric features. The problem is non trivial due to noisy low cost camera and changes in the light conditions. The aim of this project is to improve the feature extraction and description process, both in performance and quality, possible adding a more complete description or others type of features. The long term aim of the research is to have an autonomuos robot capable to create a semantic map of the envirorment, localize himself , make inference on the map, navigate into the envirorment using the objects as landmarks.

No special skills are required, except basic c and object oriented programming.

Tutor: AndreaBonarini (andrea.bonarini@polimi.it), DavideTateo (davide.tateo@polimi.it)
Start: 1/1/2015
Students: 1 - 2
CFU: 5 - 20
Research Area: Robotics
Research Topic: SLAM, Feature Extraction
Level: Ms
Type: Course, Thesis
Status: Active