|
Close Details
|
|
2012 - 11th Annual NC State Summer Undergraduate Research Symposium
|
|
Session Time :
8/1/12 3:00 PM - 8/1/12 4:14 PM
|
|
Content Area : OIA-SRE, Office of International Affairs - Summer Research Experience
|
Lead Student Presenters : Wanxi Liu
|
Abstract Title : Path Planning for Object Recognition
|
Abstract :
The object recognition problem in computer vision has several applications in security and environmental sciences to name a few. For example, one may be interested in identifying objects that can be associated with a potential threat in an airport. In this context, we aim to minimize the traveled distance of a mobile robot looking to recognize multiple targets. Our objective is optimal path planning of a mobile robot's motion for the recognition of multiple targets under the assumptions that (1) objects have circle shape and can be recognized from a certain orientation, and (2) all objects and the robot can be localized. We proceed by building a mathematical modal for the problem, finding the best strategy by geometrical derivation, analyzing the result via numerical simulations in MATLAB, and testing the result in a clearpath-chameleon robot. We use an Optitrack Motion Tracking System to localize the robot and the objects. The implementation is done using the Robot Operating System. For single object, both the simulation and the derivation show that the optimal strategy is to let the robot go directly towards the object to enter the possible recognition area and circle around the object until it is recognized. The tangential point also turns out to be a good choice when the robot is close enough to the object and the angle of recognizing the object is relatively large. The result will be extended to multiple objects by inserting intermediate target points set as end or starting points in single object situation.
|
Mentor and/or Co-Author : Edgar J Lobaton
|