Home

About

Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network
FAQHELPSITE MAPCONTACT US


  Advanced Search  

 
The Center for Control, Dynamical Systems, and Computation University of California at Santa Barbara
 

Summary: The Center for Control, Dynamical Systems, and Computation
University of California at Santa Barbara
Presents
Stable and Efficient Tracking of Multiple Dynamic
Obstacles Under Large Viewpoint Changes
Isaac Miller
(Faculty Candidate)
Tuesday, March 31, 2009 9:30 am HFH 4164
Abstract:
Despite the successful application of target tracking algorithms in long-range and even high-dynamics tracking scenarios, field robots
tasked with the similar goal of detecting and tracking moving obstacles still rely on ad hoc feature extraction algorithms and error-
prone best-fit matching schemes to track obstacles through multiple frames of sensor data. Though such schemes enable real-time
processing on field robots with limited computational budgets, they often introduce significant biases and estimation artifacts through
lossy feature extraction steps taken at the level of raw sensor data. Unfortunately, the estimation artifacts induced by these steps
are most catastrophic in scenarios of critical interest to robotic path planning: when obstacles occupy a significant portion of, or move
rapidly across, the robot's field of view. The sensor noise that lies at the root of these estimation artifacts suggests the application of
Bayesian techniques to eliminate ad hoc feature extraction in favor of robust, probabilistic target tracking. To that end, this talk presents
the LocalMap tracking algorithm: a computationally feasible, real-time solution to the joint estimation problem of data assignment and
dynamic obstacle tracking from a potentially moving robotic platform. The algorithm utilizes a Bayesian factorization to separate the
joint estimation problem into 1) a data assignment problem solved via particle filter, and 2) a multiple dynamic obstacle tracking problem

  

Source: Akhmedov, Azer - Department of Mathematics, University of California at Santa Barbara

 

Collections: Mathematics