Professional Research Projects
From ChungWiki
Still under construction but check back soon!
Research Projects
I am interested in information-processing methods for intelligent systems, which is a relevant research area for a variety of interesting applications.
Distributed Sensing
One can ask how a team of sensor agents can coordinate their individual motions to improve the quality of their sensing performance.
(back to top)
Multi-Robot Sensor Testbed
In order to validate the proposed theoretical methodologies as well as to provide insight for practical implementations and applications, we've developed a multi-robot, multi-sensor, experimental testbed at Caltech. The testbed comprises multiple ER1 robots from Evolution Robotics equipped with an overhead, indoor localization system, odometry measurements, and onboard LADAR (laser-scanning) units.
(back to top)
Distributed Shape Estimation
(Collaboration with Jeremy Ma, Caltech)
A primary motivation for employing teams of sensing agents is the ability to sense entities that are intrinsically distributed, such as a dynamically changing environment or process which cannot be easily observed by a single sensor. One such example is that of a geometric object, whose shape parameters are subject to observation uncertainty.
Mobility can be used to enhance the information gathered by the sensing agents, whether it be estimates of vertex locations or of line parameters (e.g. from the Hough transform).
We propose a model of line uncertainty which lends itself to the construction of an objective function whose minimization incorporates a reduction of local observation uncertainty along with an increase in visibility of shape edges.
(back to top)
Observability-based Motion Planning
(Collaboration with Domitilla Del Vecchio, U. Michigan)
(back to top)
Probabilistic Search for Robotic Systems
|
Various methods exist for search. Our mechanisms enable us to examine various applications, including behavior modeling of natural systems (such as foraging and human visual search). |
|
(back to top)
Modeling of Human Visual Search
(back to top)
Modeling of the Escape Response in Drosophila
(Collaboration with Gwyneth Card, Caltech)
Information-processing is relevant to natural systems, such as in <emph>Drosophila melanogaster</emph>, in executing actions and decisions. In the context of an escape response, stationary (i.e. not in flight) fruit flies respond to visual looming stimuli by a reflexive, open-loop jumping maneuver.
|
|
|
In trying to understand the relationship between the dynamics of the (apparent) approaching object and the gathering of visual information, we can begin to investigate how the integration of these data provides a decision signal for executing a jump response.
(back to top)
Modeling of Visually-Mediated Search in Drosophila
(Collaboration with Alice Robie, Caltech)
Drosophila melanogaster (a.k.a. fruit flies) exhibit very interest searching behaviors when in search of food.
|
|
|
(back to top)
Pursuit-Evasion: Reasoning about Levels of Rationality
(Collaboration with Angelo Polydoro, FGV)
How many times have you tried to out-guess an opponent in a friendly game of paper-rock-scissors?
(back to top)
