The seminar will be held in IS Lecture Hall on Friday, March 2, 15:00-16:30

Han-Lim Choi

Division of Aerospace Engineering, KAIST

Email: hanlimc@kaist.ac.kr

http://lics.kaist.ac.kr

This seminar presents high-level planning approaches for networked robotic agents when the mission objective is to extract information in a dynamic uncertain environment. The first part of the talk discusses methodologies to perform robust distributed task planning for a heterogeneous team of agents performing cooperative missions. We present the consensus-based bundle algorithm (CBBA) which is a decentralized cooperative iterative auction algorithm for assigning tasks to agents. CBBA uses two phases to achieve a conflict-free task assignment. A key feature of CBBA is that its consensus protocol aims at agreement on the winning bids and corresponding winning agents (i.e., consensus in the spaces of decision variables and objective function). This enables CBBA to create conflict-free solutions that are relatively robust to inconsistencies in the current situational awareness. Recent extensions to handle more realistic multi-UAV operational complications will also be reported. 

The second part of the talk addresses informative forecasting using mobile sensor networks. The algorithms plan maneuvers for a set of mobile observing platforms (e.g., Unmanned Aerial Vehicles) to extract the maximum information from a large-scale dynamic environment by automatically predicting when and where measurements are needed. The goal of planning is to reduce the uncertainty in the environmental quantities of interest in the far future. The primary challenge in this problem is the significant computational complexity that is incurred when planning informative flight paths, as a result of the large size of the decision space and the cost of propagating the influence of sensing into the future. This work presents a new set of methodologies that correctly and efficiently quantify the value of information in large information spaces, thus leading to a systematic architecture for planning information-gathering paths for mobile sensors in a dynamic environment.

Short Bio: Dr. Han-Lim Choi is an Assistant Professor of Aerospace Engineering at KAIST (Korea Advanced Institute of Science and Technology).  He received his B.S. and M.S. degrees in Aerospace Engineering from KAIST, Daejeon, Korea, in 2000 and 2002, respectively, and his PhD degree in Aeronautics and Astronautics from Massachusetts Institute of Technology (MIT), Cambridge, MA, USA, in 2008. He then studied for one and a half years at MIT as a postdoctoral associate until he has joined KAIST in 2010. His current research interests include estimation and control for sensor networks and decision making for multi-gent systems. He (together with Dr. Jonathan P. How) is the recipient of Automatica Applications Prize in 2011.