ECE Seminar Series Fall 2014

Thursday October 23^st, 1-2 PM, ITEB 336

Towards a formal design of distributed cooperative systems

 Hai Lin

Assistant Professor, Electrical Engineering, University of Notre Dame

 

 Abstract: A common challenge in our future engineered system design, such as power grids, intelligent transportation networks and flexible-manufacturing systems, is how to make a large number of distributed systems work together in a reliable and efficient manner. Existing methods are either only suitable for small scale systematic synthesis, oversimplifying the nodal dynamics, or fail to adapt to changing environments. This motivates our research aiming at a scalable, correct-by-construction formal design methodology for distributed cooperative systems. In particular, we focus on a formal design of multi-robot systems that can guarantee the accomplishment of high-level team missions through automatic synthesis of local coordination mechanisms and control laws. The basic idea is to decompose the team mission into individual subtasks such that the design can be reduced to local synthesis problems for individual robots. Multidisciplinary approaches combining hybrid systems, supervisory control, inference deduction and model checking are utilized to achieve this goal. The developed theory will enable robots in the team to cooperatively learn their individual roles in a mission, and then automatically synthesize local supervisors to fulfill their subtasks. A salient feature of the proposed method lies on its ability to handle environmental uncertainties and un-modeled dynamics as we do not require an explicit model of the transition dynamics of each agent and their interactions with the environment. In addition, the design is on-line and reactive enabling the robot team to adapt to changing environments and dynamic tasking.

 

Short Bio: Hai Lin obtained his B.S. degree at the University of Science and Technology Beijing and his M.S. degree from the Chinese Academy of Sciences in 1997 and 2000 respectively. In 2005, he received his Ph.D. degree from the University of Notre Dame. Dr. Lin is currently an Assistant Professor at the Department of Electrical Engineering, University of Notre Dame. Before returning to his alma mater, Hai has been working as an assistant professor in the National University of Singapore from 2006 to 2011. Dr. Lin’s teaching and research interests are in the multidisciplinary study of the problems at the intersections of control, communication, computation and life sciences. His current research thrust is on cyber-physical systems, multi-robot cooperative tasking, systems biology and hybrid control. Hai has been served in several committees and editorial board. He is the Program Chair for IEEE ICCA 2011, IEEE CIS 2011 and the Chair for IEEE Systems, Man and Cybernetics Singapore Chapter for 2009 and 2010. He is a recipient of 2013 NSF CAREER award and a senior member of IEEE.

 

 

Host: Peng Zhang, peng@engr.uconn.edu

Special ECE/CSE Seminar

Tuesday October 21^st 11 AM – 12 PM, ITEB 336

Challenges of Human-in-the-Loop Planning & Decision Support

Subbarao Kambhampati

Arizona State University

Abstract: Endowing  an automated agent with  the ability to “plan” — i.e., convert its high-level goals into an executable course of action — has been a long-standing quest in Artificial Intelligence.  For much of the history of automated planning, the dominant research theme has been efficient synthesis of plans under increasingly expressive system dynamics (classical, temporal, stochastic etc.).

 

An implicit assumption underlying  this research has been that the planner’s responsibilities start with taking a complete specification, and end with giving out a complete course of action. This assumption is no longer valid when humans are part of the decision making loop, as is the case in an increasing number of decision support and human-machine teaming scenarios.

In this talk I will identify the research challenges in human-in-the-loop planning, including the need to interpret the goals/intentions of the humans in the loop, the need to support continual planning and replanning, the need to unobtrusively support team-decision making, and above all the need to do handle pervasive incompleteness in the domain models as well as problem specification.  I will then describe some of our ongoing work in handling these challenges in the context of human-robot teaming and crowd-sourced planning.

 

Short Bio: Dr. Subbarao Kambhampati is a professor of Computer Science at Arizona State University, where he leads the Yochan research group focusing on the challenges in  automated planning and decision support, as well as information integration from structured and unstructured data sources.  He is a 1994 NSF Young Investigator, a 2004 IBM faculty fellow and thrice received Google Research Awards. He was elected fellow of AAAI (in 2004) for his contributions to automated planning. He received the 2002 college of engineering teaching excellence award, 2011 university last lecture invitation, and 2012 departmental best teacher award. Kambhampati was the co-chair of AAAI 2005, and will be the program chair of IJCAI 2016. He is an elected Trustee of IJCAI, and the president-elect of AAAI.

 

Host: Krishna Pattipati, krishna@engr.uconn.edu

ECE Seminar Series Fall 2014

Thursday October 2^nd, 1-2 PM, ITEB 336

 Time varying and sparse underwater acoustic response estimation via a hierarchical Gaussian mixture model with application to M-ary orthogonal spread spectrum signaling

Paul J. Gendron

University of Massachusetts Dartmouth

Abstract: Recent advances in modeling doubly spread propagation channels have provided solutions for diverse communication applications between mobile platforms and these advances have been extended and have had a positive impact on the particularly challenging underwater acoustic environment. In this talk a hierarchical Gaussian mixture model is proposed to characterize shallow water acoustic response functions that are time-varying and sparse. A particular ocean environment between source and receiver is predicated on a proportion of relatively coherent paths that possess an ensemble frequency-Doppler-beam spectra. Conditioned on the bulk platform speed and ensemble Doppler spread a structured field of Beta variates link the Doppler profile to the probabilities of indicator variables specifying the state of ensonification across channel frequency, Doppler and beam. Conditioned on these indicator variables the amplitude and phase of a particular frequency and Doppler slot is modeled as complex Gaussian. The remaining non-coherent multiple surface scattered paths exhibit a spectrally flat Doppler profile. The hierarchical model is flexible and naturally accommodates diverse platform motion scenarios and array orientations. Accurate estimation of the time-varying acoustic response for the full duration of the broadband transmission facilitates compensation of the bulk time-varying dilation process taking advantage of the correlated coherent paths. The model ameliorates coherence degradation and enhance coherent multi-path combining replacing conventional time recursive Kalman-like schemes for channel estimation and classic PLL structures for phase tracking. A receiver for M-ary orthogonal spread spectrum signaling is built on this model and is tested at very low signal to noise ratios without the aid of pilot symbols. Tests were conducted in shallow water ocean environments in Buzzards Bay MA and St. Margaret’s Bay. Empirical bit error rates are demonstrated at very low SNRs with coherent symbol decisions. Tests were conducted at various spreading gains and bandwidths to achieve rates up to 130 bps at 2 km with a demonstrated probability of bit error less than E-4 with 3 element combining at SNRs less than -20 dB. [This work is funded by the Office of Naval Research, SSC Pacific’s Naval Innovative Science and Engineering Basic and Applied Research Program as well as the University of Massachusetts Dartmouth].

 

Bio: Paul J. Gendron is an Assistant Professor at the University of Massachusetts Dartmouth. He received his PhD from Worcester Polytechnic Institute, his MS from Virginia Tech and his BS from the University of Massachusetts Amherst, all in Electrical Engineering. His work is broad in the fields of statistical signal processing, detection and estimation theory. His contributions range from seismic event detection and classification where he is the co-developer of the New England Seismic Network’s rapid seismic event detection algorithm to adaptive filtering, underwater acoustic communications and magnetic anomaly detection. In 2000 he was the recipient of an Office of Naval Research Fellowship award for his work with the Acoustics Division at the Naval Research Laboratory and in 2006 he was an Office of Naval Research Visiting Scientist to DRDC-Atlantic, Canada. Paul presently conducts research for the Office of Naval Research and SSC Pacific related to the discovery and invention of enabling technologies for undersea surveillance.

 

Host: Shengli Zhou, shengli@engr.uconn.edu