Research Projects

Knowledgeable Dynamic-Process Modeling and Execution (March 2009–March 2010)

Infosys Technologies/Boeing
Daniel Corkill

We are developing a dynamic-process management framework that extends our previous research on ProME and KPM. Managing and participating in complex, dynamic processes is difficult due to their inherent uncertainty, which undermines the predictability necessary for efficient planning and execution. Effective management of these processes hinges on the ability of the manager to recognize unanticipated difficulties in the process execution, determine the causes of the anomalies, and implement remedies. Current process-management approaches respond reactively to process dynamics, if they deal with them at all.

We are focusing on how human process managers and participants interact with a dynamic, on-line representation of executing dynamic processes to proactively manage and operate in dynamic business processes. We show how having the best information available about a process and its expected future can provide managers with the time needed to detect and understand impending process anomalies and to develop and implement effective interventions.

Bootstrapped Learning (April 2007–November 2009)

SRI International/DARPA
Daniel Corkill, Huzaifa Zafar

The Bootstrapped Learning program seeks to make instructable computing a reality. An “electronic student” is being developed that will learn from a human teacher who uses spoken language, gestures, demonstration, and many other methods that one would find in a human mentored relationship. We are contributing to the design and implementation of the knowledge repository, control module, and self-diagnosis component of the MABEL electronic student.

Past Projects

A Cognitive Framework for Resource-Aware Sensor Net Organizations (January 2005–December 2007)

Air Force Research Laboratory
Daniel Corkill, Victor Lesser, Steven Hoffman, Michael O'Neill, Huzaifa Zafar

CNAS (Collaborative Network for Atmospheric Sensing) is a rapidly deployable, agent-based, power-aware microsensor network for ground-level atmospheric monitoring. To conserve battery power, and hence extend the useful lifetime of the network, CNAS sensor agents must have their Wi-Fi communication adapters turned off much of the time. This complicates agent interaction and network responsiveness since an agent cannot simply turn on its Wi-Fi radio when it needs to send a message; it has to ensure that other agents also have their radios turned on in order to receive or relay the message. Such power constraints required CNAS agents to collaborate effectively at an architectural level as well as on atmospheric monitoring tasks.

CNAS research activities were performed in five main areas: 1) implementation of a blackboard-based, individual agent architecture that could be effective given the limited computing facilities at each sensor agent; 2) development of the CNAS agent and network software architecture and atmospheric sensing knowledge sources; 3) improving CNAS performance and responsiveness with limited radio availability; 4) power-aware reasoning associated with solar harvesting performed by each sensor agent; and 5) exploration of potential next-generation CNAS hardware.

Field deployments of CNAS included the 2006 PATRIOT Exercise held at Fort McCoy, Wisconsin and two deployments (drop-zone and urban) at the 2007 Talisman Saber Combined Exercise conducted in Queensland, Australia.

An Enhanced Collaborative-Software Environment for Information Fusion (September 2004–January 2007)

U.S. Army RDECOM CERDEC ARES-FIST program
Daniel Corkill, Raphen Becker, Paul Kohler, Kenneth Watts, Huzaifa Zafar

We developed a highly responsive, blackboard-system based, decision-support technology for improving the effectiveness of analysts and decision makers within the Army's Brigade Combat Teams (BCTs). The CIFA (Collaborative Information-Fusion Assistant) reasoner (CIFAR) augments and supports Army personnel in answering Priority Intelligence Requirements (PIRs) associated with monitoring, assessing, and responding to enemy courses of action and other battlespace-environment characteristics. A major challenge in CIFA is managing the combinatorial explosion of sensing and processing activities without sacrificing accurate inference. The large volumes of data, possibilities, and outcomes exceed human perceptual and cognitive abilities and require an effective human/computer partnership to make the best use of sensing, computation, and communication resources in highly dynamic and uncertain battlefield environments. Before CIFAR, time constraints and information overload resulted in hasty, partial analysis of the information available to intelligence personnel. CIFAR was designed to help Army analysts and decision makers focus their attention on appropriate data by providing spatially and temporally aggregated views of the environment and by ensuring that important information has not been overlooked.

Our research focused on representation strategies for effective reasoning, temporal and spatial aggregation techniques, and control of reasoning processes. Significant collective information is present in the stream of individual human-generated and automated-sensor intelligence, surveillance, and reconnaissance (ISR) reports that are available to analysts and decision makers. Harvesting this rich collective information required semantically aggregating individual reports in both space and time. Rather than discarding detailed spatial and temporal information in order to simplify automated reasoning, we investigated how to make use of all the information that can be obtained from ISR and other sources.

Architectural Analysis and Investigation of Command & Control, and Data Fusion Systems (September 2004–March 2008)

DND Canada TDP
Daniel Corkill

We provided AI software-architecture design recommendations and reviews to DRDC Valcartier and contractors developing above-water anti-missile threat assessment and weapon assignment systems for next generation Halifax class Canadian Navy frigates.