Chhaya Mudgal

Bilateral Negotiation with Incomplete and Uncertain Information: A Decision-Theoretic Approach Using a Model of the Opponent

The application of software agents to e-commerce has made a radical change in the way businesses and consumer to consumer transactions take place. Agent negotiation is an important aspect of e-commerce to bring satisfactory agreement in business transactions. We approach e-commerce and negotiation in the context of a distributed multiagent peer help system, I-Help, supporting students in a university course. Personal agents keep models of student preferences and negotiate on their behalf to acquire resources (help) from other agents. We model negotiation among personal agents by means of influence diagram, a decision theoretic tool. To cope with the uncertainty inherent in a dynamic market with self-interested participants, the agents create models of their opponents during negotiation, which help them predict better their opponents’ actions. We carried out experiments comparing the proposed negotiation mechanism with influence diagram, one using in addition a model of the opponent and one using a simple heuristic approach (as a base for comparison). The results show some of the advantages and disadvantages of the proposed negotiation mechanisms.

Using a fleet of slocum battery gliders in a regional scale coastal ocean observatory

Rutgers University is constructing the New Jersey Shelf Observing System (NJSOS), a regional-scale (300 km /spl times/ 300 km) observatory for the coastal ocean which includes the LEO15 site. Spatially extensive surface remote sensing systems (CODAR, satellites) are continuously collecting data in this region. However, only during the month of July in 1998, 1999, 2000 and 2001 were extensive subsurface physical and optical data collected and only in the LEO15 vicinity. The July samplings were labor and boat intensive. Obtaining the subsurface data in the LEO15 area required the use of up to seven boats and manpower to collect, process and analyze the information. To make the collection of subsurface data in the NJSOS more efficient, less expensive and more complete both spatially and temporally, Rutgers University has worked with Webb Research Corporation on their development of the Slocum Glider autonomous underwater vehicles. Currently Rutgers owns a fleet of four Gliders that can be used individually or as a group to collect subsurface data in the observatory region.

An influence diagram model for multi-agent negotiation

Bilateral negotiation is an important aspect of e-commerce to bring satisfactory agreement in business transactions. Though auction-based models are more common currently, they are unable to accommodate tradeoffs between multiple parameters. This makes them inappropriate when trading with intangible goods, like information, expertise and help. We approach e-commerce and negotiation in the context of a distributed multi-agent peer help system, I-Help, supporting students in a university course. Personal agents keep models of student preferences and negotiate on their behalf to acquire information resources (help) from other agents. Negotiation is modeled using an influence diagram, a decision theoretic tool. Agents create models of their opponents during negotiation, which help them predict better the opponent actions. The experiments show that the proposed approach brings better deals to the agent who uses it. However, if both negotiating agents use this approach, the negotiation fails more often.

Bilateral Negotiation with Incomplete and Uncertain Information

The paper describes a bilateral negotiation approach with incomplete and uncertain information that has been implemented in the context of a distributed multiagent peer help system, I-Help, supporting students in a university course. Personal agents keep models of student preferences and negotiate on their behalf to acquire resources (help) from other agents. The agents negotiate iteratively using an influence diagram, a decision theoretic tool. To cope with the uncertainty inherent in a dynamic market with self-interested participants, the agents create models of their opponents during negotiation, which help them predict better their opponents’ actions. The evaluation shows the advantages and disadvantages of the proposed negotiation mechanisms.

Transition of Slocum Electric Gliders to a sustained operational system

In the 1980s Slocum Gliders were a vision of Douglas C. Webb, which Henry Stommel promoted in a science fiction article published in Oceanography in 1989. In the early 1990s the glider concept was proven and in the late 1990s open water test flights were done at LEO15. In 2002 Rutgers University COOL Group began collaborating with Webb Research Corporation on the development and deployment of the Gliders. Initially the deployments were on the order of hours to a few days with constant human supervision. By the latter half of 2003 Slocum Gliders were routinely flying multiple week missions and calling in to the automated Glider Command Center on Rutgers main campus via satellite phone to provide a status update, download data and receive any new mission commands. The ability to operate Gliders with minimal human intervention for extended periods of time has allowed Rutgers to integrate them into the New Jersey Shelf Observing System. Since November 2003 a Glider has been occupying the Endurance Line, a 123 km track located between the LEO15 nodes and the shelf break, on a monthly basis. The sustained data set being collected permits scientists to go beyond collecting snapshots of information for short-term projects and gather long-term, expanded region data sets that would allow the tracking of trends over multiple years. While the Endurance Line Glider has been flying, additional Gliders have been operating for shorter periods of time on the West Florida Shelf, in the northwestern Mediterranean and in the Hudson River plume. Like the Endurance Line Glider, these Gliders are controlled by the Glider Command Center via satellite phone. Rutgers would be adding 2 Gliders to its fleet the end of this year bringing the total to 6 electric Gliders. One Glider would be dedicated to the West Florida Shelf Red Tide research program and the second would be used in the Mediterranean to look at the significance of Sahara Desert dust on biological and optical signals. Dedication of the new Gliders to these two research projects would enable Rutgers to have a continual presence in these regions as well as on the shelf of New Jersey.

Automated control of a fleet of Slocum gliders within an operational coastal observatory

Rutgers University, Webb Research, Dinkum Software, Wetlabs, and Mote Marine Lab have been collaborating on the development and deployment of a fleet of Slocum gliders to continuously patrol the coastal oceans. The gliders are AUVs that move up and down in the water column in a saw-toothed pattern by changing their buoyancy. Presently, during a deployment, humans must look at the data, determine if a change in the sampling protocol is indicated by the data and if so, upload a new mission to the glider. Rutgers has been focusing on the development and testing of the software to automate the control of the gliders. Using agent oriented programming, the goal of the software is to assimilate data received by the command center from the gliders and other agents, such as CODAR and satellites, and generate new missions for the glider fleet. The software is being developed on a Linux system. The ability of the command center computer to automatically analyze data from a glider or gliders and recognize thermoclines and haloclines were the first pieces of the control software to be developed. Testing of the thermocline software has been completed using both hand generated data and real data collected in January 2003 from the Gulf of Mexico. New sensors are being added for applications in the New York Bight and on the West Coast Florida Shelf.