Arrow Research search

Author name cluster

Nathan Schurr

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

9 papers
2 author rows

Possible papers

9

AAMAS Conference 2011 Conference Paper

Adaptive Decision Support for Structured Organizations: A Case for OrgPOMDPs

  • Pradeep Varakantham
  • Nathan Schurr
  • Alan Carlin
  • Christopher Amato

In today's world, organizations are faced with increasingly large and complex problems that require decision-making under uncertainty. Current methods for optimizing such decisions fall short of handling the problem scale and time constraints. We argue that this is due to existing methods not exploiting the inherent structure of the organizations which solve these problems. We propose a new model called the OrgPOMDP (Organizational POMDP), which is based on the partially observable Markov decision process (POMDP). This new model combines two powerful representations for modeling large scale problems: hierarchical modeling and factored representations. In this paper we make three key contributions: (a) Introduce the OrgPOMDP model; (b) Present an algorithm to solve OrgPOMDP problems efficiently; and (c) Apply OrgPOMDPs to scenarios in an existing large organization, the Air and Space Operation Center (AOC). We conduct experiments and show that our OrgPOMDP approach results in greater scalability and greatly reduced runtime. In fact, as the size of the problem increases, we soon reach a point at which the OrgPOMDP approach continues to provide solutions while traditional POMDP methods cannot. We also provide an empirical evaluation to highlight the benefits of an organization implementing an OrgPOMDP policy.

PDF

AAMAS Conference 2010 Conference Paper

Agent-based Coordination of Human-Multirobot Teams in Complex Environments

  • Alan Carlin
  • Jeanine Ayers
  • Jeff Rousseau
  • Nathan Schurr

Room clearing, in which building surveillance is conducted tosearch for criminals, continues to be a dangerous and difficultproblem in urban settings, for both the military as well as forpolice. In a typical setting, an unknown number of hostile forcesmay be located in a building, and they may be armed. Furthermore, there may be innocent civilians. The goal of thefriendly units is to enter the room and secure it, but without lossof life of friendly forces, hostile forces, and most especially ofinnocent civilians. It would be beneficial to allow robots to be apart of the friendly team, however it is very challenging to haverobots that do not either slow down or obstruct their humanteammate. This is especially difficult since nearly all robots in useby the military and police today are tele-operated. In this paper, we describe work we have developed in cooperation with thearmy, for the room clearing domain. We constructed an algorithmwhereby multiple agents, in the form of robots, can accomplish aroom clearing task. We augmented the agent algorithms tointroduce Adjustable Autonomy, allowing cooperation withhumans. We describe simulated results of the algorithm onbuilding maps, and furthermore we describe how we intend tonext conduct hardware tests, and eventual plans to field thesystem. This agent-based solution has great potential to increasethe acceptance and leverage of robotics in complex environments.

PDF

UAI Conference 2010 Conference Paper

ALARMS: Alerting and Reasoning Management System for Next Generation Aircraft Hazards

  • Alan Carlin
  • Nathan Schurr
  • Janusz Marecki

The Next Generation Air Transportation System will introduce new, advanced sensor technologies into the cockpit. With the introduction of such systems, the responsibilities of the pilot are expected to dramatically increase. In the ALARMS (ALerting And Reasoning Management System) project for NASA, we focus on a key challenge of this environment, the quick and efficient handling of aircraft sensor alerts. It is infeasible to alert the pilot on the state of all subsystems at all times. Furthermore, there is uncertainty as to the true hazard state despite the evidence of the alerts, and there is uncertainty as to the effect and duration of actions taken to address these alerts. This paper reports on the first steps in the construction of an application designed to handle Next Generation alerts. In ALARMS, we have identified 60 different aircraft subsystems and 20 different underlying hazards. In this paper, we show how a Bayesian network can be used to derive the state of the underlying hazards, based on the sensor input. Then, we propose a framework whereby an automated system can plan to address these hazards in cooperation with the pilot, using a Time-Dependent Markov Process (TMDP). Different hazards and pilot states will call for different alerting automation plans. We demonstrate this emerging application of Bayesian networks and TMDPs to cockpit automation, for a use case where a small number of hazards are present, and analyze the resulting alerting automation policies.

Details

AAMAS Conference 2010 Conference Paper

Function Allocation for NextGen Airspace via Agents

  • Nathan Schurr
  • Paul Picciano
  • Janusz Marecki

Commercial aviation transportation is on the rise and hasbecome a necessity in our increasingly global world. Thereis a societal demand for more options, more traffic, moreefficiency, while still maintaining safety in the airspace. Tomeet these demands the Next Generation Air Transportation System (NextGen) concept from NASA calls for technologies and systems offering increasing support from automated decision-aiding and optimization tools. Such systemsmust coordinate with the human operator to take advantage of the functions each can best perform: The automatedtools must be designed to support the optimal allocation oftasks (functions) between the system and the human operators using these systems. Preliminary function allocationmethods must be developed (and evaluated) that focus onthe NextGen Airportal challenges, given a flexible, changingConcept of Operations (ConOps). We have begun making steps toward this by leveragingwork in agents research (namely Adjustable Autonomy) inorder to allow function allocation to become more dynamicand adjust to the goals, demands, and constraints of thecurrent situation as it unfolds. In this paper we introduceDynamic Function Allocation Strategies (DFAS) that arenot static and singular, but rather are represented by allocation policies that vary over time and circumstances. TheNextGen aviation domain is a natural fit for agent basedsystems because of its inherently distributed nature and theneed for automated systems to coordinate on tasks mapswell to the adjustable autonomy problem. While current adjustable autonomy methods are applicable in this context, crucial extensions are needed to push the existing models tolarger numbers of human players, while maintaining criticaltiming. To this end, we have created an air traffic controlsystem that includes: (1) A simulation environment, (2) aDFAS algorithm for providing adjustable autonomy strategies and (3) the agents for executing the strategies and measuring system efficiency. We believe that our system is thefirst step towards showing the efficacy of agent supportedapproach to driving the dynamic roles across human operators and automated systems in the NextGen environment. We present some initial results from a pilot study using thissystem.

PDF