Coordination Loops: a New Unit of Analysis

Coordination Loops: a New Unit of Analysis

Voshell et al. Coordination Loops: A New Unit of Analysis Coordination Loops: A New Unit of Analysis for Distributed Work Martin G. Voshell David D. Woods, Cognitive Systems Engineering Laboratory Cognitive Systems Engineering Laboratory The Ohio State University The Ohio State University [email protected] [email protected] Brian Prue Lisa C. Fern Cognitive Systems Engineering Laboratory Cognitive Systems Engineering Laboratory The Ohio State University The Ohio State University [email protected] [email protected] ABSTRACT Designing effective coordination into domains of distributed decision making and decentralized control is a daunting joint cognitive systems challenge. In order to support such coordination, it is necessary to look at modeling these systems from the perspective of coordination requirements. We propose “coordination loops” as a model that enables us to expand upon measures and constructs that allow specification requirements of distributed work to cultivate effective decision making across multiple domains. Keywords Joint Activity Theory, automated systems, teamwork, coordination loops. INTRODUCTION Coordinated skilled activity is fundamentally an ability to adapt. As new information systems, robotic resources, and robust communications are introduced into mission critical settings under the auspices of improving coordination across multiple roles, it is crucial to be able to anticipate and describe how new role changes and new forms of expertise are created as the systems change. Connectivity alone will not serendipitously result in effective coordination. Our previous work looked specifically at the coordination challenges of introducing new robotic platforms and capabilities in future operations (Woods, Tittle, Feil, & Roesler, 2004; Voshell & Oomes, 2006; Woods, Voshell, Roesler Phillips, Feil, & Tittle, 2006) and we continue to observe that developers and designers in such technology driven fields tend to underestimate these new role creations and neglect to acknowledge that new forms of autonomy and connectivity change the underlying system. Technologists typically respond by increasing isolated autonomy in attempt to create more coordinative agents- however we feel that this is fundamentally flawed. Such isolated increases ignore the principal coordination requirements necessary to make automated systems team players (Christofferson & Woods, 2002; Klein, Woods, Bradshaw, Hoffman, & Feltovich, 2004). Effective joint activity is not about autonomous skills, it is about coordination skills. We contend there exists a fundamental asymmetry in the coordinative capabilities between robots and people that is illustrated in Norbert’s Contrast (Woods & Hollnagel, 2006) in that literal minded automata will always be dependent on context bound human decision makers and stake-holders to coordinate with to fulfill the context gap. Christofferson and Woods (2002) previously advised that ‘the coordination across agents in the system is at least as important as the performance of the individual agents taken in isolation’. This current work focuses specifically on the importance of that coordinated activity across groups that functions as an emergent system of distributed work and is greater than individual agent role aggregation. If there is a coordinating agent, the agent itself cannot be the basic unit of analysis- effective coordinated activity must result in a response larger than the sum of individual agent or role actions. This kind of synergy entails that a scope of activity larger than any single agent’s interaction has to be the unit of reference where agents are part of a larger entity when describing coordination as the world changes. In designing new systems to help coordinating agents function effectively, we look to analyze this joint activity in terms of multiple classes of coordination loops (analogous to control loops in distributed and hierarchical control systems for Proceedings of the Eighth International NDM Conference (Eds. K. Mosier & U. Fischer), Pacific Grove, CA, June 2007 1 Voshell et al. Coordination Loops: A New Unit of Analysis engineered processes). The concept of the coordination loop is a further extension of joint activity theory originally applied to mixed human-robot team coordination. The coordination loops provide researchers and designers with a new perspective to utilize when exploring anticipation of coordination challenges while building new systems. In such loops agents synchronize and adjust goal-oriented activity with others as situations change and updates and impasses arise, all in order to achieve mission goals. As work becomes spatially and temporally distributed, lack of synchronization will exacerbate coordination challenges. These goal-oriented adaptive systems must often operate in surprising and rapidly changing environments. Introducing new technology shifts what the goals are and how they are pursued. To make significant progress toward successfully introducing new members (be they human, computer, or robot) as coordinating agents into any domain, they must be capable of demonstrating coordinative competencies that allow them to participate in adaptive and synchronized joint activity (Klein et al., 2005). Based on previous work, we look at multiple mission critical domains that depend upon effective joint activity utilizing the coordination loop model. The model proposes an initial breakdown of three classifications of coordination- horizontal loops, vertical loops, and projective loops. Borrowing examples from such domains as nuclear power plant control rooms, automated vehicle check-zones, mixed-team human robot search and rescue operations, small scale joint ground-UAV reconnaissance missions, and stability and support operations, we describe the mechanics underlying each of these loop classifications in selections of short anecdotes and scenarios. Defining Coordination Loops The three classes of coordination loop patterns provide measures and constructs that allow us to specify requirements of distributed work and evaluate many different complex systems to test for effective coordination and collaboration. Each coordination loop is a collection of groups formed in anticipation or in reaction to an event or series of events. These events are domain driven and can range anywhere from simple information gathering to coping with unexpected surprises in the field. By quickly describing and defining the loops below, we capture the dynamics that constitute skill across a variety of coordinative capabilities. Horizontal Loops In a theater of operation, coordination is defined on a horizontal loop by the groups (be they mixed, practice-oriented or ad- hoc, homogenous or heterogeneous) that work together within a similar scope and scale. Working on these loops, teams can form groups, shift and reform coordinative relationships (such as search and rescue personnel, robots and UAVs entering into an area of interest) to identify threats, discover injured parties, and move survivors to decontamination or aid stations. For another example, if a group is conducting a robot operation for IED disposal, emphasis is placed on the disposal task. As the scope and focus of secondary and tertiary tasks shift, there may be a goal adjustment that leverages a new functional demand for securing the safety of the robot operator. One can have separate horizontal loop structures carrying out independent work in any operation (i.e. law enforcement and fire personnel rarely work together) but our goal is to move beyond simply the teams that work together in order to describe coordination writ large and be inclusive of roles and functions that are not always acknowledged in planning and training (such as victims in an emergency that can adapt, have communications, and will act of their own volition). In our first short story below, we describe the startup process of a nuclear power-plant looking at coordination loops in a horizontal and technically coupled supervisory control setting. Vertical Loops Vertical loops are coordination connections and changes that act across echelons. Most simply this is a communication or modification of plans across levels. The vertical component refers to the mutual exchange of information between more distant levels of control with in-scene operations. For example, an incident command center needs to funnel aid resources toward the hot zone in an emergency response and this process depends on updates about the threats, needs, and activities in the search and rescue area. Similarly, in a vertical loop, activity of in-scene mixed teams depends on updated information from incident command centers that indicate what threats may be faced (key decisions about protective gear are crucial here), updates on what and where aid resources will be deployed, updates that indicate the potential for secondary attacks, or updates on urgency and priorities as events play out. Speaking in terms of echelons to describe the vertical component implies hierarchies, however, many systems are relatively uncoordinated and do not have rigid hierarchies. Search and rescue is often quite uncoordinated and military organizations often do not have strict hierarchies. Different domains have adopted and utilize different command systems (i.e. Incident Command System, Command and Control) that attempt to be flexible enough to adapt to changing situations. For this reason we have defined vertical loops in terms of changes in levels as descriptors acknowledging that in any of these systems there

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    6 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us