SMC-IT 2011 HRI workshop

Efficient control of multiple semi-autonomous robotic agents

NOTE: This workshop has passed. To find information about the current SMC-IT Conference, please visit the SMC-IT Conference Webpage.

Workshop abstract

In the near future, teams or “swarms” of robots may regularly perform complex, coordinated tasks (e.g., survey or reconnaissance, sensor placement, construction, search and rescue) on Earth and in space. These operations require the development of information systems and user interfaces (UIs) that allow humans to efficiently monitor, understand, and control the activity of many robots simultaneously. This workshop is intended to identify challenges in this domain, constraints on what can be achieved, and promising approaches to this kind of Human-Robot Interaction (HRI).

Workshop goals

  • Identify problems or challenges associated with command and control of robot teams or swarms in existing or novel operations.
  • Propose solutions or general approaches to problems associated with existing or novel operations concepts for robot teams or swarms.
  • Suggest measures or criteria for evaluating different approaches to the control of robot teams or swarms.
  • Describe how past or current operations involving single robots can inform an approach to controlling robot teams or swarms.
  • Explore user interface and/or information systems concepts that solve (or attempt to solve) problems analogous to those associated with existing or novel operations concepts for robot teams or swarms.

Workshop theme

Robotics technology is advancing rapidly, resulting in robotic agents that are capable, flexible, durable, and affordable. In the near future, teams or swarms of robots may regularly perform complex, coordinated tasks (e.g., survey or reconnaissance, sensor placement, construction, search and rescue) on Earth and in space.

Human operators will play an important role in the activities of these robot teams. Human involvement may be at the level of planning and scheduling tasks to be executed by robot teams that are almost entirely autonomous. Alternatively, human operators may interact with robots at a lower level, making navigational choices or initiating simple tasks for robots with limited autonomous capability. In between these extremes, a human operator may supervise a robotic team and assert control of the whole team, sub-teams, or individuals as necessary to coordinate behavior, implement strategic decisions, or to otherwise augment the robots’ capabilities with human capabilities.

In any case, the question arises: what information systems and user interfaces best support a human operator’s interaction with a mutli-robot team or swarm? The answer to this question must be informed by research in human-computer interaction (HCI) and human-robot interaction (HRI) (which can provide guidance on how humans make and implement decisions and how humans and user interfaces can improve or compromise efficiency) and also by research in robotics (which can provide guidance on the requirements of control systems, challenges associated with robot control, and experience with successful and unsuccessful approaches to control systems and interfaces).

Some of the factors that should drive the design and development of systems and interfaces for multi-robot control are also present for single-robot control. For example, a human operator needs information about the actions or tasks that are possible for a robot (e.g., what commands are acceptable), information about the robot’s state (e.g., power, position of actuators, instrument function), and information about the robot’s environment (e.g., instrument readings, sensor data).

When multiple robots are present in single operation, additional considerations must shape the operator’s control systems and interfaces. For example, if the actions of robot team members are tightly coupled (e.g., formation flying, coordinated grasping) then the human operator’s task is different than if the robots’ actions are loosely coupled (e.g., flocking, collective but uncoordinated mapping) or are completely independent, merely sharing an operational environment. Adding to this complexity is the possibility that the robots on a single team may be heterogeneous both in their capabilities and in the extent to which they couple their behaviors to that of other team members.

An information system designed to provide operator control of a robotic team might be considered successful if it accounted for factors like those mentioned above in a single context. That is, a system and interface might be designed to work well with a specific robot team. However, there are a number of potential advantages to designing more general systems and interfaces that could apply to a broad variety of teams (offering flexibility in team composition in terms of the types and numbers of robots). For example, there is significant potential for the accumulation and preservation of operator expertise over multiple operations if those operations employ a common approach to robot team command and control. In addition, reuse of a single system might provide cost savings (or other pragmatic efficiencies) to organizations that run several robotic operations.

This workshop is intended to bring together researchers from HCI, HRI, and robotics to discuss the factors that should shape the design and development of information systems and user interfaces to control multi-robot teams. To that end, workshop participants will identify challenges in this domain, constraints on what can be achieved, measures or criteria for evaluation of systems, and promising approaches to this kind of Human-Robot Interaction (HRI).