With robots entering our every day lives the question of how to share autonomy in joint actions comes into the focus of robotics research. Shared autonomy arises, if two autonomous agents interact to coordinate their capabilities. This coordination can occur with regard to different aspects, such as cooperative task scheduling to minimize interference, temporal alignment to create synergies from behavior synchronization, or task partitioning to gain from complementary capabilities. Moreover, such coordination benefits if the two agents have knowledge about their mutual capabilities, resource constraints or internal states, such as beliefs, interests, intentions or even moods - if one of the agent is a human or a robot capable of social behavior. Some of this knowledge may be given, but some may be obtainable only through learning or interactive shared actions during which new information and insights arise. Therefore, Shared Autonomy opens the wider scope of understanding task-directed optimization at the autonomy level of each agent in its coupling with a spectrum of interaction-related constraints and value functions that arise when the autonomous agents have rich representations that include the representation of their self in its interaction with a partner, jointly embedded in a shared context. This entails a range of consequences for further research:

  1. How can we achieve a stronger focus on the perception of user intention and user state and their integration with the robot's world perception, intentions and state? By better recognizing the user's (local) intentions the robot can provide local support for achieving joint goals. This entails both, physical as well as communicative intentions and thus requires a broader model of the user that unifies task and communication models in a coherent way.
  2. Further, how can we achieve a relaxation of the constraints on robot autonomy? By allowing for less than optimal plan solutions and executions by the robot new space for emergent collaborative behavior is opened up that may exceed the robot's world model. However, if this criterion is relaxed a new optimization criterion is needed. How can this be achieved?
  3. Related to this, how can we achieve learning capabilities that extend the robot's active world model? In other words, the robot should be able to take advantage of emerging joint goal-directed behavior by enlarging its world model. How can we achieve to break this research barrier?

With this workshop we want to bring together robotics researchers with different perspectives in order to discuss and challenge existing concepts of shared autonomy and, stimulated by above considerations, provide a platform for the formulation of new ideas and proposals to overcome existing limitations. We see this as an important topic for robotic research in the future and expect continued workshops on this topic to build a platform where future research directions and strategies will be discussed.


IROS 2016 website

Please Note: due to last minute cancellations this workshop has been changed to a half day workshop starting at 12:30!