Objectives

Autonomy in robotic grasping and manipulation applications remains to be an elusive goal. A key challenge derives from the uncertainties arising before and during interactions with real objects. There are various sources of uncertainty involved for an autonomous robotic agent to succeed in targeted tasks. Environments and events are observed by agents through sensors, and sensory data is noisy, blurry, occluded, and an incomplete snapshot of a 3D world. In addition, other aspects leading to high uncertainty exist, such as lack of data, unmodelled effects (e.g. friction), partially known, or novel environments/situations, modeling errors. For the past three decades, these uncertainties have foiled best efforts to create automated systems for understanding the world as we experience it. Various approaches have been proposed to utilize uncertainty to drive exploration for information gathering. Despite many studies and significant progress over the last decades regarding different steps in robotic applications, such as grasping and manipulation, a robust and general approach for a wide variety of tasks and objects encountered in dynamic and unstructured environments and novel situations, which is close to human skills, does not exist yet. Current systems have limitations in terms of dealing with novelty, uncertainty and unforeseen situations. In this workshop, we will discuss challenges stemming from different sources of uncertainties involved in solving robotic grasping and manipulation tasks, and how to overcome them in the context of perception, learning and control, also taking into account human inspired studies and providing insights into how to further improve the systems defining how to best direct efforts.

Topics of Interests

The workshop will focus on the following areas, including but not limited to:

• Various approaches (learning, analytical, hybrid) dealing with the uncertainty or unobservability in sensing, actuation and modeling, Simulation to reality transfer

• Modeling, representation and integration of sensing modalities, e.g. proprioceptive, visual, force/torque, tactile, proximity sensing

• Grasping and manipulation of known, partially known or novel objects

• New quality measures for grasping under uncertainty

• Learning-based approaches for grasp planning and manipulation: e.g. model-based, model-free, data-efficient, multi-task, transfer, meta learning, reinforcement learning, learning from demonstration

• Analytic and hybrid approaches for grasping and manipulation

• Integration of data-driven with physics-based models for grasping and manipulation

• Modelling (object/hand/scene) interaction dynamics for grasp and manipulation tasks

• Integrating learning and control for grasping and manipulation

• Generalization and scalability of approaches to a variety of hands and objects

• Approaches addressing deformable/flexible object manipulation, dexterous grasping and manipulation, in-hand manipulation, bi-manual manipulation, mobile manipulation (e.g. legged, wheeled, aerial, underwater manipulation)

Organizers

Important Dates

Abstract submission deadline: October 13

Acceptance notification: October 20

Final materials due: October 26

Workshop date: October 27