Antonio Bicchi – University of Pisa
Title: Body+Environment=Function. Studying the embodied intelligence equation
Abstract: Recent experimental evidence of subjects using anthropomorphic artificial hands as a replacement of their own showed that humans exhibit an impressive capability of grasping and manipulation. This surprising result is in part related to the design of the hand, which uses soft robotics and synergy-inspired motor control to embody a intelligent physical interaction with the environment. To exploit the potential offered by such hands in an autonomous robot, however, the classical grasp planning paradigm has to be shifted completely. In this talk we will present our recent results in this direction.
Aaron Dollar – Yale University
Title: Dexterous, Within-Hand Manipulation Leveraging Compliance and Underactuated Mechanisms.
Abstract: In this talk I will overview my group’s work on the design and utilization of underactuated robot hands that utilize their under-constrained nature in order to passively adapt to added external and internal constraints. Beyond fundamental grasping, we will demon- strate a number of different within-hand manipulation modalities, and hand designs geared towards each.
Matei Ciocarlie – University of Columbia
Title: Contact Sensing for Manipulation: Thoughts on Why and How.
Abstract: In this talk I will describe very recent progress from our lab in the area of contact sensing. From a theoretical perspective, I will discuss how sensory data could fit into the established framework of grasp quality metrics. From an implementation perspective, I will discuss new methods for achieving high resolution tactile sensing. One traditional approach is to fabricate a large number of taxels, each delivering an individual, isolated response to a stimulus. In contrast, we propose a method where the sensor simply consists of a continuous volume of piezoresistive elastomer with a number of electrodes embedded inside. In addition to extracting rich location information from few wires, this approach lends itself to simple fabrication methods and makes no assumptions about the underlying geometry, simplifying future integration with robot fingers.
Danica Kragic – KTH
Title: Grasp planning and control using Grasp Moduli Spaces
Abstract: In this talk, we present our approach for modelling grasping using an integrated space of grasps and shapes. In particular, we introduce an infinite dimensional space, the Grasp Moduli Space, which represents shapes and grasps in a continuous manner. We define a metric on this space allowing us to formalize “nearby” grasp/shape configurations and we discuss continuous deformations of such configurations. In the case of surfaces of revolution, we determine stable grasps which correspond to grasps used by humans and develop an efficient algorithm for generating those grasps in the case of three contact points. We show that sufficiently stable grasps stay stable under small deformations. For larger deformations, we develop a gradient-based method that can transfer stable grasps between different surfaces. Additionally, we show in experiments that our gradient method can be used to find stable grasps on arbitrary surfaces with cylindrical coordinates by deforming such surfaces towards a corresponding “canonical” surface of revolution.
Domenico Prattichizzo – University of Siena
Title: Modeling and Mapping Human Strategies to Robot Hands.
Abstract: In this talk I will describe methods to map human hand motions to robotic hand/arm systems. The mapping procedure output is a representation, in the robot configuration space, of the primitives coming from the motions and interactions observed in humans, which can be used as robot actuation reference values, to be sent to the robotic control systems. A novel interaction of the human with a robotic sixth finger will be presented as an application of the mapping procedure.
Maximo Roa – DLR
Title: Moving grasp and manipulation planning closer to reality
Abstract: Traditional grasp planning uses different as- sumptions on critical aspects of the object-hand inter- action, such as the type of contact or the maximum force that the fingers can apply. Unfortunately, several of these assumptions are very abstract for most of the real grasping and fixturing cases. In this talk, relaxation of the assumptions on punctual contacts and on normalized forces applied at the fingertips will be shown, to move the planning aspects closer to real executions. The second part of the talk will show how the grasp planner is integrated with an assembly and a path planner, therefore allowing the generation of realistic assembly plans for modular structures, which are then executed on a setup using an LBR-iiwa arm for assembling structures using ITEM profiles.
Alberto Rodriguez – MIT
Title: Exploiting the environment with hardness. Lessons learned the hard way.
Abstract: In this presentation I will describe a simple and robust approach to bring dexterous manipulation to part handling. The key idea is to manipulate grasped objects by pushing them against the environment, harnessing extrinsic dexterity. This approach leads to levels of dexterity with simple grippers not yet demonstrated by much more complex robotic hands. The precision and control of these actions relies on understanding the interaction between the gripper, the part, and the environment, which builds on assumptions of hard-contact rigid-body coulomb-friction interactions. Hardness has its benefits.
But the world is messy. It is often different than we expect in more dimensions that I can comfortably list, which makes interaction often unpredictable. Softness, along with other means of mechanical intelligence, makes it easier by absorbing all or part of the messiness. Softness has its benefits.
Given the little impact that hard robotic manipulation has had in real applications, it seems advisable to explore new paradigms. Soft manipulation is a great candidate. But we face the danger of breaking ties and creating yet another isolated island of knowledge, instead of expanding on the one we have been living in.
Oliver Brock – Technical University of Berlin
Title: Can We Solve Manipulation Planning by Switching from Configuration Space to Interaction Space?
Abstract: Planning manipulation in configuration space is notoriously difficult. The nature of the manipulation problem requires us to consider complex contact states and the many transitions between them. To be realistic, we’d even have to consider dynamics, surface properties, friction, etc. Furthermore, we must account for uncertainty as minor variations in the configuration can decide between success and catastrophic failure. But is maybe some part of the challenges we have encountered in manipulation the result of the space we use to represent manipulation states? In this talk, I will think out loud about the consequences of attempting to plan in the space of interactions instead of in configuration space.