New challenges in humanoid grasping and manipulation

Objectives

Notwithstanding the many advances, achieving dexterous grasping and manipulation capabilities with robotic hands remains one of the biggest challenges in robotics, and numerous researchers around the globe are trying to tackle this issue under different aspects. All these attempts cannot leave out of consideration that robotics has matured from the industry setup and started to make a break into our daily lives. Accordingly, a paradigm shift also occurred in the research on grasping and manipulation that has evolved from rigid manipulators and specialized grippers, to dexterous human-like robotic arms and hands that are capable of grasping and manipulating a wide range of daily living objects, performing vastly different tasks in unstructured environments. A key enabler for this technological change is the embodiment of mechanical intelligence in the system design, through the purposeful introduction of soft elements in the structure of the hand, implementing a low level physical control loop to take care of local uncertainties during grasping and manipulation. The new requirements come along with fresh challenges and opportunities. In response to the uncertainty in the perception and dynamics of the physical interaction with the unstructured environment, new research work in hardware design and intelligent software, sensing and perception, control has grown and flourished. Robotic arms and hands become more dexterous, tactile sensitive, robust to uncertainty, durable, lightweight, and inexpensive. At the same time flexibility, adaptability, and learning ability have been enhanced, also capitalizing on recent advances on software and machine learning.


The aim of this workshop is to collect these experiences, and to serve as a platform to discuss and connect among researchers in different sub-fields of grasping and manipulation. Fully achieving this goal through a one day workshop is a very ambitious task, which we aim at reaching moving from two pillars. First – in accordance with the main topic of the conference – we will focus on the specific case of humanoid and human inspired design solutions and control/perception strategies. Second, we propose this workshop as an extension of the Technical Session of Grasping and Manipulation which will be part of the main conference.

The workshop is articulated into three main sessions

  1. Body (Design)
  2. Brain (Control, Learning, Planning)
  3. Perception (Sensors, and Sensing strategies).

In this way, we aim at covering the whole area of grasping and manipulation, encouraging discussions between researchers working in different fields. To further foster discussions and interchange of ideas, the workshop will include a mix of widely recognized experts in the field, younger researchers, and students. Student talks will be selected among the most exciting results submitted to the technical session.

Topics of interest

· Design solutions: soft hands, anthropomorphic hands, under actuated hands, iper-actuated hands, bio-inspired mechanisms, variable-stiffness joints

· Control strategies: deep learning, machine learning, model based control, model based planning, model based optimization, bio-inspired control, loco-manipulation in humanoids, motor control of the human hand, datasets, grasping with soft hands

· Sensing strategies: tactile sensors, flexible sensors, haptic exploration, strain sensors, kinematic measurements, sensor fusion, vision, perception for grasping in humans, datasets, sensing with soft hands

Endorsing Technical Committee:

This workshop has been endorsed by the IEEE-RAS Technical Committees on:

Robotic Hands Grasping and Manipulation, Mobile Manipulation, Bio Robotics, Human-Robot Interaction & Coordination, Soft Robotics, Robot Learning, Mechanisms and Design.

We are grateful for this support.

Organizers

Dr. Cosimo Della Santina

Massachusetts Institute of Technology (MIT-CSAIL), 22 Vassar street, Cambridge, MA, USA.

Email: cosimodellasantina@gmail.com


Prof. Matteo Bianchi

Centro E. Piaggio, Department of Information Engineering, University of Pisa, Largo Lucio Lazzarino 1, Pisa, Italy.

Email: matteo.bianchi@centropiaggio.unipi.it


Prof. Yu Sun

Department of Computer Science and Engineering at the University of South Florida, 4202 E Fowler Ave., ENB 118, Tampa, FL, USA.

Email: yusun@mail.usf.edu