Two papers on robotic hand at Humanoids 2022 for APRIL

The picture of great minds mutually sharing their work and passion is not enough to describe what happened at Humanoids 2022, where Manuela Iuliano and Lucia Angelini went to represent the APRIL project in front of hundreds of scientists, researchers and robotic enthusiasts. 

In that environment, specifically created to spread how the paradigm of robotics is changing all around the world, we as APRIL had the opportunity to speak our truth through two scientific papers on the functioning of our robotic hand that we are incredibly proud of.

Self-collision avoidance algorithm for bimanual hand-arm robotic platform

The first paper, named “Self-collision avoidance in bimanual teleoperation using CollisionIK: algorithm revision and usability experiment”, was presented during an interactive session on November 29th, 2022 from 11 to 12 a.m. and was written by Lucia Angelini, Manuela Iuliano, Angela Mazzeo, Mattia Penzotti, and Marco Controzzi.

The work focuses on the development of a bimanual hand-arm robotic platform with a self-collision avoidance control algorithm. One of the challenges in teleoperation is avoiding self-collisions, which is particularly critical in bi-manual systems. This is why the authors propose a revised version of the CollisionIK algorithm, dubbed revised_CollisionIK, to solve this issue. 

The algorithm was tested in a bi-manual system teleoperated by naïve users and compared with the original version CollisionIK monitored by a standard emergency brake strategy. In particular, the authors focused on understanding what was the impact of the modification introduced in revised_CollisionIK during teleoperation. Based on objective and subjective metrics, results show that the proposed revised version of CollisionIK algorithm can be successfully used for teleoperating bimanual pick-handover-place tasks. Participants find the manipulation of small objects easier with this strategy and don’t perceive any difference in terms of accuracy and delay, despite these being significantly worse than CollisionIK combined with a standard emergency brake strategy.

A modular approach to the embodiment of hand motions

The second paper, named “A Modular Approach to the Embodiment of Hand Motions from Human Demonstrations”, was presented during an oral session (Grasping, Manipulation and Hands) on November 30th, 2022 at 16:40 p.m. from Alexander Fabisch of the DFKI team. The paper was written by Alexander Fabisch, Manuela Uliano, Dennis Marschner, Melvin Laux, Johannes Brust, and Marco Controzzi.

The starting point of the work is that manipulating objects with robotic hands is a complicated task. Not only the fingers of the hand, but also the pose of the robot’s end effector need to be coordinated. Using human demonstrations of movements is an intuitive and data-efficient way of guiding the robot’s behavior. In this work, a modular framework with an automatic embodiment mapping to transfer recorded human hand motions to robotic systems is proposed. In particular, a motion capture is used to record human motion. The approach is evaluated on eight challenging tasks (the APRIL use cases tasks), in which a robotic hand needs to grasp and manipulate either deformable or small and fragile objects. A subset of trajectories are tested both in simulation and on a real robot and the overall success rates are promising and aligned.

The outcome for APRIL and its robotic hand

As for every large event (such as IROS 2022), Humanoids revealed its relevance on the dissemination and communication side. When you are in line to get the world to know your project, the effort you and other 13 partners are constantly making to ensure its progress and the path you are pursuing, you are making a step in the right direction.

This is what Humanoids represents for APRIL: one of the many steps which are going to lead this project on top of the hill.