Periodic Reporting for period 3 - An.Dy (Advancing Anticipatory Behaviors in Dyadic Human-Robot Collaboration)
Reporting period: 2019-07-01 to 2021-08-31
AndyControl (milestone 1). Another important achievement of the first year is the implementation of the first milestone: a reactive controller for physical human interaction. Details on the implementation of this controller have been published outside the consortium in the paper by F.Romano et al., published on Robotics and Automation Letters [13]. The paper represents a bond with CoDyCo (Project nr.: 600716, FP7-ICT-2011.2.1 Cognitive Systems and Robotics), the FP7 project which shares with AnDy 3 out of 4 scientific partners (IIT, INRIA, JSI).
In the manufacturing domain, the AndySuit has been tested in conjunction with the Ottobock passive exoskeleton to evaluate the impact of the exoskeleton in reducing the muscular-skeletal stress during overhead works. Quantitative analyses of these evaluation are already planned in the second year of the project.
Hereafter, an overview of the project results towards the objective of the project, with references to the structure of the Annex 1 to the Grant Agreement.
• AndySuit (objective 1). During the first year of the project An.Dy the consortium carried out activities according to what detailed in the Description of Action (Annex 1 to the Grant Agreement). In particular, the consortium developed the AndySuit, a wearable technology for real-time monitoring humans when performing whole-body physical interaction tasks. Measured quantities include human kinematics (position, velocity, acceleration) and dynamics (contact forces, joint torques).
• AndyControl (milestone 1). Another important achievement of the first year is the implementation of the first milestone: a reactive controller for physical human interaction. Details on the implementation of this controller have been detailed confidentially within the consortium (Deliverable 4.1) and publicly outside the consortium with a paper by F.Romano et al., published on Robotics and Automation Letters [13]. The paper represents a bond with CoDyCo (Project nr.: 600716, FP7-ICT-2011.2.1 Cognitive Systems and Robotics), the EU project which shares with AnDy 3 out of 4 scientific partners (IIT, INRIA, JSI).
These achievements are the prerequisite for enabling the following stages of the AnDy project. In particular, the AndySuit will be exploited during the second year for the AndyDataset collection (objective 2) and the implementation of the ergonomic collaboration (milestone 2).
Objectives
In order to achieve the aforementioned first year objectives, the consortium focused its activities on the design of the necessary software and hardware components for the AndySuit (objective 1) and for implementing the reactive collaboration (milestone 1). At first, the whole consortium contributed to defining the foundations of the upcoming activities. These foundations are summarized in the following achievements.
A first group of achievements concern the ethic aspects.
• Dissemination plan. Publication of the AnDy dissemination plan.
• EPQ-POPD-M-OEI-DU-GEN requirements. A document on the SSH/RRI dimensions of this research and its outcomes, as it encompasses extensively human-robot interaction and includes qualitative assessments of the technologies developed.
• Ethics guidelines. The plan and procedures for managing ethical issues within the AnDy project.
• Data management plan. The plan for the management of data, with specific focus on the AndyDataset.
A second group of deliverables define the technical specifications for technologies used and developed within the project.
• Specifications for the AndySuit. A data-sheet describing the quantitative specifications of the AndySuit (i.e. desiderata). Specifications primarily consider the goals to be achieved in the following stages (see in particular WP4 and WP5).
• AndyDataset specifications. This deliverable concerns the AndyDataset specifications as defined in T2.1. It consists of a technical report containing specifications such as: adopted standards for sensor positioning (e.g. motion capture markers), sensor modalities, acquired biometric data, considered interactive and collaborative motions.
• AndyModel: dynamic models. This deliverable is mainly related to T3.1 and it is a scientific report or paper on musculo-skeletal models for real-time force and motion tracking. Existing detailed musculoskeletal models have been computationally enhanced to meet as many of the project requirements as possible, and model reduction is used to provide accurate data for the simpler human models e.g. for real-time execution.
Finally, a third set of deliverables coincides with the aforementioned objective and milestone for year one.
• Reactive control. This deliverable is the theoretical foundation of partner-aware reactive control.
• AndySuit. This deliverable is about the first prototype of the suit. At month 12, force and motion estimations will be based on wearable motion tracking system with wearable force platforms.
Impact on the scientific domain
In the scientific domain, the AnDy consortium participated at 19 international scientific events and published 17 scientific papers in international peer-reviewed conference proceedings and journals. Also, the IIT scientists are members of the Co-Aware team, finalist of the Kuka innovation award 2018 (https://www.kuka.com/en-de/technologies/research-and-innovation/kuka-innovation-Award).
Impact on media
The AnDy project already attracted a significant interest from media, especially those that target the wide public. Members of the AnDy consortium have been involved in more than 36 national and international events (newspapers, news tv-channels, etc.).
Impact on the manufacturing domain
In the manufacturing domain, the AndySuit has been tested in conjunction with the OBGH passive exoskeleton to evaluate the impact of the exoskeleton in reducing the muscular-skeletal stress during overhead works. Quantitative analyses of these evaluation are already planned in the second year of the project.
Impact on big data
INRIA performed some preliminary data collections with the XSens suit in the laboratory (indoor) and on an industrial factory floor of Thyssenkrupp Presta France. Data have been collected and are now available to the consortium. This preliminary dataset is the seed of an extensive dataset to be used in the context of big data.