Periodic Reporting for period 2 - SoftEnable (Towards Soft Fixture-Based Manipulation Primitives Enabling Safe Robotic Manipulation in Hazardous Healthcare and Food Handling Applications)
Reporting period: 2023-10-01 to 2025-03-31
Fresh food processing is a labour intensive task requiring workers to operate in hazardous environments such as cold rooms while handling delicate produce. Industry KPIs however include reducing human exposure to these harsh conditions and avoiding injury while maintaining high levels of productivity and employee satisfaction. These challenges are in principle well-suited for robotics, but the handling of delicate items such as meat and fish requires new types of robotic manipulation primitives to secure the manipulated object safely. SoftEnable proposes a new framework for manipulation actions involving deformable delicate objects based on the notion of soft fixture-enabled manipulation primitives. While fixtures by means of robotic grasps or specifically designed workpiece fixtures are routinely applied in industrial robotics applications involving rigid objects such as metal parts, there is a need to generalize these robotic methods to materials and situations that require the robust manipulation of soft or fragile items. SoftEnable will extend existing robotic grasping and manipulation techniques based on a novel soft fixture-based framework that generalize the notion of caging grasps together with the co-development of a soft fixture-optimized robotic manipulator and tooling for this purpose. Besides the fresh food handling use case, SoftEnable will focus on the healthcare sector where the project will demonstrate how soft fixtures can be utilized by a robotic system assisting health care workers by reducing stress and risk exposure during the handling, preparation, dressing, undressing and safe disposal of personal protective equipment (PPE).
During the first 30 months of the project, the SoftEnable project has completed several key achievements resulting in a total of 16 scientific publications at journals, conferences and workshops as well as 6 preprint works.
During the startup phase of the project in month 1-12, a regular schedule of working group and consortium meetings to facilitate collaboration was adopted. Since SoftEnable has a strong focus on providing benefits to human workers, an ethical project framework towards human study participants (D1.1) was established, ensuring the highest ethical standards are followed during the project run-time. The framework established a clear approach to procedures for human involvement for both intended use cases of our project. We completed and established a data management plan (D6.2) which will be revised as the project progresses. To further strengthen the ethics and GDPR expertise available to the consortium, an external ethics advisory board and external export advisory board was furthermore established (D8.1). To ensure that the technical project results are aligned with the use case requirements and to track progress in an appropriate manner, we completed work on use case requirements and KPIs (D1.2) and project partner OCADO provided the consortium with silicone phantoms of food items for experimentation. A Software integration framework based on containerized individual software components was adopted (D4.1).
During months 13-30, significant hardware and methodology progress was achieved. This included a patent application for a novel thin layer separation tool developed by CSIC and a gripper design by DLR and the development of a scooping tool by Technion (D3.2). Several gripper design iterations with novel components, including an extendable palm and finger approach (DLR) as well as the use of high-friction contacts (CSIC) were proposed and reported on in D3.1. Partners significantly increased collaborative research efforts, for example with methodology for co-design of grippers being developed jointly, and parties working closely on joint demonstrator and integration efforts. A novel taxonomy for deformable object manipulation was developed collaboratively to guide classification of manipulation primitives under a unified framework. The consortium had a significant overall research output in the current review period, with the publication of 11 scientific works and 6 preprint works listed in the technical report during M11-30. Publication venues included IEEE Robotics and Automation Letters as well as IEEE ICRA and IEEE/RSJ IROS. Work towards two use case demonstrators – one per use case - has advanced significantly, with results being reported in deliverables D4.2 and D5.2. The consortium held regular integration meetings for this purpose and identified and developed key software components and datasets. Further details can be found in deliverables D2.1 D2.2 and D2.3.
During the startup phase of the project in month 1-12, a regular schedule of working group and consortium meetings to facilitate collaboration was adopted. Since SoftEnable has a strong focus on providing benefits to human workers, an ethical project framework towards human study participants (D1.1) was established, ensuring the highest ethical standards are followed during the project run-time. The framework established a clear approach to procedures for human involvement for both intended use cases of our project. We completed and established a data management plan (D6.2) which will be revised as the project progresses. To further strengthen the ethics and GDPR expertise available to the consortium, an external ethics advisory board and external export advisory board was furthermore established (D8.1). To ensure that the technical project results are aligned with the use case requirements and to track progress in an appropriate manner, we completed work on use case requirements and KPIs (D1.2) and project partner OCADO provided the consortium with silicone phantoms of food items for experimentation. A Software integration framework based on containerized individual software components was adopted (D4.1).
During months 13-30, significant hardware and methodology progress was achieved. This included a patent application for a novel thin layer separation tool developed by CSIC and a gripper design by DLR and the development of a scooping tool by Technion (D3.2). Several gripper design iterations with novel components, including an extendable palm and finger approach (DLR) as well as the use of high-friction contacts (CSIC) were proposed and reported on in D3.1. Partners significantly increased collaborative research efforts, for example with methodology for co-design of grippers being developed jointly, and parties working closely on joint demonstrator and integration efforts. A novel taxonomy for deformable object manipulation was developed collaboratively to guide classification of manipulation primitives under a unified framework. The consortium had a significant overall research output in the current review period, with the publication of 11 scientific works and 6 preprint works listed in the technical report during M11-30. Publication venues included IEEE Robotics and Automation Letters as well as IEEE ICRA and IEEE/RSJ IROS. Work towards two use case demonstrators – one per use case - has advanced significantly, with results being reported in deliverables D4.2 and D5.2. The consortium held regular integration meetings for this purpose and identified and developed key software components and datasets. Further details can be found in deliverables D2.1 D2.2 and D2.3.
The scientific results of SoftEnable currently consist of 16 scientific publications and 6 preprints. The periodic report lists key advances for each publications and progress during the period touches upon method and hardware development for SoftEnable as well as progress towards the definition of a taxonomy for deformable object manipulation. Two separate hardware patent applications were submitted based on tool and gripper design developments undertaken as part of the project.