The collection of delicate deep-sea specimens of biological interest with remotely operated vehicle (ROV) industrial grippers and tools is a long and expensive procedure. Industrial grippers were originally designed for heavy manipulation tasks, while sampling specimens requires dexterity and precision. We describe the grippers and tools commonly used in underwater sampling for scientific purposes, systematically review the state of the art of research in underwater gripping technologies, and identify design trends. We discuss the possibility of executing typical manipulations of sampling procedures with commonly used grippers and research prototypes. Our results indicate that commonly used grippers ensure that the basic actions either of gripping or caging are possible, and their functionality is extended by holding proper tools. Moreover, the approach of the research status seems to have changed its focus in recent years: from the demonstration of the validity of a specific technology (actuation, transmission, sensing) for marine applications, to the solution of specific needs of underwater manipulation. Finally, we summarize the environmental and operational requirements that should be considered in the design of an underwater gripper.
In order to develop a gripping system or control strategy that improves scientific sampling procedures, knowledge of the process and the consequent definition of requirements is fundamental. Nevertheless, factors influencing sampling procedures have not been extensively described, and selected strategies mostly depend on pilots’ and researchers’ experience. We interviewed 17 researchers and remotely operated vehicle (ROV) technical operators, through a formal questionnaire or in-person interviews, to collect evidence of sampling procedures based on their direct field experience. We methodologically analyzed sampling procedures to extract single basic actions (called atomic manipulations). Available equipment, environment and species-specific features strongly influenced the manipulative choices. We identified a list of functional and technical requirements for the development of novel end-effectors for marine sampling. Our results indicate that the unstructured and highly variable deep-sea environment requires a versatile system, capable of robust interactions with hard surfaces such as pushing or scraping, precise tuning of gripping force for tasks such as pulling delicate organisms away from hard and soft substrates, and rigid holding, as well as a mechanism for rapidly switching among external tools.
17 August 2023 in IEEE Robotics and Automation Letters
This paper proposes an adaptive method for robot-to-human handovers under different scenarios. The method combines Dynamic Movement Primitives (DMP) with Preference Learning (PL) to generate online trajectories that are reactive to human motion, modulating the speed of the robot. The PL allows for tuning the coupling parameters of the DMP, tailoring the interaction to each participant personally, and allowing for qualitative analysis of user preferences. Simulation of an interaction-constrained learning task with different optimization techniques is performed to determine an appropriate learning approach for a handover task. The validity of the approach is demonstrated through experiments with participants on two handover tasks, with results indicating that the proposed method leads to seamless and pleasurable interactions.
Robotic avatar technology has the potential to impact the future of human connectivity, transporting the sense of a human’s presence to a remote location anywhere and in real-time. In this regard, the recent ANA Avatar XPRIZE challenge fostered the development of the ultimate generation of non-autonomous robotic avatar designs. This paper is devoted to introducing our system proposal, allowing intuitive motion-based control and multi-modal feedback. The teleoperated robot endorses an anthropomorphic upper body with dual arms and dexterous hands for fine manipulation of even small objects, as well as an omnidirectional platform for improved and safe locomotion. Special focus is pointed at the crucial challenge of providing self-body perception to the operator, in particular regarding control of the arms and rendering of haptic sensations. To this end, we propose an upper-limb exoskeleton and a teleoperation architecture allowing retargeting of the operator’s skills and receiving tactile and kinesthetic force feedback with realistic and informative perception of the own’s arms. Lastly, the platform was validated during laboratory and challenge missions mimicking several social, cooperative, and cultural scenarios from which we report on the lesson learned and future improvements.
31 December 2024 in IEEE Transactions on Neural Systems and Rehabilitation Engineering
The evaluation of hand function is of great importance to both clinical practice and biomedical research and is frequently evaluated by manual dexterity. Most of the assessment procedures evaluate the gross or the fine dexterity of the hand, but few of them are devoted to the assessment of both. We developed the Virtual Eggs Test (VET): it resembles the task of transporting fragile and robust objects, thus requiring both gross and fine dexterity. The test is composed of 11 Virtual Eggs that collapse if the grasping force exceeds their breaking thresholds, ranging from 0.4 N to 11.5 N. The test aims to transport each Virtual Egg over the barrier in the centre of the test platform without breaking it and as fast as possible. The metrics measured during the test are combined and provide two indexes that evaluate, respectively, gross and fine dexterity. We verify the concurrent validity and the construct validity of the VET with a target population of 30 trans-radial amputees wearing a myoelectric hand and the test-retest reliability on a control population of 35 healthy individuals. The results suggest the ability of the VET to assess hand function specifically in handling breakable objects, using both gross and fine dexterity over time. However, further research is needed to verify its correlation with other tests and the ability of amputees to perform activities of daily living.
Robotic literature widely addresses deformable object manipulation, but few studies analyzed human manipulation accounting for different levels of deformability and task properties. We asked participants to grasp and insert rigid and deformable objects into holes with varying tolerances and depths, and we analyzed the grasping behavior, the reaching velocity profile, and completion times. Results indicated that the more deformable the object is, the nearer the grasping point is to the extremity to be inserted. For insertions in the long hole, the selection of the grasping point is a trade-off between task accuracy and the number of re-grasps required to complete the insertion. The compliance of the deformable object facilitates the alignment between the object and the hole. The reaching velocity profile when increasing deformability recalls the one observed when task accuracy and precision decrease. Identifying human strategy allows the implementation of human-inspired high-level reasoning algorithms for robotic manipulation.