• autonomous control
• dexterous grasp
• grasping trajectories models
• hand transport-based
• multi-articulated
• prosthetic hand
• reaching-to-grasp
• shared control

Nowadays prosthetic hands can be multi-articulated and thus capable of dexterous movements. However, enabling a reliable and repeatable control of these robotic devices is still an open challenge. An emerging human-machine interfacing approach is the so-called shared-control, by which the prosthesis is co-controlled by the user and the machine. In this setup, the user can provide basic input and then supervise (and correct if necessary) the robotic decisions, leaving to the artificial intelligence most of the control burden. In this work it was proposed and explored a shared-control approach for reaching-to-grasp actions. Here, the goal was to develop and test a new approach to autonomously generate grasping trajectories from the mere movement of the residual limb, by means of human-derived models. Offline testing of such trajectories prediction models appeared promising. Moreover, the models were further validated by a human-in-the-loop bench test with an actual prosthetic hand. Promising results were found also for the bench validation of these trajectory predictions models. While further validation is imperative, this work might have shed light on a rather simple and yet effective approach which aims to restore seamless object grasping in individuals with upper limb amputation, ultimately promoting full exploitation of modern prosthetic hands functionalities.