Haptic Shared-Control Methods for Robotic Cutting

Shared control allows to share the available degrees of freedom of a robotic system between the operator and an autonomous controller, to facilitate the task and improve the efficiency of the system [1], [2], e.g., in robotic teleoperation [3]–[5]. Robotic cutting is particularly interesting for shared control, as it is employed in various sensitive applications which range from surgical cutting [6] to nuclear decommissioning [7] and disaster response [8]. Moreover, cutting applications feature a variety of constraints which can have a high impact on the task. For example, to avoid damaging the environment, the cutting tool should neither perform pure lateral motion nor rotate in place. Indeed, unicycle and car-like kinematic models have been used for modeling the cutting task to reflect its nonholonomic nature [9], [10]. Several shared-control architectures have been proposed in the literature to tackle different cutting applications [11]. One example is enforcing nonholonomic constraints on multipurpose robots through control [12], [13]. In [13], Vozar et al. design four shared-control approaches for cutting straight lines into MLI blankets under a time delay. In all modes, the master interface was free to move in any direction, and the constraints were implemented only at the remote side. This work targets the limitations of the above-described architectures. It presents the design and evaluation of two shared-control approaches for commanding a manipulator in a cutting scenario while enforcing constraints associated with the task itself, e.g., limiting lateral motions, rotations in place, and sharp turns of the tool. The user is provided with information about the enforced nonholonomic constraints (alongside contact forces) via haptic feedback on the master device.