Robot-Safe Impacts with Soft Contacts Based on Learned Deformations

Abstract

Safely generating impacts with robots is challenging due to subsequent discontinuous velocity and high impact forces. We aim at increasing the impact velocity -the robot's relative speed prior to contact -such that impact-tasks like grabbing and boxing are made with the highest allowable speed performance when needed. Previous works addressed this problem for rigid bodies' impacts. This letter proposes a control paradigm for generating intentional impacts with deformable contacts that incorporates hardware and task constraints. Based on data-driven learning of the shock-absorbing soft dynamics and a novel mapping of joint-space limits to contact-space, we devise a constrained model-predictive control to maximize the intentional impact within a feasible, robot-safe level. Our approach is assessed with real-robot experiments on the redundant Panda manipulator, demonstrating high pre-impact velocities (up to 0.9 m/s) of a rigid end-effector on soft objects and an end-effector soft suction-pump on rigid or deformable objects.

Dual-Arm Box Grabbing With Impact-Aware MPC Utilizing Soft Deformable End-Effector Pads

“…Basically, there are no equations of motion for the rigid environment. We obtain the single-arm model described in [9], with…”

“…The prediction of impact-induced state jumps caused by rigid contacts is formulated in [23] to limit the end-effector speed prior to making contact. For impacting soft contacts with a manipulator, we inferred in [9] the configuration-dependent maximum feasible impact velocity. This upper velocity bound can be tracked with the robot arm before making contact, guaranteeing the robot's hardware limits during the following deformation phase.…”

“…Let us elaborate further the mapping of the manipulator's structural hardware limits onto the contact-space, initially described in [9] and inspired by [25], [26]. This configurationdependent mapping is not trivial when dealing with redundant robots.…”

“…In [9], we proposed the bounded deformation dynamics model for poking soft material with a single robot, cf. Fig.…”

“…Our main contribution is the derivation of bounded deformation dynamics for a dual-arm system with deformable end-effector pads. These constrained deformation dynamics provide the basis for a model predictive control (MPC) scheme [6], [7], [8], generalized from single-arm poking [9] to dual-arm grabbing. The MPC optimizes the deformation trajectory, which is to be initiated with the highest possible end-effector velocities.…”