Dexterous reaching, grasp transfer and planning using electrostatic representations

Abstract

Abstract-We propose a novel approach to transfer reach and grasp movements while being agnostic and invariant to finger kinematics, hand configurations and relative changes in object dimensions. We exploit a novel representation based on electrostatics to parametrise the salient aspects of the demonstrated grasp. By working in this alternate space that focuses on the relational aspects of the grasp rather than absolute kinematics, we are able to use inference based planning techniques to couple the motion in abstract spaces with trajectories in the configuration space of the robot. We demonstrate that our method computes stable grasps that generalise over objects of different shapes and robots of dissimilar kinematics while retaining the qualitative grasp type -all without expensive collision detection or re-optimisation.

Egocentric Mapping of Body Surface Constraints

“…Recently, a spatial parameterization based on electric coordinates has been proposed [29]. Although this parameterization can be used to define grasp postures and to robustly map them to a variety of target hands and objects [30], the computation of the electric coordinates is not fast enough for interactive applications.…”

Space-time area coverage control for robot motion synthesis

“…The representations discussed here have been also applied to solve robotics and computer animation problems, such as path planning with winding constraints using the winding numbers [9], character motion control preserving spatial relationships using the Gauss Linking Integral [10], and capturing the relationship between two objects by parametrizing the space around one of these objects using the analogy of the electric field and computing position within this field [4] and its integral through the surface of the second object [3]. These methods capture the relationship between objects by describing how much one object wraps around another while ignoring the finer geometric detail.…”

“…We then track the position of these points over time and create a virtual surface constructed through triangulation. Our previous work [4] has shown that the electric flux space has very few local minima, which makes it a suitable task space for local optimisation methods. To synthesise robot motion, we use the Approximate Inference Control, a local trajectory optimisation 978-1-4673-7509-2/15/$31.00 ©2015 Crown method, proposed in [5], which we use to optimise the shape of the virtual surface to achieve a prescribed amount of electric flux through it.…”