Human Initiated Grasp Space Exploration Algorithm for an Underactuated Robot Gripper Using Variational Autoencoder
This addresses the open issue of grasp planning in robotics, specifically for multifingered adaptive grippers, but is incremental as it builds on existing data-driven and analytic approaches.
The paper tackles the problem of grasp space exploration for an underactuated robot gripper by developing a method that uses a limited dataset of expert grasps, a grasp quality metric, and variational autoencoders to generate reliable grasps. It achieves a grasp success rate of 99.91% on 7000 trials in simulation for three objects.
Grasp planning and most specifically the grasp space exploration is still an open issue in robotics. This article presents an efficient procedure for exploring the grasp space of a multifingered adaptive gripper for generating reliable grasps given a known object pose. This procedure relies on a limited dataset of manually specified expert grasps, and use a mixed analytic and data-driven approach based on the use of a grasp quality metric and variational autoencoders. The performances of this method are assessed by generating grasps in simulation for three different objects. On this grasp planning task, this method reaches a grasp success rate of 99.91% on 7000 trials.