Stable Robotic Grasping of Multiple Objects using Deep Neural Networks

Abstract

SUMMARY Optimal grasping points for a robotic gripper were derived, based on object and hand geometry, using deep neural networks (DNNs). The optimal grasping cost functions were derived using probability density functions for each local cost function of the normal distribution. Using the DNN, the optimum height and width were set for the robot hand to grasp objects, whose geometric and mass centre points were also considered in obtaining the optimum grasping positions for the robot fingers and the object. The proposed algorithm was tested on 10 differently shaped objects and showed improved grip performance compared to conventional methods.

Study of grasp-energy based optimal distribution of contact forces on a humanoid robotic hand during object grasp

“…Müller et al [25] describe several approaches for the design of a soft robotic gripper including selection of the appropriate model for characterization of the maximum payload on the gripper, material characterization, and simulation of grasp to determine its performance at various loads. Kim et al [26] developed a Deep-Neural-Network-based algorithm to identify optimal grasping points based on an object's geometric features for obtaining a stable grasp. Researchers from the University of Brazil developed an anthropometric robotic hand gripper, the UnB-Hand [27], which was designed with bio-inspired optimization algorithms, and was proven to be able to successfully perform grasps according to the Cutkosky grasping taxonomy, that is, power and precision grasps necessary for machining operations.…”

Robot Grasping System and Grasp Stability Prediction Based on Flexible Tactile Sensor Array

“…Yao [28] concentrates a on grasp configuration planning method based on a three-fingered robot hand, which depends on the grasping data described with the configuration of fingers and the attribute of the object. Kim [29] trained a deep neural network based on information of object and robot hand geometryand obtained the optimal grasping points between object and fingers. The geometry and attributes of objects and fingers are the key to achieve multi-fingered form-closure grasping on multiple kinds of objects, such as a polygon-type form-closure grasping achieved with a four-fingers parallel-jaw gripper [30].…”

Assist system for remote manipulation of electric drills by the robot “WAREC-1R” using deep reinforcement learning

“…Thus, grasp (a) with success rate high enough for application and (b) is easy to operate with less burden for operators are highly needed for remote operation of WAREC-1R and its application in tool operation. As the solution, in this paper, we propose an assist system using Deep Reinforcement Learning (DRL in short for the following contents), a subtype of reinforcement learning [6,7,8,9,10] for both reaching and grasping process before the end effector of WAREC-1R holds the target tool. The contribution and originality of this paper focus on the following points:…”