Fast Convolutional Neural Network for Real-Time Robotic Grasp Detection

Abstract

The development of the robotics field has not yet allowed robots to execute, with dexterity, simple actions performed by humans. One of them is the grasping of objects by robotic manipulators. Aiming to explore the use of deep learning algorithms, specifically Convolutional Neural Networks (CNN), to approach the robotic grasping problem, this work addresses the visual perception phase involved in the task. To achieve this goal, the dataset “Cornell Grasp” was used to train a CNN capable of predicting the most suitable place to grasp the object. It does this by obtaining a grasping rectangle that symbolizes the position, orientation, and opening of the robot's parallel grippers just before the grippers are closed. The proposed system works in real-time due to the small number of network parameters. This is possible by means of the data augmentation strategy used. The efficiency of the detection is in accordance with the state of the art and the speed of prediction, to the best of our knowledge, is the highest in the literature.