Dexin Wang , Faliang Chang, Chunsheng Liu, Hengqiang Huan, Nanjun Li, Rurui Yang
School of Control Science and Engineering, Shandong University
Links
Paper: PLGP-Dataset
PLGP-Dataset: README
(Please copy the following link and open it in a new page. Clicking "Open link in new tab" may fail.)
Whole (Baidu Netdisk): part1, part2, part3, part4, part5, part6, part7, part8
Grasp (Baidu Netdisk): test, train_part1, train_part2, train_part3, train_part4
Code (github): ODG-Genaration; AFFGA-NET-PLGP; AFFGA_binpicking_ur5e
Abstract
Grasp detection in cluttered scenes is a very challenging task for robots. Generating synthetic grasping data is a popular way to train and test grasp methods, as is Dex-Net; yet, these methods sample training grasps on 3D synthetic object models, but evaluate at images or point clouds with different sample distributions, which reduces performance due to covariate shift and sparse grasp labels. To solve existing problems, we propose a novel on-policy grasp detection method for parallel grippers, which can train and test on the approximate distribution with dense pixel-level grasp labels generated on RGB-D images. An Orthographic-Depth Grasp Generation (ODG-Generation) method is proposed to generate an orthographic depth image through a new imaging model of projecting points in orthographic; then this method generates multiple candidate grasps for each pixel and obtains robust positive grasps through flatness detection, force-closure metric and collision detection. Then, a comprehensive Pixel-Level Grasp Pose Dataset (PLGP-Dataset) is constructed, which is the first pixel-level grasp dataset, with the on-policy distribution. Lastly, we build a grasp detection network with a novel data augmentation process for imbalance training. Experiments show that our on-policy method can largely overcome the gap between simulation and reality, and achieves the best performance.
1. Approach
2. PLGP-Dataset
3. Experiments
3.1 Augmentation and Loss
3.2 Experiments of Sparse Label
3.3 On-policy VS Off-policy
3.4 Experiments of networks
3.5 Compared with 6-DOF grasping
3.5.1 AFFGA-Net (ours) in single object scenarios.
Screwdriver
Toy (blue)
Toy (Yellow)
Stone (White)
Stone (Black)
U-disk
Tape
Fastenings
Conduit
Stapler
Wire
3.2 AFFGA-Net (ours) in clutter scenarios.
3.3 GraspNet in single object scenarios.
Screwdriver
Toy (blue)
Toy (yellow)
Stone (White)
Stone (Black)
U-disk
Tape
Fastenings
Conduit
Stapler
Wire
3.4 GraspNet in clutterscenarios.
Contact
DexinWang: dexinwang@mail.sdu.edu.cn