Enhancing 6-DoF Object Pose Estimation through Multiple Modality Fusion: A Hybrid CNN Architecture with Cross-Layer and Cross-Modal Integration
Author:
Wang Zihang1ORCID, Sun Xueying12ORCID, Wei Hao3, Ma Qing1, Zhang Qiang12ORCID
Affiliation:
1. College of Automation, Jiangsu University of Science and Technology, No. 666 Changhui Road, Zhenjiang 212100, China 2. Systems Science Laboratory, Jiangsu University of Science and Technology, No. 666 Changhui Road, Zhenjiang 212100, China 3. Shenlan College, Jiangsu University of Science and Technology, No. 666 Changhui Road, Zhenjiang 212100, China
Abstract
Recently, applying the utilization of RGB-D data for robot perception tasks has garnered significant attention in domains like robotics and autonomous driving. However, a prominent challenge in this field lies in the substantial impact of feature robustness on both segmentation and pose estimation tasks. To tackle this challenge, we proposed a pioneering two-stage hybrid Convolutional Neural Network (CNN) architecture, which connects segmentation and pose estimation in tandem. Specifically, we developed Cross-Modal (CM) and Cross-Layer (CL) modules to exploit the complementary information from RGB and depth modalities, as well as the hierarchical features from diverse layers of the network. The CM and CL integration strategy significantly enhanced the segmentation accuracy by effectively capturing spatial and contextual information. Furthermore, we introduced the Convolutional Block Attention Module (CBAM), which dynamically recalibrated the feature maps, enabling the network to focus on informative regions and channels, thereby enhancing the overall performance of the pose estimation task. We conducted extensive experiments on benchmark datasets to evaluate the proposed method and achieved exceptional target pose estimation results, with an average accuracy of 94.5% using the ADD-S AUC metric and 97.6% of ADD-S smaller than 2 cm. These results demonstrate the superior performance of our proposed method.
Funder
National Natural Science Foundation of China Jiangsu Province Industry University Research Cooperation Project: Research on key technology of robot navigation and following in human–robot integration environment
Subject
Electrical and Electronic Engineering,Industrial and Manufacturing Engineering,Control and Optimization,Mechanical Engineering,Computer Science (miscellaneous),Control and Systems Engineering
Reference68 articles.
1. Huang, X., Mei, G., Zhang, J., and Abbas, R. (2021). A Comprehensive Survey on Point Cloud Registration. arXiv. 2. Zhu, Y., Li, M., Yao, W., and Chen, C. (2022, January 17–19). A Review of 6D Object Pose Estimation. Proceedings of the 2022 IEEE 10th Joint International Information Technology and Artificial Intelligence Conference (ITAIC), Chongqing, China. 3. Vision-Based Robotic Grasping from Object Localization, Object Pose Estimation to Grasp Estimation for Parallel Grippers: A Review;Du;Artif. Intell. Rev.,2021 4. T6D-Direct: Transformers for Multi-Object 6D Pose Direct Regression;Amini;DAGM German Conference on Pattern Recognition,2021 5. He, Y., Huang, H., Fan, H., Chen, Q., and Sun, J. (2021, January 20–25). FFB6D: A Full Flow Bidirectional Fusion Network for 6D Pose Estimation. Proceedings of the 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Nashville, TN, USA.
|
|