Fast and Accurate Category-Level Object Pose Estimation Without Shape Priors for Robotic Grasp Detection

Category-level object pose estimation is crucial for enabling robot grasping. Currently, many methods rely on 3D shape priors for pose estimation, but obtaining priors for specific categories often requires a significant amount of time to generate. Although some methods do not rely on priors, However, these methods struggle to achieve a balance between speed and accuracy. Achieving fast and accurate category-level object pose estimation remains a challenging issue. In this paper, we propose an algorithm called FAPose, which aims to simultaneously achieve speed and accuracy in category-level object pose estimation without any shape priors. Firstly, we design an RGB-point cloud feature aggregation method based on transformers to fuse RGB and point cloud features. Secondly, we develop a dual-constraint object pose estimation method to effectively leverage both feature space and geometric space information. This approach constructs pose constraints in both feature space and geometric space to achieve optimal pose prediction. Finally, we validate FAPose through benchmark datasets and real robot grasping experiments. The experimental results demonstrate that our proposed method surpasses most existing state-of-the-art pose estimation methods, achieving superior pose estimation performance. The code for FAPose will be available after the paper is accepted for publication. Note to Practitioners—Robotic grasp detection based on category-level 6D object pose estimation is an important way to realize reliable robotic grasp. How to achieve fast and accurate object pose estimation is important for the real-world application. Currently, there are many category-level object pose estimation methods, which can be divided into two categories: methods with priors and methods without priors. Methods with priors rely on object prior models, but they require a significant amount of time to generate, which is not ideal for practical applications. As for methods without priors, although they do not rely on object priors, they struggle to achieve a balance between real-time performance and accuracy. To solve these problems, in this paper, we try to develop a fast and accurate category-level object pose estimation network and apply it to the robotic grasp tasks in the real world. Experimental results show that our proposed method outperforms most realated methods on benchmark datasets. What’s more, the proposed network consumes less time in the pose estimation process, and achieves a fast and accurate pose estimation in the real world.