Abstract
This paper presents a novel approach to implement hierarchical, dense and dynamic reconstruction of 3D objects based on the VDB (Variational Dynamic B + Trees) data structure for robotic applications. The scene reconstruction is done by the integration of depth-images using the Truncated Signed Distance Field (TSDF). The proposed reconstruction method is based on dynamic trees in order to provide similar reconstruction results to the current state-of-the-art methods (i.e., complete volumes, hashing voxels and hierarchical volumes) in terms of execution time but with a direct multi-level representation that remains real-time. This representation provides two major advantages: it is a hierarchical and unbounded space representation. The proposed method is optimally implemented to be used on a GPU architecture, exploiting the parallelism skills of this hardware. A series of experiments will be presented to prove the performance of this approach in a robot arm platform.
Highlights
Industrial robotic research has been extremely prolific in the last decades, with a special interest in applications such as welding, painting and pick-and-place of objects
A novel dense and dynamic 3D reconstruction method has been implemented based on a hierarchical database structure (GPU oriented) for integrating depth images by truncated signed distance field theory
Future directions in our research explore the use of this method to simulate material dynamics in situ, taking advantage of the GPU-optimized VDB data structure
Summary
Industrial robotic research has been extremely prolific in the last decades, with a special interest in applications such as welding, painting and pick-and-place of objects. When robots need to manipulate deformable objects, current reconstruction methods fail since they are based on the assumption of the presence of rigid objects in static scenarios (Zeng et al (2013), Whelan et al (2016) and Puri et al (2017)). Another well-known problem is drifting in textureless scenarios during the camera pose estimation, which implies erroneous reconstructions. We propose to use a new generation consumer depth camera (such as the Intel RealSense D435) installed on the robot so that they can output live half-HD depth maps at high-frequency rates with a low price to implement a precise reconstruction of the objects to be manipulated (Figure 1)
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
