A5086900004- Robot Manipulation and Learning
- Advanced Vision and Imaging
- Image Enhancement Techniques
- Advanced Neural Network Applications
- Hand Gesture Recognition Systems
- Multimodal Machine Learning Applications
- Computational and Text Analysis Methods
- Optical measurement and interference techniques
- Robotics and Sensor-Based Localization
- Image Processing and 3D Reconstruction
- Neural Networks and Applications
- Remote Sensing and LiDAR Applications
- Machine Learning and Data Classification
- Computer Graphics and Visualization Techniques
- Robotic Path Planning Algorithms
- Data Visualization and Analytics
Technical University of Munich
2021-2025
Inferring geometrically consistent dense 3D scenes across a tuple of temporally consecutive images remains challenging for self-supervised monocular depth prediction pipelines. This paper explores how the increasingly popular transformer architecture, together with novel regularized loss formulations, can improve consistency while preserving accuracy. We propose spatial attention module that correlates coarse predictions to aggregate local geometric information. A temporal mechanism further...
The progressive prevalence of robots in human-suited environments has given rise to a myriad object manipulation techniques, which dexterity plays paramount role. It is well-established that humans exhibit extraordinary when handling objects. Such seems derive from robust understanding properties (such as weight, size, and shape), well remarkable capacity interact with them. Hand postures commonly demonstrate the influence specific regions on objects need be grasped, especially are partially...
We present FuncGrasp, a framework that can infer dense yet reliable grasp configurations for unseen objects using one annotated object and single-view RGB-D observation via categorical priors. Unlike previous works only transfer set of poses, FuncGrasp aims to infinite parameterized by an object-centric continuous function across varying instances. To ease the process, we propose Neural Surface Grasping Fields (NSGF), effective neural representation defined on surface densely encode...
Forestry constitutes a key element for sustainable future, while it is supremely challenging to introduce digital processes improve efficiency. The main limitation the difficulty of obtaining accurate maps at high temporal and spatial resolution as basis informed forestry decision-making, due vast area forests extend over sheer number trees. To address this challenge, we present an autonomous Micro Aerial Vehicle (MAV) system which purely relies on cost-effective light-weight passive visual...
The progressive prevalence of robots in human-suited environments has given rise to a myriad object manipulation techniques, which dexterity plays paramount role. It is well-established that humans exhibit extraordinary when handling objects. Such seems derive from robust understanding properties (such as weight, size, and shape), well remarkable capacity interact with them. Hand postures commonly demonstrate the influence specific regions on objects need be grasped, especially are partially...