The field of 3D perception and robotics is rapidly advancing, with a focus on developing more effective and efficient methods for understanding and interacting with complex environments. One of the key directions in this area is the integration of deep learning techniques with traditional computer vision and robotics approaches. This has led to significant improvements in tasks such as 3D object recognition, scene understanding, and robotic manipulation. Notable papers in this area include OTAS, which proposes a novel method for open-vocabulary token alignment for outdoor segmentation, and 3DGAA, which introduces a realistic and robust 3D Gaussian-based adversarial attack for autonomous driving. Other noteworthy papers include SegVec3D, which presents a method for vector embedding of 3D objects oriented towards robot manipulation, and rt-RISeg, which proposes a real-time interactive perception framework for active instance-level object understanding. These papers demonstrate the innovative and advancing nature of the field, with a focus on developing practical and effective solutions for real-world problems.
Advances in 3D Perception and Robotics
Sources
OTAS: Open-vocabulary Token Alignment for Outdoor Segmentation
Learning and Transferring Better with Depth Information in Visual Reinforcement Learning
Stereo-based 3D Anomaly Object Detection for Autonomous Driving: A New Dataset and Baseline
Geometric Generative Modeling with Noise-Conditioned Graph Networks
SegVec3D: A Method for Vector Embedding of 3D Objects Oriented Towards Robot manipulation
Hierarchical Abstraction Enables Human-Like 3D Object Recognition in Deep Learning Models
3DGAA: Realistic and Robust 3D Gaussian-based Adversarial Attack for Autonomous Driving
Tool-to-Tool Matching Analysis Based Difference Score Computation Methods for Semiconductor Manufacturing
Spatial Reasoners for Continuous Variables in Any Domain
rt-RISeg: Real-Time Model-Free Robot Interactive Segmentation for Active Instance-Level Object Understanding
Subgraph Generation for Generalizing on Out-of-Distribution Links
MS-DGCNN++: A Multi-Scale Fusion Dynamic Graph Neural Network with Biological Knowledge Integration for LiDAR Tree Species Classification
A Spectral Interpretation of Redundancy in a Graph Reservoir
3DKeyAD: High-Resolution 3D Point Cloud Anomaly Detection via Keypoint-Guided Point Clustering