AKB-48: A Real-World Articulated Object Knowledge Base

Liu, Liu; Wang, Xu; Haoyuan, Fu,; Qian, Sucheng; Han, Yu; Lu, Cewu

doi:10.48550/arxiv.2202.08432

Cited by 2 publications

(2 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The knowledge base provides the semantic context for the robots' input and output in their tasks, including defining the meaning or function of the manipulated objects. The early robot knowledge bases primarily focused on static objects, such as RoboEarth [1,2], KnowRob [3,4], RoboBrain [5], and the recently developed articulated object knowledge base AKB-48 [6]. However, all the aforementioned knowledge bases for robots are large-scale static repositories primarily focused on describing stationary objects in robotic manipulation tasks.…”

Section: Knowledge Representation In Robotic Manipulationmentioning

confidence: 99%

Hierarchical Understanding in Robotic Manipulation: A Knowledge-Based Framework

Miao,

Jia,

Sun

et al. 2024

Actuators

View full text Add to dashboard Cite

In the quest for intelligent robots, it is essential to enable them to understand tasks beyond mere manipulation. Achieving this requires a robust parsing mode that can be used to understand human cognition and semantics. However, the existing methods for task and motion planning lack generalization and interpretability, while robotic knowledge bases primarily focus on static manipulation objects, neglecting the dynamic tasks and skills. To address these limitations, we present a knowledge-based framework for hierarchically understanding various factors and knowledge types in robotic manipulation. Using this framework as a foundation, we collect a knowledge graph dataset describing manipulation tasks from text datasets and an external knowledge base with the assistance of large language models and construct the knowledge base. The reasoning tasks of entity alignment and link prediction are accomplished using a graph embedding method. A robot in real-world environments can infer new task execution plans based on experience and knowledge, thereby achieving manipulation skill transfer.

show abstract

Section: Knowledge Representation In Robotic Manipulationmentioning

confidence: 99%

Hierarchical Understanding in Robotic Manipulation: A Knowledge-Based Framework

Miao,

Jia,

Sun

et al. 2024

Actuators

View full text Add to dashboard Cite

show abstract

“…More recently, there has been increasing interest in data-driven methods for studying articulated objects and estimating motion parameters [13,34]. To support these data-driven approaches, there has been concurrent development of datasets of annotated part articulations for synthetic [34,38,41] and reconstructed [15,18] 3D objects. predicts segmentation together with motion parameters, for 2.5D inputs [41,43], 3D point clouds [34] or for sequences of RGBD scans [12].…”

Section: Related Workmentioning

confidence: 99%

OPD: Single-view 3D Openable Part Detection

Jiang¹,

Yongsen²,

Savva³

et al. 2022

Preprint

View full text Add to dashboard Cite

Part-based cross-category generalizationSingle-view image bbox, mask, category axis origin Per part InputOutput detection type motion parametersFigure 1. We address the task of openable part detection (OPD). The input is a single view image, and the outputs are detected openable parts as well as their motion parameters. We design OPDRCNN: a neural architecture for this task that operates at the part level, allowing generalization across diverse object categories.

show abstract

AKB-48: A Real-World Articulated Object Knowledge Base

Abstract: Figure 1. AKB-48 consists of 2,037 articulated object models of 48 categories scanned from the real world. The objects are annotated with ArtiKG, and can support a full task spectrum from computer vision to robotics manipulation.

Cited by 2 publications

References 23 publications

Hierarchical Understanding in Robotic Manipulation: A Knowledge-Based Framework

Hierarchical Understanding in Robotic Manipulation: A Knowledge-Based Framework

OPD: Single-view 3D Openable Part Detection

Contact Info

Product

Resources

About