Using Multi-Relational Embeddings as Knowledge Graph Representations for Robotics Applications
Author(s)
Daruna, Angel Andres
Advisor(s)
Editor(s)
Collections
Supplementary to:
Permanent Link
Abstract
User demonstrations of robot tasks in everyday environments, such as households, can be brittle due in part to the dynamic, diverse, and complex properties of those environments. Humans can find solutions in ambiguous or unfamiliar situations by using a wealth of common-sense knowledge about their domains to make informed generalizations. For example, likely locations for food in a novel household. Prior work has shown that robots can benefit from reasoning about this type of semantic knowledge, which can be modeled as a knowledge graph of interrelated facts that define whether a relationship exists between two entities. Semantic reasoning about domain knowledge using knowledge graph representations has improved the robustness and usability of end user robots by enabling more fault tolerant task execution. Knowledge graph representations define the underlying representation of facts, how facts are organized, and implement semantic reasoning by defining the possible computations over facts (e.g. association, fact-prediction).
This thesis examines the use of multi-relational embeddings as knowledge graph representations within the context of robust task execution and develops methods to explain the inferences of and sequentially train multi-relational embeddings. This thesis contributes: (i) a survey of knowledge graph representations that model semantic domain knowledge in robotics, (ii) the development and evaluation of our knowledge graph representation based on multi-relational embeddings, (iii) the integration of our knowledge graph representation into a robot architecture to improve robust task execution, (iv) the development and evaluation of methods to sequentially update multi-relational embeddings, and (v) the development and evaluation of an inference reconciliation framework for multi-relational embeddings.
Sponsor
Date
2022-05-19
Extent
Resource Type
Text
Resource Subtype
Dissertation