Seamless Integration of Knowledge Acquisition for Autonomous Systems by Domain Users with Prudence Capability
Abstract
Knowledge-based systems are typically constrained by their ability to acquire new knowledge, thus limiting their applicability to autonomous systems. This work developed an extensive, easily maintainable hierarchical Knowledge-base System (KBS) for Autonomous Systems (AS) technologies trained by Knowledge Domain Experts (KDE) using a Natural language (NL) interface for communication. The system implements an abstracted architecture, taking a layer-based approach to separate data and hardware, information, and services, each with an associated, contextual knowledge base. The developed process, Contextual MCRDR, improves upon classical Multiple Classification Ripple Down Rules (MCRDR), with constrained natural language conversation systems associated with querying of in-situ databases of pre-existing information. This was then expanded to support Automatic Speech Recognition (ASR). Finally, the work was extended to a semi-autonomous system (Robotis Turtlebot3). The full effort produced three published journal/conference papers, and two additional papers in the review process.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 12, 2018
- Accession Number
- AD1071250
Entities
People
- Byeong Ho Kang
- Caren Han
- David Herbert
Organizations
- University of Tasmania