Hybrid Cognitive System for Smart Navigation
Traditional navigation systems and applications often face limitations due to their over-reliance on GPS signals and maps, as well as lack of reasoning, adaptive planning, and cognitive capabilities. Cognitive architectures such as Soar have aimed to develop more general intelligent autonomous agents by modelling human-inspired capabilities for diverse application domains including smart navigation. Despite the potential benefits of Soar architecture to address these previously mentioned challenges, there persists gaps for the improvement of its overall effectiveness including limited integration with algorithms and external systems, absence of real-time knowledge graph visualisation, memory reliance for knowledge integration and access, and lack of sophisticated approaches such as proactive tailored for specific challenges and domains, among others. This research introduces a hybrid cognitive architecture that extends Soar's cognitive capabilities accompanied with novel implementations, integrations, and techniques such as predictive, deliberative, and proactive demonstrated through a smart navigation use case proof of concept. Several research key contributions include extending Soar's capabilities through integrating specialised path planning algorithm with enhanced computation runtime, knowledge extension through external system integrations, and implementation of interactive and dynamic knowledge and matrix graphs, among many others. Testing and evaluation results provide validation and verification of the proof of concept, that is, cognitive autonomous agent’s performing navigation-related tasks within a simulated environment, with an extended overall knowledge base.
History
Qualification name
- PhD
Supervisor
Ah-Lian KorAwarding Institution
Leeds Beckett UniversityCompletion Date
2024-06-13Qualification level
- Doctoral
Language
- eng