Spiking Neural Networks for Scalable Visual Place Recognition
Year: 2024
Type: dissertation
Abstract: This thesis investigates how Spiking Neural Networks, inspired by our brains, can serve as an alternative to current systems for Visual Place Recognition, a critical component in autonomous robotic navigation. This research introduces a spiking network tailored for place recognition that is scalable to large environments and integrates strategies that enhance robustness against appearance changes. Once deployed on specialised brain-inspired neuromorphic hardware, spiking networks promise high energy efficiency, paving the way for robots that are efficient and can operate autonomously over long periods, especially in scenarios with resource constraints.
Author: Somayeh Hussaini
Language: English
Subfield: Aerospace Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W4404665245