Date of Award

2024

Document Type

Open Access Master's Thesis

Degree Name

Master of Science in Electrical and Computer Engineering (MS)

Administrative Home Department

Department of Electrical and Computer Engineering

Advisor 1

Hongyu An

Committee Member 1

Tan Chen

Committee Member 2

Yan Zhang

Abstract

Deep Neural Networks (DNNs) have come a long way in many cognitive tasks by training on large, labeled datasets. However, this method has problems in places with limited data and energy, like when planetary robots are used or when edge computing is used [1]. In contrast to this data-heavy approach, animals demonstrate an innate ability to learn by communicating with their environment and forming associative memories among events and entities, a process known as associative learning [2-4]. For instance, rats in a T-maze learn to associate different stimuli with outcomes through exploration without needing labeled data [5]. This learning paradigm is crucial to overcoming the challenges of deep learning in environments where data and energy are limited. Taking inspiration from this natural learning process, recent advancements [6, 7] have been made in implementing associative learning in artificial systems. This work introduces a pioneering approach by integrating associative learning utilizing an Unmanned Ground Vehicle (UGV) in conjunction with neuromorphic hardware, specifically the XyloA2TestBoard from SynSense, to facilitate online learning scenarios. The system simulates standard associative learning, like the spatial and memory learning observed in rats in a T-maze environment, without any pretraining or labeled datasets. The UGV, akin to the rats in a T-maze, autonomously learns the cause-and-effect relationships between different stimuli, such as visual cues and vibration or audio and visual cues, and demonstrates learned responses through movement. The neuromorphic robot in this system, equipped with SynSense’s neuromorphic chip, processes audio signals with a specialized Spiking Neural Network (SNN) and neural assembly, employing the Hebbian learning rule to adjust synaptic weights throughout the learning period. The XyloA2TestBoard uses little power (17.96 µW on average for logic Analog Front End (AFE) and 213.94 µW for IO circuitry), which shows that neuromorphic chips could work well in places with limited energy, offering a promising direction for advancing associative learning in artificial systems.

Creative Commons License

Creative Commons Attribution 4.0 License
This work is licensed under a Creative Commons Attribution 4.0 License.

Share

COinS