Enhancing Robotic Tactile Perception through Biomimetic Approaches : From Neuron Modeling to Spiking Neural Networks

목차 (Table of Contents)

• List of Figures Ⅳ
• List of Tables Ⅵ
• Nomenclature Ⅶ
• Chapter 1 General Introduction 1
• 1.1 Background 2
• 1.1.1 Motivation 2
• 1.1.2 Introduction to biomimetic approaches in tactile sensing 3
• 1.1.3 Introduction to Spiking Neural Network 3
• 1.2 Objective and specific aims of each chapter 5
• Chapter 2 Modeling Long-term Spike Frequency Adaptation in SA-I Afferent Neurons Using an Izhikevich-based Biological Neuron Model 7
• 2.1 Introduction 8
• 2.2 Materials and methods 10
• 2.2.1 Overview of animal experiments 10
• 2.2.2 Animals and surgical procedure 10
• 2.2.3 Setup for spike data acquisition 10
• 2.2.4 Ex-vivo single fiber recordings 10
• 2.2.5 Identification of single A-fibers 11
• 2.2.6 Mechanical stimulator 11
• 2.2.7 Modeling SA-I firings using interspike intervals 12
• 2.2.8 Izhikevich biological neuron model 12
• 2.2.9 Long-term adapting Izhikevich model 15
• 2.2.10 Evaluating the similarity between model and biological spike trains 16
• 2.3 Results 17
• 2.3.1 Firing patterns of SA-I afferent 17
• 2.3.2 Simulation of the vanilla Izhikevich model 20
• 2.3.3 Simulation of the vanilla Izhikevich model 20
• 2.3.4 Simulating the dynamics of long-term spike frequency adaptation in SA-I neurons .. 22
• 2.3.5 Optimization of parameters for long-term spike frequency adaptation Izhikevich neuron model 24
• 2.3.6 Performance of the long-term spike frequency adaptation Izhikevich. 25
• 2.4 Discussion 27
• Chapter 3 Biomimetic Spiking Neural Network Model for Multi-dimensional Tactile Perception 29
• 3.1 Introduction 30
• 3.2 Materials and methods 33
• 3.2.1 Multi-layered SNN architecture 33
• 3.2.2 First layer: SA-1 and RA-1 afferent neurons 33
• 3.2.3 Second layer: cuneate nucleus simulation 34
• 3.2.4 Third layer: emulation of the S1 cortex 36
• 3.2.5 Biological neuron models implementation 37
• 3.2.6 Experiment design and tactile stimuli 40
• 3.3 Results 44
• 3.3.1 Spike responses through layers 44
• 3.3.2 Role of SA and RA in static and dynamic stimulation 46
• 3.3.3 One-dimension tactile stimulus information processing 47
• 3.3.4 Multi-dimension tactile stimulus information processing 51
• 3.4 Discussion 54
• Chapter 4 Advancing Robotic Tactile Perception: Adaptive Tactile Perception 56
• 4.1 Introduction 57
• 4.2 Materials and methods 60
• 4.2.1 Experimental setup for tactile robotic system 60
• 4.2.2 Experimental setup for tactile stimuli 61
• 4.2.3 Tactile data acquisition 63
• 4.2.4 Data preparation and classifier training 64
• 4.2.5 Adaptive Tactile Perception and Decision-Making System 65
• 4.3 Results 68
• 4.3.1 Object Classification Accuracy for Single Actions 68
• 4.3.2 Object Classification Accuracy for Multiple Actions 71
• 4.3.3 Evaluation of Adaptive Decision-Making Strategy 73
• 4.4 Discussion 75
• Chapter 5 Concluding remarks 77
• 5.1 Main findings 78
• 5.1.1 Transformation of tactile stimuli into human-like spike patterns: Potential of biomimetic approaches 79
• 5.1.2 Processing tactile information using spiking neural networks: Suitability for real-time applications 80
• 5.1.3 Adaptive strategy selection for object recognition: Enhancing classification accuracy with prior tactile knowledge 81
• 5.1.4 Contributions of current dissertation 82
• 5.2 Limitations 83
• 5.3 Implications 84
• REFERENCES 85