Introducing Memory and Association Mechanism Into a Biologically Inspired Visual Model

Qiao, Hong; Li, Yinlin; Tang, Tang; Wang, Peng

doi:10.1109/tcyb.2013.2287014

Cited by 46 publications

(4 citation statements)

References 59 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The retained image information after early processing is eventually transmitted to the visual cortex through the optic nerve to achieve further processing and understanding of the sensed image information. With the early processing occurring in the retina, the redundant information irrelevant to the image can be discarded and consequently the understanding of the image is accelerated in the visual cortex, which has inspired a hierarchical model of object recognition that has been widely used in computer vision [ 41 , 42 ]. By closely mimicking the HVS, we propose a neuromorphic vision system composed of a retinomorphic sensor and a memristive network, as schematically shown in Fig.…”

Section: Resultsmentioning

confidence: 99%

Networking retinomorphic sensor with memristive crossbar for brain-inspired visual perception

Wang

et al. 2020

National Science Review

104

View full text Add to dashboard Cite

Comparing to human vision, conventional machine vision composed of image sensor and processor suffers from high latency and large power consumption due to physically separated image sensing and processing. Neuromorphic vision system with brain-inspired visual perception provides a promising solution to solve the challenge. Here we propose and demonstrate a prototype neuromorphic vision system by networking retinomorphic sensor with a memristive crossbar. We fabricate the retinomorphic sensor by using WSe2/h-BN/Al2O3 van der Waals heterostructures with gate-tunable photoresponses, to closely mimic the human retinal capabilities in simultaneously sensing and processing images. We then network such sensor with a large-scale Pt/Ta/HfO2/Ta one-transistor-one-memristor (1T1R) memristive crossbar, which serves as the role similar to the visual cortex in human brain. The realized neuromorphic vision system allows for fast letter recognition and object tracking, indicating the capabilities of image sensing, processing and recognition in the full analog regime. Our work suggests that such neuromorphic vision system may open up unprecedented opportunities in future visual perception applications.

show abstract

Section: Resultsmentioning

confidence: 99%

Networking retinomorphic sensor with memristive crossbar for brain-inspired visual perception

Wang

et al. 2020

National Science Review

104

View full text Add to dashboard Cite

show abstract

“…Therefore, similarly, networks of this type are usually adopted for low-level sensory acquisition in robotic systems, such as vision (Perrinet et al, 2004 ), tactile sensing (Rochel et al, 2002 ), and olfaction (Cassidy and Ekanayake, 2006 ). For example, inspired by the structures and principles of primate visual cortex, Qiao et al ( 2014 , 2015 , 2016 ) enhanced the feed-forward models including Hierarchical Max Pooling (HAMX) model and Convolutional Deep Belief Network (CDBN) with memory, association, active adjustment, semantic and episodic feature learning ability etc., and achieved good results in visual recognition task.…”

Section: Modeling Of Spiking Neural Networkmentioning

confidence: 99%

A Survey of Robotics Control Based on Learning-Inspired Spiking Neural Networks

et al. 2018

View full text Add to dashboard Cite

Biological intelligence processes information using impulses or spikes, which makes those living creatures able to perceive and act in the real world exceptionally well and outperform state-of-the-art robots in almost every aspect of life. To make up the deficit, emerging hardware technologies and software knowledge in the fields of neuroscience, electronics, and computer science have made it possible to design biologically realistic robots controlled by spiking neural networks (SNNs), inspired by the mechanism of brains. However, a comprehensive review on controlling robots based on SNNs is still missing. In this paper, we survey the developments of the past decade in the field of spiking neural networks for control tasks, with particular focus on the fast emerging robotics-related applications. We first highlight the primary impetuses of SNN-based robotics tasks in terms of speed, energy efficiency, and computation capabilities. We then classify those SNN-based robotic applications according to different learning rules and explicate those learning rules with their corresponding robotic applications. We also briefly present some existing platforms that offer an interaction between SNNs and robotics simulations for exploration and exploitation. Finally, we conclude our survey with a forecast of future challenges and some associated potential research topics in terms of controlling robots based on SNNs.

show abstract

“…Compliance is one special characteristic of the robotic system during interaction tasks (Huang et al , 2020; Yu et al , 2019). On the one hand, compliance mechanism is discussed and applied to a robotic system to adapt to the contact force generated by the human or the environment (Li et al , 2015; Zhong et al , 2020; Qiao et al , 2014). On the other hand, control algorithms are widely researched to achieve compliant manipulation.…”

Section: Introductionmentioning

confidence: 99%

Optimal grasp force for robotic grasping and in-hand manipulation with impedance control

Chen

2021

View full text Add to dashboard Cite

Purpose The purpose of this paper is to achieve stable grasping and dexterous in-hand manipulation, the control of the multi-fingered robotic hand is a difficult problem as the hand has many degrees of freedom with various grasp configurations. Design/methodology/approach To achieve this goal, a novel object-level impedance control framework with optimized grasp force and grasp quality is proposed for multi-fingered robotic hand grasping and in-hand manipulation. The minimal grasp force optimization aims to achieve stable grasping satisfying friction cone constraint while keeping appropriate contact forces without damage to the object. With the optimized grasp quality function, optimal grasp quality can be obtained by dynamically sliding on the object from initial grasp configuration to final grasp configuration. By the proposed controller, the in-hand manipulation of the grasped object can be achieved with compliance to the environment force. The control performance of the closed-loop robotic system is guaranteed by appropriately choosing the design parameters as proved by a Lyapunove function. Findings Simulations are conducted to validate the efficiency and performance of the proposed controller with a three-fingered robotic hand. Originality/value This paper presents a method for robotic optimal grasping and in-hand manipulation with a compliant controller. It may inspire other related researchers and has great potential for practical usage in a widespread of robot applications.

show abstract

Introducing Memory and Association Mechanism Into a Biologically Inspired Visual Model

Cited by 46 publications

References 59 publications

Networking retinomorphic sensor with memristive crossbar for brain-inspired visual perception

Networking retinomorphic sensor with memristive crossbar for brain-inspired visual perception

A Survey of Robotics Control Based on Learning-Inspired Spiking Neural Networks

Optimal grasp force for robotic grasping and in-hand manipulation with impedance control

Contact Info

Product

Resources

About