Autonomous learning of disparity–vergence behavior through distributed coding and population reward: Basic mechanisms and real-world conditioning on a robot stereo head

Gibaldi, Agostino; Canessa, Andrea; Solari, Fabio; Sabatini, Silvio P.

doi:10.1016/j.robot.2015.01.002

Cited by 9 publications

(6 citation statements)

References 49 publications

(63 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Because the trained network produces coordinated movements of both eyes to the same target, it is able to perform vergence eye movements in addition to saccades. Other models of vergence control rely on disparity detection (Gibaldi, Canessa, Chessa, Solari, & Sabatini, 2013; Gibaldi, Canessa, Solari, & Sabatini, 2015; Gibaldi, Chessa, Canessa, Sabatini, & Solari, 2009; Patel, Öğmen, White, & Jiang, 1997; Theimer & Mallot, 1994; Vikram, Teuliere, Zhang, Shi, & Triesch, 2014; Zhao, Rothkopf, Triesch, & Shi, 2012). The approach proposed in this article is complementary to these disparity-based methods.…”

Section: Discussionmentioning

confidence: 99%

A neural model of binocular saccade planning and vergence control

Muhammad

Spratling

2015

Adaptive Behavior

View full text Add to dashboard Cite

The human visual system uses saccadic and vergence eye movements to foveate visual targets. To mimic this aspect of the biological visual system the PC/BC-DIM neural network is used as an omni-directional basis function network for learning and performing sensory-sensory and sensory-motor transformations without using any hardcoded geometric information. A hierarchical PC/BC-DIM network is used to learn a head-centred representation of visual targets by dividing the whole problem into independent subtasks. The learnt headcentred representation is then used to generate saccade and vergence motor commands. The performance of the proposed system is tested using the iCub humanoid robot simulator.

show abstract

Section: Discussionmentioning

confidence: 99%

A neural model of binocular saccade planning and vergence control

Muhammad

Spratling

2015

Adaptive Behavior

View full text Add to dashboard Cite

show abstract

“…For example, the recent approach by Konda and Memisevic (2014) also learns disparity and motion representations, but it does so from a fixed set of training videos via supervised learning and it does not include the learning of any behavior which would change the statistics of the sensory signals. Conversely, the approach by Gibaldi et al (2015) learns to execute vergence eye movements, but the set of filter banks which are used to process the input images is predefined and does not adapt to the statistics of the visual input. Indeed, the majority of existing models for learning vergence or smooth pursuit have a much narrower focus than our work.…”

Section: Discussionmentioning

confidence: 99%

Autonomous Development of Active Binocular and Motion Vision Through Active Efficient Coding

et al. 2019

View full text Add to dashboard Cite

We present a model for the autonomous and simultaneous learning of active binocular and motion vision. The model is based on the Active Efficient Coding (AEC) framework, a recent generalization of classic efficient coding theories to active perception. The model learns how to efficiently encode the incoming visual signals generated by an object moving in 3-D through sparse coding. Simultaneously, it learns how to produce eye movements that further improve the efficiency of the sensory coding. This learning is driven by an intrinsic motivation to maximize the system's coding efficiency. We test our approach on the humanoid robot iCub using simulations. The model demonstrates self-calibration of accurate object fixation and tracking of moving objects. Our results show that the model keeps improving until it hits physical constraints such as camera or motor resolution, or limits on its internal coding capacity. Furthermore, we show that the emerging sensory tuning properties are in line with results on disparity, motion, and motion-in-depth tuning in the visual cortex of mammals. The model suggests that vergence and tracking eye movements can be viewed as fundamentally having the same objective of maximizing the coding efficiency of the visual system and that they can be learned and calibrated jointly through AEC.

show abstract

“…As a matter of fact, scholars from all over the world have already begun to explore the methods to improve channel utilization and realize multiple access in a communication system to overcome the problem of signal collision [21]. Therefore, it is necessary to make a simple analysis of these methods.…”

Section: Information Channel and Channel Capacitymentioning

confidence: 99%

“…The research improves the system's understanding of the environment from a one-sided, discrete, and passive perception level to a global, related, and active perception level. This research explores the establishment of a method system suitable for autonomous perception and control in a dynamic environment, which is a robust modeling requirement for the system to achieve adaptive autonomous control [21][22][23][24].…”

mentioning

confidence: 99%

Overview of RFID System Anti-Collision Technology

Zhao

Zhang³

2021

Physical Anti-Collision in RFID Systems

View full text Add to dashboard Cite

This book is based on the basic principles of physics (including electromagnetics, optics, thermodynamics, etc.), with engineering mathematical methods as the core of detection and control algorithm design. Furthermore, this book innovatively applies semi-physical verification and detection technology to the dynamic performance testing of radio-frequency identification (RFID) systems. This book proposes a series of new theories and methods for physical collision prevention, as well as related test verification methods for building semi-physical hardware platforms based on photoelectric sensing technology, which will provide important theories and technical supports for the applications of the internet of things (IoT) systems in smart logistics, car networking, food traceability, anti-counterfeiting, and other livelihood fields. The main contributions of this book are as follows:(1) It introduces a new theory of physical anti-collision for improving the integral performance of multi-tag RFID systems.(2) It studies the principles and optimization algorithms of innovative multi-tag network topologies, which enhance the sensing capability of the RFID system. (3) It provides valuable guidance for RFID system physical anti-collision based on deep learning. (4) It combines physical and electrical expertise from an interdisciplinary standpoint toward the analysis and design of dynamic performance testing systems. RFID Collision Problem and Research Progress of Anti-CollisionInternet of things (IoT) as a new technology has developed rapidly in recent years, which is an important part of a new generation of information systems. The emergence of IoT is another information technology revolution after computers, the Internet, and mobile communications [1]. As one of the core technologies in the field of perception of IoT, radio-frequency identification (RFID) technology is a

show abstract

Autonomous learning of disparity–vergence behavior through distributed coding and population reward: Basic mechanisms and real-world conditioning on a robot stereo head

Cited by 9 publications

References 49 publications

A neural model of binocular saccade planning and vergence control

A neural model of binocular saccade planning and vergence control

Autonomous Development of Active Binocular and Motion Vision Through Active Efficient Coding

Overview of RFID System Anti-Collision Technology

Contact Info

Product

Resources

About