Emergence of Content-Agnostic Information Processing by a Robot Using Active Inference, Visual Attention, Working Memory, and Planning

Emergence of Content-Agnostic Information Processing by a Robot Using Active Inference, Visual Attention, Working Memory, and Planning
复制标题

机器人使用主动推理、视觉注意力、工作记忆和规划进行与内容无关的信息处理的出现

DOI:
10.1162/neco_a_01412
复制
发表时间:
2021
期刊:
影响因子:
2.9
通讯作者:
Tani Jun
Tani Jun
中科院分区:
计算机科学4区
文献类型:
--
作者:
Queisser Jeffrey Frederic;Jung Minju;Matsumoto Takazumi;Tani Jun

文献摘要

参考文献

相似文献

通过学习进行概括是人类基本的认知能力。例如,我们甚至可以操纵不熟悉的物体,并在制定预先计划之前产生心理图像。这怎么可能?我们的研究通过回顾我们以前的研究(Jung,松本和Tani)来研究这个问题,该研究研究了机器人执行块堆叠任务的基于视觉的目标导向规划问题。通过扩展以前的研究,我们的工作介绍了一个大型网络,包括动态交互的子模块,包括视觉工作记忆(VWM),视觉注意模块,和执行网络。执行网络预测运动信号、视觉图像和各种注意力控制,以及视觉信息的掩蔽。与以前的研究最显着的区别是,我们目前的模型包含一个额外的VWM。整个网络通过使用预测编码进行训练,并使用主动推理推断出实现给定目标状态的最佳视觉计划。结果表明,我们目前的模型性能明显优于Jung等人,特别是在操纵未学习的颜色和纹理块。仿真结果表明,所观察到的泛化,因为内容不可知的信息处理通过第二VWM和其他模块在学习过程中的协同作用,其中记忆图像内容和转换它们是分离的发展。这封信通过对模拟结果进行定性和定量分析来验证这一说法。
Generalization by learning is an essential cognitive competency for humans. For example, we can manipulate even unfamiliar objects and can generate mental images before enacting a preplan. How is this possible? Our study investigated this problem by revisiting our previous study (Jung, Matsumoto, & Tani, ), which examined the problem of vision-based, goal-directed planning by robots performing a task of block stacking. By extending the previous study, our work introduces a large network comprising dynamically interacting submodules, including visual working memory (VWMs), a visual attention module, and an executive network. The executive network predicts motor signals, visual images, and various controls for attention, as well as masking of visual information. The most significant difference from the previous study is that our current model contains an additional VWM. The entire network is trained by using predictive coding and an optimal visuomotor plan to achieve a given goal state is inferred using active inference. Results indicate that our current model performs significantly better than that used in Jung et al. , especially when manipulating blocks with unlearned colors and textures. Simulation results revealed that the observed generalization was achieved because content-agnostic information processing developed through synergistic interaction between the second VWM and other modules during the course of learning, in which memorizing image contents and transforming them are dissociated. This letter verifies this claim by conducting both qualitative and quantitative analysis of simulation results.
认知神经科学的关注:概述。
DOI: --
发表时间: 1995
期刊:
影响因子: --
作者:
M. Posner
通讯作者: M. Posner
论机器人的代理感和客体持久性
DOI: --
发表时间: 2016
期刊: Joint IEEE International Conference on Development and Learning and on Epigenetic Robotics
影响因子: --
作者:
Sarah Bechtle;G. Schillaci;V. Hafner
通讯作者: V. Hafner
DOI: 10.1142/s0129065707001081
发表时间: 2007
影响因子: 8
作者:
H. Wersing;Stephan Kirstein;Michael Götting;H. Brandl;Mark Dunn;Inna Mikhailova;C. Goerick;Jochen J. Steil;H. Ritter;E. Körner
通讯作者: E. Körner
视觉工作记忆操纵的神经生理学标志
DOI: 10.1167/17.10.1116
发表时间: 2017
期刊: Journal of Vision
影响因子: 1.8
作者:
Hrag Pailian;Viola S. Störmer;G. Alvarez
通讯作者: G. Alvarez
DOI: 10.1038/s42003-021-02994-2
发表时间: 2022-01-14
影响因子: 5.9
作者:
Isomura T;Shimazaki H;Friston KJ
通讯作者: Friston KJ