ADM-201 dump PMP dumps pdf SSCP exam materials CBAP exam sample questions

虚拟生物在三维世界学习导航 – 译学馆
未登陆,请登陆后再发表信息
最新评论 (0)
播放视频

虚拟生物在三维世界学习导航

Digital Creatures Learn to Navigate in 3D | Two Minute Papers

亲爱的学霸同学们 这里是Károly Zsolnai-Fehér带来的两分钟论文
Dear Fellow Scholars, this is Two Minute Papers with Károly Zsolnai-Fehér.
早些时候 我们讨论过一些很不错的用来教虚拟人物行走的算法
Earlier, we have talked about a few amazing algorithms to teach digital creatures to walk.
今天 我们感兴趣的是如何控制虚拟人物的关节
And this time, we’re interested in controlling the joints of a digital character
从而使它不仅能正确地行走 还能考虑到周围的环境
to not only walk properly, but take into consideration its surroundings.
这个新技术可以让虚拟人物在有静态或是动态移动的障碍物的三维世界中行进
This new version can navigate in 3D with static and dynamic moving obstacles,
甚至可以朝着目标带球前进
or even dribble a ball toward a target.
超喜欢这篇论文的执行过程和最后成果
Loving the execution and the production value of this paper.
它是通过一套非常有效率的系统来完成的 此系统
This is accomplished by an efficient system that consists of two controllers
由代表学习算法的两个控制器组成
that are represented by learning algorithms.
第一个是低级控制器 通过调节关节的位置和速度
One, the low level controller is about about maintaining balance and proper limb control
来维持平衡和控制四肢
by manipulating the joint positions and velocities appropriately.
这个控制器是在精细时间刻度上运作的 每秒运作三十次
This controller operates on a fine time scale, thirty times per second
它是经由一个四层神经网络来训练的
and is trained via a 4-layer neural network.
第二个是高级控制器 用来达成一些更宏观更智能的目标
Two, the high level controller can accomplish bigger overarching goals,
比如跟随路径 或者避开静态或动态的障碍物
such as following a path, or avoiding static and dynamic obstacles.
这个控制器不需要经常运作 因此可以节省资源
We don’t need to run this so often, therefore to save resources.
它在粗略时间刻度上运作 每秒运作两次
this controller operates on a coarse time scale, only twice each second
经由一个深卷积神经网络来训练
and is trained via a deep convolutional neural network.
它还支持少量的迁移学习
It also has support for a a small degree of transfer learning.
迁移学习是表示在成功学习解决一个问题以后
Transfer learning means that after successfully learning to solve a problem,
再对下一个问题 我们不需要完全从头学起
we don’t have to start from scratch for the next one,
而是可以重复利用之前已学到的有用的知识 从而抢占先机
but we can reuse some of that valuable knowledge and get a headstart.
这是现在非常热门的研究领域
This is a heavily researched area
而且可能会是机器学习研究中的下一个最主要的前沿课题
and is likely going to be one of the major next frontiers in machine learning research.
现在 不要弄错了 这里并不是说迁移学习这一难题已经被解决了
Now, make no mistake, it is not like transfer learning is suddenly to be considered a solved problem
而是在这个例子中 它最终成为可能
– but in this particular case, it is finally a possibility.
太酷了! 我希望这段简短的讲解也能让你热血沸腾
Really cool! I hope this brief exposé fired you up too.
这篇论文赞爆了 请一定看一下
This paper is a bomb, make sure to have a
像以前一样 论文链接在视频描述栏可以找到
look, as always, the link is available in the video description.
顺便说一下 由于有了你们在Patreon上的支持 我们很快就可以
And by the way, with your support on Patreon, we will soon be able
将使用部分经费来加强研究项目
to spend part of our budget on empowering research projects.
是不是太棒了?
How amazing is that?
新的两分钟论文衬衫已经发售了!
The new Two Minute Papers shirts are also flying off the shelves!
很高兴知道你们那么喜欢它
Happy to hear you’re enjoying them so much.
如果你有兴趣并且位于美国 请点击http://twominutepapers.com
If you’re interested, hit up http://twominutepapers.com if you’re located in the US.
欧盟和全球销售点店的链接也可在视频描述栏找到
The EU and worldwide store’s link is also available in the video description.
感谢收看和支持 下次再见
Thanks for watching and for your generous support, and I’ll see you next time!

发表评论

译制信息
视频概述

本集介绍了一种新的技术,使得虚拟人物不仅可以在三维世界正确行进,还能对周围环境做出正确应对。

听录译者

收集自网络

翻译译者

豆子

审核员

知易行难

视频来源

https://www.youtube.com/watch?v=2vnLBb18MuQ

相关推荐