TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Robot Learns to See in 30 Minutes (2022)

229 点作者 stoniejohnson大约 2 年前

12 条评论

stoniejohnson大约 2 年前
Via simulation + meta-learning they can make a <i>blind</i> robot walk over arbitrary terrains solely via proprioception.<p>They add a visual component, which with 30 minutes of real world data, through a technique called Cross-Modal Supervision (CMS), is able to improve walking performance significantly!<p>They can even distort its vision (turn the camera) and it learns to adapt!<p>Paper: <a href="https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;2211.03785" rel="nofollow">https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;2211.03785</a><p>Advisor Presentation: <a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=f6fDpKDxpL0">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=f6fDpKDxpL0</a><p>Note: This is not my work; I originally found the presentation above.
评论 #35320594 未加载
vessenes大约 2 年前
What seems clever to me is the super cool hack of getting sensor data while the video data is captured, then training a model on realtime video and backward time-shifted sensor data; result is a network that can predict sensor data 30 seconds ahead of time with visual data. This is then used to feed the walking model. That&#x27;s the sort of super &#x27;dumb&#x27; thing ML folks can do now; that task would have taken many many person years of work before, now, I speculate thinking of the idea to testing a network would be under a week. Verry cool.
评论 #35320991 未加载
评论 #35321129 未加载
AiaAidan大约 2 年前
Meanwhile, it takes the human brain months to adapt to visual input after blindness: &quot;This research suggests that the brain is able to adapt to getting visual input, even after 10 years of blindness. That said, it takes a little while before this ability to really process the images kicks in. Right after the second surgery, patients are getting input to the brain from their retinas, but their visual memory is still poor. After about a month, though, the brain has found ways to process this visual information, and, a year later, their brains do just about as well as those of people who had sight from birth.&quot; Source: <a href="https:&#x2F;&#x2F;psychologytoday.com&#x2F;us&#x2F;blog&#x2F;ulterior-motives&#x2F;202210&#x2F;what-happens-when-blind-people-get-vision-correction" rel="nofollow">https:&#x2F;&#x2F;psychologytoday.com&#x2F;us&#x2F;blog&#x2F;ulterior-motives&#x2F;202210&#x2F;...</a>.
评论 #35321387 未加载
评论 #35325155 未加载
fwlr大约 2 年前
“After training our robot on a large set of terrains in the Berkeley campus, we verified generalization of the visual policy in the Stanford campus.”<p>Sublime. Almost a koan for meditating on AI research.
spiritplumber大约 2 年前
We did something similar but much simpler in 2009 where the robot learned where its sensors where and which way its motors moved it, and how to use them.<p><a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=kk-H--e6PUs">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=kk-H--e6PUs</a>
pfd1986大约 2 年前
Last author, J Malik, has made many important contributions to computer vision. One I&#x27;m very grateful for is spectral manifolds.
评论 #35324372 未加载
评论 #35320843 未加载
chiefalchemist大约 2 年前
It&#x27;s not the learning that&#x27;s the magic. It can be 30 minutes or 30 days. It hardly matters. Where the sparks fly is that learning can he transfered in .3 seconds to 300,000 other robots. And then updated as needed.<p>Now compare the to humans.
评论 #35320698 未加载
评论 #35320600 未加载
评论 #35320625 未加载
评论 #35320714 未加载
29athrowaway大约 2 年前
Humans not only &quot;learn to see&quot;. They learn sensor fusion, which is the aggregation of your senses.<p>Read about vertigo and the ear, mirror box therapy, etc.
margorczynski大约 2 年前
I can&#x27;t find it in the paper - is the inference ran on the robot directly or remotely on a more powerful computer?
评论 #35320663 未加载
precompute大约 2 年前
&gt;This work was supported by the DARPA Machine Common Sense program<p>Coming soon to a neighbourhood near you!
mrnonchalant大约 2 年前
Looking forward to seeing the code! Any sense how soon “soon” might be?
extrememacaroni大约 2 年前
ok, which jobs is this one going to steal now?
评论 #35320792 未加载
评论 #35320780 未加载
评论 #35320718 未加载
评论 #35321045 未加载
评论 #35320894 未加载