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 teardown, stripping industrial collaborative robots for good

29 点作者 vmayoral将近 4 年前

1 comment

LeifCarrotson将近 4 年前
Their research paper:<p><a href="https:&#x2F;&#x2F;aliasrobotics.com&#x2F;files&#x2F;robot_teardown_paper.pdf" rel="nofollow">https:&#x2F;&#x2F;aliasrobotics.com&#x2F;files&#x2F;robot_teardown_paper.pdf</a><p>might be a better link than their front page, though the UR3 teardown is interesting to me.<p>I find their claim that &quot;several robot manufacturers apply planned obsolescence practices today in an attempt to discourage repairs and evade competition&quot; in need of support. Their main supporting argument seems to be that manufacturers exclusively go through certified dealers (full disclosure: I&#x27;m a Fanuc certified integrator), therefore malice.... I&#x27;m not sure I follow.<p>As an industrial controls engineer, who has used a couple of UR3s and a ton of other industrial components sold under similar schemes, as well as designed, built and serviced fully custom CNC routers and laser cutters, the reason that old robots are obsoleted, that internal parts are undocumented, and that certified dealers are employed is to maximize the typical use case and minimize downtime.<p>I can bolt down a robot in front of a CNC and have it loading and unloading parts in something like 400 man-hours. It will do that day-in and day-out for a decade with occasional service. People will be put out of work and I&#x27;ll be getting a call at midnight if it goes down, expected to fix it over the phone or at absolute worst within 24 hours. I could do that work with servos and scales and custom control boards, building a Cartesian (or more complicated) system from $20k parts and $80k in labor, and be the only guy on the planet who understands it, spending weeks reverse engineering it when it goes down in the future...or I could buy a $30k 6-axis and put in $20k of labor, and use a known network of repair techs.<p>The state of robotics manufacturing comes from economics, which may produce unfortunately closed-source, manufacturer-locked machines, but doesn&#x27;t have to be motivated by malice. Even if I got a box of parts as well as full documentation from a UR3 for free, and my machine never needed servicing or adjustment, I couldn&#x27;t build a better machine for a typical integration application with those components than the robot they were designed to go into.<p>Sure, there&#x27;s a little vexing price segmentation where they&#x27;ll sell you a software bit flip to unlock a feature for $5k. Yes, the main point of the &quot;call us for pricing&quot; is to see whether you&#x27;re an OEM or primary integrator who will be locking in whole factories to sunk costs with a particular vendor of orange or yellow robots, so they&#x27;ll sell nearly at cost, whether you&#x27;re a second contractor pricing a model from quote requirements, or whether you have no choice but to buy that robot. Yeah, it sucks to pay $4k for a software license because the software division needs to be a profit center just like the hardware division. And yes, there&#x27;s a lot of crap buried inside the undocumented black box instructions, in most non-collaborative robots it&#x27;s just that the NRE costs were too high to make it better.