HN user
robotics/ML researcher: http://members.loria.fr/jbmouret/
Provide examples of uses of words and possibly help us rephrase in a more idiomatic way. Basically, answer "how do people use this word usually?"
A bit like linguee but for scientists: (at the bottom of the page): https://www.linguee.fr/francais-anglais/search?source=auto&q...
We're doing exactly this to teleoperate humanoid robots on high-latency networks!
Paper: https://arxiv.org/abs/2107.01281
Video: https://www.youtube.com/watch?v=N3u4ot3aIyQ
"We introduce a system in which a humanoid robot executes commands before it actually receives them, so that the visual feedback appears to be synchronized to the operator, whereas the robot executed the commands in the past. To do so, the robot continuously predicts future commands by querying a machine learning model that is trained on past trajectories and conditioned on the last received commands. In our experiments, an operator was able to successfully control a humanoid robot (32 degrees of freedom) with stochastic delays up to 2 seconds in several whole-body manipulation tasks, including reaching different targets, picking up, and placing a box at distinct locations."
Related (not historic stuffs necessarily): https://www.softwareheritage.org/
Here is an open-source, 3D-printable quadruped with (almost) direct-drive actuation:
https://open-dynamic-robot-initiative.github.io
we are also happy with waf, wich gives us a lot of flexibility (e.g., use python libs that are not build-related) and use a language that the team already know. (https://waf.io)
Gaussian Processes are very good when you do not have much data (< 500 points) and and your data are low-dimensional (<10 dimensions). For this, they are more accurate than anything else, and they provide both a prediction and variance.
Most journals are OK with authors posting a paper on their own website, especially the 'pre-print' (pre-review, pre-layout). See: http://www.sherpa.ac.uk/romeo/index.php
(even Nature is OK for all the pre-prints, arxiv and your own website).
This is the "net" of a senior researcher (very competitve position, people are about 35-40 year old, about 10-15 years after a good PhD) in public institutions (see http://www.emploitheque.org/grille-indiciaire-etat-Directeur...) and more than most "researchers" in public institutions (http://www.emploitheque.org/grille-indiciaire-etat-Charges-d... -- 5-10 years after a PhD).
This includes researches in CS and AI.
My definition: there is no clear limit, but there is a spectrum of ‘roboticity’ that corresponds to the spectrum of ‘versatility’. The more versatile a machine is, the more a robot it is.
For instance, a blender is not versatile. But a cooking robot can do more things (it blends, but also cooks, mixes, etc.). This is why the cooking robot is more a robot than a blender.
This is similar in industry. You have specialized machines, which can do a single thing. Industrial robots are more versatile because we can program them to achieve different task (e.g. when there is a new model of car). The ultimate robot would be as versatile as a human. This kind of humanoid would have the highest level of ‘roboticity’.
... or more than 100 top French researchers for 3 years (Research Director, top pay at the end of the career is about 6000 euros / ~10000 for the company).
[source : https://fr.wikipedia.org/wiki/Directeur_de_recherche_au_CNRS ]
check also iceSL, which is developped with 3D printing in mind: https://members.loria.fr/Sylvain.Lefebvre/icesl/
Original article: http://science.sciencemag.org/content/354/6312/aaf5239
The actual rule for European grant (H2020) is "either publish in open-access or put the paper on arxiv or similar archives". We can still publish in non-open access journals provided that we put the paper on an open access repository.
In H2020 projects (EU projects), the maximum embargo that can be "tolerated" is 6 months (then it has to be on arxiv) [12 months for social sciences]. However, all journals are OK with an arxiv post of the submitted (pre-review) manuscript, and this is what every authors should do.
Nature (and the vast majory of journals) is OK with posting on arxiv the pre-review manuscript, but not the post-review (ie, with the changes made after the comments by the reviewers) and not the version they edited (and added their page layout).
Some journals allow us to post the post-review paper (at any rate, the changes are usually mostly cosmetic).
Most of the robots in the manufacturing industry are following a pre-set sequence. Why do we call them a robot whereas most people would not call a dishwasher a robot?
I think what we (as a society) call a robot is a machine that has some degree of versatility: - an industrial robot can be programmed to achieve a different task (even if the sequence is pre-set, it is easy to set a different sequence for a different need), and the same kind of robots can be used to achieve many different tasks - a food processor can make many different recipes - a Roomba can adapt its behavior to the room (it has some degree of versatility because it can adapt to the conditions)
... but a dishwasher has a single purpose, which is why we usually do not call it a robot (even if it has sensors, actuators, some algorithms, etc.). If the dishwasher was also capable of cooking dishes, it would be more versatile and we would call this a robot (think of a humanoid torso that could do the dishes but also cook your eggs).
And the most versatile robots like a fully-featured humanoid is probably what we all have in mind when talk about robots.
Overall, we could say that we have degrees of versatility and therefore degrees of 'roboticity'. The lower level is the dishwasher, the highest level is the humanoid.
We discuss a lot on mailing lists, live in conferences, and with personnal e-mails.
By finding an existing system that shares common properties with the academic system but in which there is no competition for prestige.
I don't know, maybe documentation writing? Or classified papers?
In France (and I suspect in many European countries), most modern buildings/renovated buildings have large mailbox in which many packages fit (e.g. books). The mailman has a 'universal key' and drop the packages there.
A while ago, Internet was about people sharing things, freely. And it was an interesting place!
Not all the content has to be written by professional writers and a lot of code is actually written for free by volunteers...
As a scientist, I am most efficient when I skim with a goal. Typically, I would search for an answer to a very specific question (related to my own research) and just ignore whole parts of the book/article.
4. The grant agencies (e.g. the NSF) could use some of its money to pay some labs to reproduce the study during the end-of-grant review process, in the same way as it already pays reviewers and committees for the initial review process. This would add a direct pressure to publish things that are easy to reproduce.
Of course, that would mean even less grants...
5. Each PI should have a single grant at a time (and a long one), to end the "money competition" (that mirrors the publication competition) and spread the grants better.
Why can't people simply publish negative results?
In addition to the other comments: because a negative result is usually much more likely than a negative one, it is very hard to be sure that the experimenter did not simply screw up the experiment.
... and we still need to understand what these "costs of peer review" are! We are never paid for reviewing papers and most editors work for free. There is free software to handle the peer review process (e.g. http://myreview.sourceforge.net) which are used by some serious journals for years (e.g. http://ecj.fhv.at / http://www.mitpressjournals.org/loi/evco).
A few more points: - people outside of computer sciences (and maybe physics) do not want to deal with LaTeX. As a result, they want to submit ugly word files and get a nicely formatted paper out of it (this has a cost, but not in CS where we submit nicely formatted LaTeX papers)
- most journals do not do copy-editing, but the big ones actually work on the figures, the text, fix the references, etc. Again, this is not common in CS.
I also don't like it but the paper needs be printed and reviewed.
- Reviews are always free (nobody gets paid for reviewing academic paper, in contrast to grant proposal for which we are sometimes paid)
- I have not seen a printed version of a journal for a while (except the very big ones like Nature or Science, but we buy them mostly for the news/view section, not that much for the academic papers at the end)
- Copy editors never did anything useful to my papers. In CS, we usually submit a nice latex file and they do not really do any work on the layout (it might be different in other fields, where people submit crappy MS Word files).
... so, let's say that the added value of the publisher (even for PLOS) is marginal...
... with a 6 months embargo (at best).
Also: do not let the Uni PR office write the press release. Write it yourself and insist they use yours.
A good thing that we did for our last paper [1]: we have set up a FAQ [2] with the most common questions (we updated the FAQ when we received new questions). Thanks to the FAQ, the journalists do not "invent" too much and they have quotes that they can use.
Overall, I think the FAQ helped a lot and it avoided some common mis-interpretations.
[1] Paper: http://www.nature.com/nature/journal/v521/n7553/full/nature1... | arxiv version & video: http://chronos.isir.upmc.fr/~mouret/website/nature_press.xht...
[2] FAQ: http://chronos.isir.upmc.fr/~mouret/website/nature_press.xht...
It is usually said that the English built the first railways in France and that they built them so that trains drive left, out of habit.