On Tuesday, 23 March 2021 at 15:12:54 UTC+1 Nil wrote:
> On 3/23/21 1:01 PM, Michele Thiella wrote: > > - In reply to Nil: (on Slack i'm named Raschild) > > I had seen the beginning of the work and it is very interesting. In the > > next few days I will look at the current state. > > Two quick questions: > > 1) How complicated is it to work directly with Ros + Gazebo compared to > > Malmo and Gym? > > I don't know as I've never used Ros or Gazebo. Working with Malmo and > Gym is fairly easy but that's cause they have been designed this way, > and as a result they are fairly limited. For instance the communication > protocol is completely synchronous, it's certainly not something you'd > want to use to control a robot. > > > 2) Are Values already usable instead of OctoMap and SpaceTime server? > > At the current stage ROCCA uses neither values, nor even octomap or > spacetime server. Everything, including spacetime events, lives in the > atomspace, which is extremely inefficient but is not my concern for now. > My concern is to build an agent that makes decisions as rationally as > possible, in unknown and uncertain environments. > > > In conclusion: > > * My master's proposal has already been accepted with the proviso to get > > a feasible project. So, aim for a goal and try to achieve it. If so, > > excellent; if not, show why it failed (I will try to avoid it). > > > > * The direction of the project is still incomplete. Unfortunately, i > > can't figure out if it takes me 1 day, 1 month or 1 year to > > understand/implement a certain code. > > I started with the reasoning: I am currently learning the inference > > rules and how they work with the atomspace, I have seen part of the > > examples in ure and pln and I was trying to understand the blocksworld > > problem developed by Anatoly Belikov here: > > https://github.com/noskill/ure/tree/planning/examples/ure/planning > > > > * Ideally my goal was to extend the "model of the world" to work more > > with objects than people and to extend the "self-model" to execute > > navigation and manipulation plans. In all of this, I haven't yet > > explored the learning. > > Based on what you're saying I think ROCCA would be a good fit. Consider > of course that it is very early stage. I don't mind to semi-mentor you > as long as you're somewhat autonomous (which you seem to be). > Just to throw in my 3 cents here, I have done some work on moving ROCCA to use the MineRL’s Gym API to access Minecraft, instead of using a separate Malmo wrapper. It works in the sense that I can run the example code in that way and it simplifies the whole design somewhat, as interfacing with Minecraft is reduced to relying on a familiar structure of a Gym API. I also have a setup where I can run the whole thing in a Docker container, so it’s almost reproducible (there is a minor file edit needed). An idea that I wrote about on Discord was to bring in some unsupervised image segmentator (like MONet), train it on the MineRL dataset (they have a dataset of Minecraft traces) and then use that to have some information about visual objects available to the agent. For now I am stuck a bit though, as sadly the loading code they provided for their dataset has memory leak issues, so I will have to write my own, it shouldn’t prove too difficult as I just need the video frames but I have to get around to implement it yet. > > Do you want to have a call (say Friday, as it's ROCCA day for me)? I > could walk you through the code, to help you decide whether you want to > work on it, or else work on Eva. > > Nil > > > > > > > Michele > > > > Il giorno martedì 23 marzo 2021 alle 06:19:25 UTC+1 Nil ha scritto: > > > > Forwarding to opencog as I forgot to reply-all. > > > > -------- Forwarded Message -------- > > Subject: Re: New user [was Re: [opencog-dev] Problem in atom deletion > > from postgreSQL > > Date: Tue, 23 Mar 2021 07:15:10 +0200 > > From: Nil Geisweiller <[email protected]> > > To: Michele Thiella <[email protected]> > > > > Hi Michele, > > > > I'm working on something that might be relevant to your work, see > > > > https://github.com/opencog/rocca <https://github.com/opencog/rocca> > > https://github.com/singnet/rocca <https://github.com/singnet/rocca> > > (mirror) > > > > it's a WIP and advancing slowly due to myself being involved with other > > projects, but it has IMO much potential, especially in term of > > leveraging OpenCog reasoning capabilities, if that's what your into. > > > > Nil > > > > On 3/21/21 1:10 PM, Michele Thiella wrote: > > > You're right, I didn't introduce myself properly! I mentioned my > > plans > > > in response to one of the previous conversations but it was the > > wrong > > > place so it got lost. > > > Let's start again in the right way: > > > > > > Hello everyone, I'm Michele Thiella from Italy (Padua). > > > I am about to graduate in Computer Engineering at the University > > of Padua. > > > I have always been passionate about artificial intelligence and, > > more > > > generally, about everything that has not yet been > > > invented/discovered/solved. > > > About 6 months ago I chose the path for my master's thesis: AGI. I > > > started reading around until I got to you and I was blown away. > > > I presented your work (roughly) to my Thesis Supervisors, trying > > to get > > > approval for a thesis involving OpenCog. > > > And here we are. > > > > > > For my thesis the proposal would be to use OpenCog to do TAMP. > > > More precisely, using a simplified version of Sophia's architecture, > > > switching Blender with Ros and Sophia with a much simpler robot. > > > Leaving out the emotional sphere of the robot and aiming at the > > > resolution of objectives, perhaps achieving cooperation between > > robots > > > through a single knowledge base. > > > Or at least that was the idea. > > > > > > I don't think I explained it the way I wanted but I hope it goes > > well. > > > > > > Il giorno sabato 20 marzo 2021 alle 20:04:05 UTC+1 linas ha scritto: > > > > > > So, out of curiosity, what are you using OpenCog for? What are you > > > attempting to do? You're new here, you should introduce yourself! > > > > > > -- Linas > > > > > > On Thu, Mar 11, 2021 at 12:42 PM Michele Thiella > > > <[email protected]> wrote: > > > > > > > > > You were right! After the pull everything works! It works so > > > well that it has kept the backwards-compat layer and it works > > > even without the exclamation mark haha > > > > > > Regarding RockDB, I tried to install it but now I don't remember > > > at what point I stopped and why .. but I definitely had space > > > problems. Too bad because it inspired me a lot as storage, > > > anyway 50% slower is really a lot! > > > However (unfortunately) I have skills with postgresql so for now > > > I am satisfied, since everything works. > > > Thanks a lot for the answer! Good work! > > > > > > Michele Thiella > > > Il giorno mercoledì 10 marzo 2021 alle 23:35:42 UTC+1 linas ha > > > scritto: > > > > > > Hi Michele, > > > > > > Just around the same time that you sent this message (or a > > > day before?) I spotted and fixed a problem with atom > > > deletion. So, please try rebuilding and reinstalling the > > > atomspace. (git pull; cd build; make -j; sudo make install) > > > -- let me know if that solves the problem. > > > > > > Note also, the proper name is cog-delete! (with an > > > exclamation mark at the end) I may have removed the > > > backwards-compat layer that allowed it to work without the > > > exclamation mark.. > > > > > > Note also: there is a RocksDB backend too. It might be > > > easier to use (no config required). In some synthetic > > > benchmarks, its 2x or 3x faster than postgres. In the one > > > "real-life" app that I'm using, its 10%-50% slower. Go > > > figure. Anyway its at > > > https://github.com/opencog/atomspace-rocks > > <https://github.com/opencog/atomspace-rocks> > > > <https://github.com/opencog/atomspace-rocks > > <https://github.com/opencog/atomspace-rocks>> > > > > > > (There is also a network server backend: > > > https://github.com/opencog/atomspace-cog > > <https://github.com/opencog/atomspace-cog> > > > <https://github.com/opencog/atomspace-cog > > <https://github.com/opencog/atomspace-cog>> ... the README > > > explains more) > > > > > > -- Linas > > > > > > On Mon, Mar 8, 2021 at 3:58 AM Michele Thiella > > > <[email protected]> wrote: > > > > > > Goodmorning everyone, > > > i have a simple problem about deleting atoms in postgresql. > > > According to what I understand, to delete an atom saved > > > in postgres I should use: > > > (cog-delete (Concept "asdf")) > > > The command is successful, the atom is removed from the > > > atomspace but not from the postgresql database. Could > > > someone kindly tell me why? > > > The prostgresql backend should be configured correctly, > > > according to the guide on the wiki ... saving new atoms > > > in the database works. > > > Thanks in advance, I apologize for the inconvenience > > > > > > -- > > > You received this message because you are subscribed to > > > the Google Groups "opencog" group. > > > To unsubscribe from this group and stop receiving emails > > > from it, send an email to [email protected]. > > > To view this discussion on the web visit > > > > > > https://groups.google.com/d/msgid/opencog/8dd1cfa3-0e8a-4d85-ba2a-6b38fcd2aea7n%40googlegroups.com > > > < > https://groups.google.com/d/msgid/opencog/8dd1cfa3-0e8a-4d85-ba2a-6b38fcd2aea7n%40googlegroups.com> > > > > > > > > > < > https://groups.google.com/d/msgid/opencog/8dd1cfa3-0e8a-4d85-ba2a-6b38fcd2aea7n%40googlegroups.com?utm_medium=email&utm_source=footer > > > < > https://groups.google.com/d/msgid/opencog/8dd1cfa3-0e8a-4d85-ba2a-6b38fcd2aea7n%40googlegroups.com?utm_medium=email&utm_source=footer>>. > > > > > > > > > > > > > > > > -- > > > Patrick: Are they laughing at us? > > > Sponge Bob: No, Patrick, they are laughing next to us. > > > > > > > > > -- > > > You received this message because you are subscribed to the > > > Google Groups "opencog" group. > > > To unsubscribe from this group and stop receiving emails from > > > it, send an email to [email protected]. > > > > > > To view this discussion on the web visit > > > > > > https://groups.google.com/d/msgid/opencog/f5726803-91a0-405d-99b4-deefeb48486dn%40googlegroups.com > > > < > https://groups.google.com/d/msgid/opencog/f5726803-91a0-405d-99b4-deefeb48486dn%40googlegroups.com> > > > > > > > > > < > https://groups.google.com/d/msgid/opencog/f5726803-91a0-405d-99b4-deefeb48486dn%40googlegroups.com?utm_medium=email&utm_source=footer > > > < > https://groups.google.com/d/msgid/opencog/f5726803-91a0-405d-99b4-deefeb48486dn%40googlegroups.com?utm_medium=email&utm_source=footer>>. > > > > > > > > > > > > > > > > -- > > > Patrick: Are they laughing at us? > > > Sponge Bob: No, Patrick, they are laughing next to us. > > > > > > > > > -- > > > You received this message because you are subscribed to the Google > > > Groups "opencog" group. > > > To unsubscribe from this group and stop receiving emails from it, > > send > > > an email to [email protected] > > > <mailto:[email protected]>. > > > To view this discussion on the web visit > > > > > > https://groups.google.com/d/msgid/opencog/94d7c89f-5547-4ef8-9de6-668972bfb385n%40googlegroups.com > > > < > https://groups.google.com/d/msgid/opencog/94d7c89f-5547-4ef8-9de6-668972bfb385n%40googlegroups.com> > > > > > > > > > < > https://groups.google.com/d/msgid/opencog/94d7c89f-5547-4ef8-9de6-668972bfb385n%40googlegroups.com?utm_medium=email&utm_source=footer > > > < > https://groups.google.com/d/msgid/opencog/94d7c89f-5547-4ef8-9de6-668972bfb385n%40googlegroups.com?utm_medium=email&utm_source=footer>>. > > > > > > > > -- > > You received this message because you are subscribed to the Google > > Groups "opencog" group. > > To unsubscribe from this group and stop receiving emails from it, send > > an email to [email protected] > > <mailto:[email protected]>. > > To view this discussion on the web visit > > > https://groups.google.com/d/msgid/opencog/e2f4037e-daa8-4308-99dd-74f41130745bn%40googlegroups.com > > > < > https://groups.google.com/d/msgid/opencog/e2f4037e-daa8-4308-99dd-74f41130745bn%40googlegroups.com?utm_medium=email&utm_source=footer>. > > > -- You received this message because you are subscribed to the Google Groups "opencog" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/opencog/171ac33d-856c-4be5-8815-a9c3ce522b2fn%40googlegroups.com.
