It's realistic to do this using Original OpenCog, and a fairly interesting challenge. Whether the current Hyperon prototype is suitable for this sort of experimentation right now is not clear to me , Alexey or Vitaly could tell you...
On Thu, Apr 29, 2021 at 2:16 AM Michele Thiella <[email protected]> wrote: > > Hi Ben, a great pleasure to meet you! > > Il giorno mercoledì 28 aprile 2021 alle 21:42:02 UTC+2 Ben Goertzel ha > scritto: >> >> I was referring to Michele's student project which is clearly not >> aimed at scalable production code... whether Michele uses Hyperon or >> Original OpenCog, it's a student research experiment on >> BlocksWorld.... > > > you are right, my project is absolutely not aimed at a development of > scalable production code (it would be nice but I would not graduate anymore). > Broadly speaking, the idea is to bring a simple example of the potential of > an architecture towards AGI in an "industrial" environment. > Now, what potential can it ever have? > > My supervisors say that if I solved this simple plan, I could add > communication with a human and > thus get a planner that is certainly inefficient, restricted to the domain of > blocks, etc. > But a planner based on the semantic aspect of the problem, compared to all > those solvers who solve problems without understanding their meaning. > > Simple example: > > let's say that the red cube is a book and the green cube is a glass. > The robot knows this (for now I'm using apriltag above the cubes, but with > pointcloud and some good classifiers I think we can recognize objects without > labeling them, or better yet with NN on the images). > Anyway, the robot's actions are pick-up and put-down of cubes. > The idea is to tell the robot "Drink" or "Read" and see it take the glass > rather than the book. > Finally it would be nice that, if the plan fails (there isn't a book on > table), the robot interacts with the human to increase its knowledge base and > fill in the gaps that led to the plan to fail. > > I'm not sure if this all makes sense, I'm still learning and understanding > this wide world. > Maybe you will be able to comment on this idea. > > >> >> Hyperon is ultimately intended as an alternative to current OpenCog >> Atomspace, yes. However the current crude Hyperon prototype code is >> definitely NOT intended as an alternative to Atomspace -- yet can >> still be used for research experimentation. >> >> > (1) ASP will solve most planning and constraint satisfaction problems in >> > milliseconds. So you can do a 60-frames-per-second update rate if you wish. >> >> Ah cool, good to know!!! My intuition on this is obsolete, >> apparently ;p ... will look up some references... > > > So ultimately, could it make more sense for me to work with the current > OpenCog or approach Hyperon? > > Thank you all for your interest. > > Michele > > > -- > 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/eb3f85de-53b1-48f4-a802-338949316938n%40googlegroups.com. -- Ben Goertzel, PhD http://goertzel.org “He not busy being born is busy dying" -- Bob Dylan -- 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/CACYTDBdGQNB-HgTftdok%2B1%3DB7j49Wo%3Dr2Ax8zDcNLU_52vHAbA%40mail.gmail.com.
