- Joined
- Aug 8, 2022
- Messages
- 1,252
Hi Amberwolf! Be well!
Robotics Project: Snuggles, The Wolf
- Thread starter amberwolf
- Start date
nicobie
100 MW
"Yes. Care for yourself first so you can help others later. The logic."
Hang in there AW.
Hang in there AW.
nicobie
100 MW
Maybe next time try a smaller dog?
marty
1 MW
Women Like Small Dogs and Men Like Big Dogs
Is It True That Women Like Small Dogs and Men Like Big Dogs?
The popular stereotype is that women prefer small dogs and men prefer large dogs. New data tests whether this is true.
I truly mean it when I say that I seriously doubt your behavior could ever be so wrong that it went beyond "just rude and/or insensitive." You don't give yourself nearly enough credit, my friend. I just can't imagine you being so at fault that you have reason to beat yourself up over it. Personally, my method of dealing with people is literally this: "frock 'em, they're irrationally sensitive and I'm not going to let them grind me down any more than they already have." We have just as much right to be ourselves as they have to be offended by that. I know that's probably of no help to you, but it usually gets me through the day. Sure, there are times when I can't just will myself out of dwelling or feeling ashamed, but usually it works pretty well for me. Try it sometime, it might work for you. Besides, do you really want to be friends with people who blame you instead of considering that they might also be at fault? C'mon, people like that are worse than useless.
Find some people from foreign countries to be friends with. They're few and far between, but in my experience, they're much less critical of social customs.
As for the rest of your post... I don't even know what to say. I'm sorry that life can be so difficult sometimes. I wish I had more than superficial words to offer you, but I really do wish you only the best in life. You improved mine just by being around here. Please don't forget that.
Some pics of the "new" face of the current prototype, then replies to the above. I changed my avatar to the happy-face straight-on close-up.
Well, at least dental cleaning is something you might be able to do yourself, a bit at a time, if you have the time and patience for it, unless Lady is one of those that will eat you rather than let you help her, and has to be sedated to do the cleanings.
I'm not sure what a smaller dog would change in any of the stuff I posted about though? (other than potentially bloat, so not needing the tummytack, whcih is something that happens more to deep-chested dogs than others, but can happen to any dog of any size).
I did, but there's not much help anyone can give me on it (though from various places around the web and locally that I've asked, I get suggestions to not bother trying to do what I need it to do, but maybe just make it bark, or other similar "simple" things, which is not useful, or to just use one of the existing "robot dog" things that are absolutely NOTHING like this project and cannot do what it must do).
The problem is that if everyone just acts they way they act, and no one steps back when behavior problems are pointed out and tries to fix them, the world just sucks and gets suckier all the time. So I want to at least fix myself as much as possible, since I know that very very few others are ever going to do that.
But mostly my problems are not with people I'm friends with, it just with people that come here for help that don't know me or anyone else here. Usually I just hit the "ignore" button when they act in ways I don't like, but when they point out that I'm acting improperly, they're probably right--I can't tell most of the time so I have to trust what they say if I don't just know that they're wrong and it's just them.
Life is life; it is what it is. That's why this project is important to me so I can have something to help me with that (since other people don't usually help, just make things worse usually, and not all dogs I end up with can/will help either). I also know that it would help other people in such situations, if I can ever get far enough along with it.
You are a genuine sight for sore eyes bro.... just tickled pink to see your text
I think I understand a 'dogs' needs... I've owned one since I don't remember when. My current 'Lady' needs dental cleaning... sadly, I now live about 100 mi round trip to the closest vet... and not only is she's not accepting new patients... I'm also being quoted $500 to $900. What to do.
Well, at least dental cleaning is something you might be able to do yourself, a bit at a time, if you have the time and patience for it, unless Lady is one of those that will eat you rather than let you help her, and has to be sedated to do the cleanings.
I don't pick them; I end up with the ones that need me. Since i'm "partnered" with the local Saint Bernard rescue, it's unlikely they'd be smaller than say, Tiny, who was about a hundred pounds.
I'm not sure what a smaller dog would change in any of the stuff I posted about though? (other than potentially bloat, so not needing the tummytack, whcih is something that happens more to deep-chested dogs than others, but can happen to any dog of any size).
I try to.Hang in there AW.
Thanks.
I did, but there's not much help anyone can give me on it (though from various places around the web and locally that I've asked, I get suggestions to not bother trying to do what I need it to do, but maybe just make it bark, or other similar "simple" things, which is not useful, or to just use one of the existing "robot dog" things that are absolutely NOTHING like this project and cannot do what it must do).
The problem is that if everyone just acts they way they act, and no one steps back when behavior problems are pointed out and tries to fix them, the world just sucks and gets suckier all the time. So I want to at least fix myself as much as possible, since I know that very very few others are ever going to do that.
I do just be myself, since I can't be anyone else, but when that's a problem for enough people then I have to try to figure out how to fix it. Since they don't tell me exactly what I did wrong so I can fix it (they never understand that I can't know unless they tell me, since normal people don't have that problem), I have to figure it out myself (which is usually impossible) then try different things until it's fixed. It always bothers me when I cause other people problems, and I can't get away from it; it's always in my nightmares and my thoughts all the time unless I'm concentrating on some specific problem (which I can't do for long in isolation since in my brain everything is related to everything else, and at some point all that stuff comes back into the process too).
No, I don't try to be friends with anyone like that; it would be pointless. But few normal people even accept that people like me exist, they just think I'm normal but an asshole or whatever because I don't understand them and can't "read" them like other normal people do, and am not interested in the same boring things that most people are and can't stand spending so much time on all that when there are more important things to do and say. (though almost none of the things that are important to me are important to almost anyone else).
Um, this place is full of "people from foreign countries", and there's no detectable (by me) difference in how they accept or reject me vs anyone from this country (same in all the other places I've been).
But mostly my problems are not with people I'm friends with, it just with people that come here for help that don't know me or anyone else here. Usually I just hit the "ignore" button when they act in ways I don't like, but when they point out that I'm acting improperly, they're probably right--I can't tell most of the time so I have to trust what they say if I don't just know that they're wrong and it's just them.
I'm glad I helped.
Life is life; it is what it is. That's why this project is important to me so I can have something to help me with that (since other people don't usually help, just make things worse usually, and not all dogs I end up with can/will help either). I also know that it would help other people in such situations, if I can ever get far enough along with it.
The fingers
10 TW
Used to be able to buy war surplus dental tools to do it yourself.
Maybe a Rockchip 3588 based SBC would work. in addition to an 8 core 64bit processor it has a 6tops NPU. being arm based it doesn't use as much power as computers of the 90's.Ok. I don't yet know how to do this stuff, or which things are good for what. So whatever alternatives you (or anyone else) can think of for processing this data, I'm up for trying out.
Later floyd
like this?
not sure it would help more than the assorted computers i already have, though; i still don't know how to write software that would run on it. i'm more or less still just at the (not useful) stage where i can directly take the 3-axis sensor readouts and feed them to the servo controllers in realtime to make the servos move when the axis tilts. that is, when i can get the compiler to even run instead of crashing out on dependency failures, even though the stuff is all there where it should be (the ide is just too poorly written to look for it in the places it's already been told it's at, and there is no way for me to fix that).
in otther news, jellybeantheperfectlynormalschmoo is doing "fine" so far, but the vets don't know what causes the siezures, and it could cost many thousands of dollars (that i'll never have) to go to all the other different vets and run all the expensive tests they want like cat scans and mris and stuff, which will probably not show anything anyway, so she's just going to have to stay on the keppra anticonvulsive meds; at least that is relatively cheap, if not as cheap as the phenobarbitol that tiny was on for hers.
not sure it would help more than the assorted computers i already have, though; i still don't know how to write software that would run on it. i'm more or less still just at the (not useful) stage where i can directly take the 3-axis sensor readouts and feed them to the servo controllers in realtime to make the servos move when the axis tilts. that is, when i can get the compiler to even run instead of crashing out on dependency failures, even though the stuff is all there where it should be (the ide is just too poorly written to look for it in the places it's already been told it's at, and there is no way for me to fix that).
in otther news, jellybeantheperfectlynormalschmoo is doing "fine" so far, but the vets don't know what causes the siezures, and it could cost many thousands of dollars (that i'll never have) to go to all the other different vets and run all the expensive tests they want like cat scans and mris and stuff, which will probably not show anything anyway, so she's just going to have to stay on the keppra anticonvulsive meds; at least that is relatively cheap, if not as cheap as the phenobarbitol that tiny was on for hers.
Yes. if you don't need wifi, although it is just an inexpensive card to add bluetooth, and wifi if needed. There are several OS's that run on the orangepi 5 most are linux based. And should have the tools to take advantage of the NPU.
Later floyd
marty
1 MW
Lady was driving her car. She had a accident and drove over a squirrel. She picked up the squirrel and was not sure what to do. Across the street was a veterinarian. She took the squirrel the the vet. Veterinarian looked at the squirrel and said "The squirrels dead and that will be $50 for the examination." Lady said "How do you know the squirrels dead? Aren't you going to do any tests?". Veterinarian said OK and he took the squirrel into a room and put it on a examination table. A Labrador Retriever dog walked in. The Lab Dog sniffed the squirrel, barked and walked out of the room. Then a cat walked in the room. The cat sniffed the squirrel, meowed and walked out of the room.
Veterinarian said "Your squirrels dead and and now you owe me $450." Lady said $450? Why? Dog? Cat? I don't understand?" Vet said "$200 for the lab work, $200 for the cat scan, and $50 for the examination, and now you owe me $450.
Last edited:
nicobie
100 MW
Hey marty, quit being such an idiot.
The fact that this lovely, deserving man's creation would have applications far exceeding the monetary value of anything I currently have available is irrelevant. I'm still saddened that nobody seems to take him at his word when he simply says, "I need this."
There hasn't been so many of the "market" working robot dogs but the one company that started it all in the US is the "look up to" and standard one wants I would guess. The Chinese seems to have started building copies and copies of copies but sometime soon as their development if there is any there might come something cool out of it. For me as a hacker (of time) what I would do is use a working robot as a skeleton and then add the fluff to the outside via pins, glue and whatever other ideas that are good for that implementation. For me myself I would stretch to put a teddybear on a small toy robot just as a project and for fun.
The synergies of a working real feel dog would be a new era of "pet keeping" and the psychological effects are a bit of a unknown. I was just thinking today as I write this how other people would feel having me go around the town with a Boston dynamics level performer of a robot dog. Would they be scared or feel happy about this new entity of a being?
Unitree Go2 - The ULTIMATE AI Robot Dog is the Chinese version I just saw some payed commercial from while browsing the Youtube. Expensive. Sure, for a dog who can't walk stairs or even easy steps. Could it work as a fake guard dog, probably? Still I don't know about the battery time and I don't think it can charge itself yet as autonomous electric lawnmowers can.
The synergies of a working real feel dog would be a new era of "pet keeping" and the psychological effects are a bit of a unknown. I was just thinking today as I write this how other people would feel having me go around the town with a Boston dynamics level performer of a robot dog. Would they be scared or feel happy about this new entity of a being?
Unitree Go2 - The ULTIMATE AI Robot Dog is the Chinese version I just saw some payed commercial from while browsing the Youtube. Expensive. Sure, for a dog who can't walk stairs or even easy steps. Could it work as a fake guard dog, probably? Still I don't know about the battery time and I don't think it can charge itself yet as autonomous electric lawnmowers can.
this isn't a thread for discussing stuff like that
it's a thread for helping me make mine
other people's or companies stuff hasn't got anything to do with mine and doesn't help me make mine since nothing does even remotely what i want or i wouldn't have to make it myself in the first place
if posters want to discuss that stuff go make a thread for it elsewhere
only post here if you're willing to help me do this project the way i need it to work
i expect that means i'll be the only one posting (very infrequently) here, but at least it will keep the thread clean for documenting it
it's a thread for helping me make mine
other people's or companies stuff hasn't got anything to do with mine and doesn't help me make mine since nothing does even remotely what i want or i wouldn't have to make it myself in the first place
if posters want to discuss that stuff go make a thread for it elsewhere
only post here if you're willing to help me do this project the way i need it to work
i expect that means i'll be the only one posting (very infrequently) here, but at least it will keep the thread clean for documenting it
if i ever get far enough to do anything ai related, are systems like any of those that may be listed here
useful for running them?
i can't afford the hardware yet but maybe by the time (years?) i would have something to run on it it might be cheap enough to get, or budget might've improved, etc
Gpu Server for sale | eBay
Get the best deals on Gpu Server and find everything you'll need to improve your home office setup at eBay.com. Fast & Free shipping on many items!
www.ebay.com
i can't afford the hardware yet but maybe by the time (years?) i would have something to run on it it might be cheap enough to get, or budget might've improved, etc
No significant progress on the software front.
Hardware-wise I found that some 8mm rod-end (rose-heim) joints have about the right lateral (non-rotational) movement range to emulate most of the joints of a wolf (dog) skeleton that aren't ball joints, and that two of them in "series" will emulate the hip joints better than a simple single ball (as far as complexity of manufacture for me, vs required movement ability).
So I picked some cheap ones up from aliexpress for less than a couple bucks each, and am just waiting for the cooler weather (whcih did arrive and then changed back to realy hot before I could do any work) to be able to do the metalwork (cutting, welding) outside during the daytime when I can clearly see what I'm doing.
Have had a couple more people "offer" help with the programmng then stop replying to me as soon as I describe what's involved; seems there's a pattern.... But I stopped expecting direct help with this project, so whatever. I'll get there eventually on my own; I just don't have time to help with anyone else's stuff until then.
In between my fitful attempts at figuring the software stuff out, I've destressed by working on my music, and have made a few new songs, which are linked over here in this thread:
endless-sphere.com
Hardware-wise I found that some 8mm rod-end (rose-heim) joints have about the right lateral (non-rotational) movement range to emulate most of the joints of a wolf (dog) skeleton that aren't ball joints, and that two of them in "series" will emulate the hip joints better than a simple single ball (as far as complexity of manufacture for me, vs required movement ability).
So I picked some cheap ones up from aliexpress for less than a couple bucks each, and am just waiting for the cooler weather (whcih did arrive and then changed back to realy hot before I could do any work) to be able to do the metalwork (cutting, welding) outside during the daytime when I can clearly see what I'm doing.
Have had a couple more people "offer" help with the programmng then stop replying to me as soon as I describe what's involved; seems there's a pattern.... But I stopped expecting direct help with this project, so whatever. I'll get there eventually on my own; I just don't have time to help with anyone else's stuff until then.
In between my fitful attempts at figuring the software stuff out, I've destressed by working on my music, and have made a few new songs, which are linked over here in this thread:
Amberwolf's Music Studio Technical Stuff
Would like to listen, but there's nothing there. Must've been a glitch when you tried; it was working before and it's working now. ;) (or else there's something between you and the server keeping you away from it). There's also all the other stuff still on the main Amberwolf page...
endless-sphere.com
I got the worktable cleared off this past weekend (it had been filled with the 3D printer and associated stuff), by building a new table for the 3D printers and stuff out of a shipping crate. So now I have a space to work on (and leave setup) some of the wolfy-bot related project pieces, *and* a dedicated space to work on 3D printed stuff (some of which is for the wofly, most of which wont' be). (see the What shall we do with a 3d printer thread for pics, etc)
Should make it easier since I rarely have consecutive days (or even weeks) where I can go back to something, either because there isn't time or energy, or because I haven't learned whatever new thing(s) I need to in order to continue.
Should make it easier since I rarely have consecutive days (or even weeks) where I can go back to something, either because there isn't time or energy, or because I haven't learned whatever new thing(s) I need to in order to continue.
"what about moving to a higher level with python instead? The devt environment needs less setting up than for Arduino and python's expressiveness is better for data crunching"
I should have mentioned that microcontrollers can run micropython interactively. You code directly on the hardware and get immediate feedback, giving a 'malleable' feel different from using a compiler and with less to go wrong.
- getting started
- coding demo for an IMU to control a servo (7 minutes)
MicroPython - Python for microcontrollers - supported hardware
Buy well buy once, the smaller the microcontroller the more work it is to make it consistently useful.
I should have mentioned that microcontrollers can run micropython interactively. You code directly on the hardware and get immediate feedback, giving a 'malleable' feel different from using a compiler and with less to go wrong.
MicroPython - Python for microcontrollers - supported hardware
Buy well buy once, the smaller the microcontroller the more work it is to make it consistently useful.
aside from actual coding and development help from other people, what i really need is a well-designed dev environment ui, thought out and created for the end user (not the programmer that created it) to navigate around, to set it up in just one place, that knows where all it's libraries are, that provides useful and helpful diagnosis of problems with feedback to the user that actually helps the user find and fix the problems.
arduino most definitley is not any of those, though for someone that already knows coding in the language(s?) it's using it's probably good enough.
also if python is realtime compiled as it runs, it means it takes more memory and cpu time to do, so it takes more of a computer that uses more power to run it to do the same job as something that's precompiled as machine code. unless the way this is done has changed since i started learning what littel i know of coding, many years ago.
what i need most is direct actual help from people that are willing to work with me on creating this in the way that i need it to work. so far that's a bust everywhere i've tried.
(at best i get people helping me figure out what i might be able to use to do it, which is good...but mostly i get people telling me i don't need to create this thing at all, and that i should just use some other existing "dogbot" that isn't anything like a dog. or people that tell me i shouldn't even be trying to make a companion robot but should just deal with my "problems" and live alone and be lonely and stressed out all the time.
not really for my purposes. bedause of what i need hte sytem to do, a single computer that does eveyrhting would be too expensive and too difficult to deal wiht.
breaking it down into small jobs that subprocessors can do, and pass data up/down the chain, means that the main system can be much simpler and only do main decision making. it also means i can make subsystems that do various jobs as i learn how to do them, affordably for me.
all the stuff for processing the local data inputs to filter for the main system, then to take commands from the main system and do little local controls, can be done by super cheap small systems.
a much bigger system, like a "desktop" or rack system with a lot of memory and cpu power, can have the imu sensor network data passed on to it, after it's been prefiltered by the network of smaller mpus, to do all the things I want it to do. I'm pretty sure from comments by various people that none of the smaller stuff including hte rpi could do it in realtime as required.
maybe a smaller rpi or even part of that same desktop/rack could do the audio processing for speech recognition, and then generate (from "phonemes") the sounds (not speech) needed as responses or just general behaviors.
the individual small sub processors can deal with reflex movements that have to be fast, like blinking an eye or twitching an ear in response to a touch right there, if i can work out a locallly-processed touch detection that doesn't require waiting for the main imu network to figure it out. but even if local imu processing can't be done that way, the local mpus can still handle all the detailed motion control. (then the imu network will tell the main system that stuff is happening, and it can work out what else to do from there).
it will probably require another separate desktop/rack system to do the visual processing if i ever get far enough to do that stuff (unlikely).
**** (after covid some costs of living have doubled or tripled, and pay hasn't gone up, actually gone down a bit in that there's a few less hours of work available, so i can no longer easily save up for things like i was jsut beginning to when covid started. now it takes much longer to save up for even smaller amounts, and i spent what savings i'd already had on helping jellybeantheperfectlynormalschmoo earlier this year).
Last edited:
(I think you edited this post and my reply about programmers got wiped?) Anyway #2...
Sensory processing like speech recognition and vision need oomph, the rest not so much. Try hard to ditch preconceptions about how to build. Breaking a task down helps learn but that can happen stepwise in s/w, it doesn't require small h/w. Micropython will need a bigger processor and that's what you want: if devt takes an extra 3 hours in Arduino that's another sandwich gone and a bigger processor would have been the lesser investment. Co-ordinating networking is a significant overhead and the job would be easier with all the fast movement, servos, imus etc. handled by one up to date micro from Adafruit or Raspberry Pi Pico series, figure around $20 for such a board e.g. Adafruit Feather RP2040
Sensory processing like speech recognition and vision need oomph, the rest not so much. Try hard to ditch preconceptions about how to build. Breaking a task down helps learn but that can happen stepwise in s/w, it doesn't require small h/w. Micropython will need a bigger processor and that's what you want: if devt takes an extra 3 hours in Arduino that's another sandwich gone and a bigger processor would have been the lesser investment. Co-ordinating networking is a significant overhead and the job would be easier with all the fast movement, servos, imus etc. handled by one up to date micro from Adafruit or Raspberry Pi Pico series, figure around $20 for such a board e.g. Adafruit Feather RP2040
no, i haven't edited anything?
you sent me a pm about some stuff i haven't read yet; was your reply about programmers in that?
the imu processing will need it too, according to what little info i've been able to get from people about what i want to do with that data. it has to realtime process the full 3-axis accelerometer data from at least a dozen (probably more than two or three times that many, as some will just be under the fur and not attached to skeleton, some places get more of them, some not many) imu's as if it were an earthquake network, along with the full 3 or 6 axis position/velocity data from at least a dozen of those imus (one in each segment of each limb, plus at least a few along the spine and ribs, at least one in the head, and at least a couple in the tail).
what it has to do is first map all the position / velocity data to keep track of where every part of the wolfy is. that is used for both motion control and for touch feedback.
teh touch feedback is used with the touch detection which is done via the accelerometer data on the imus detecting every vibration. it has to first filter out all the extraneous data such as motor/skeletal vibrations from moving, by comparing sensor data from different ones and eliminating stuff that's identical in those in the same area or connected to the same part.
it has to compare the same touch detected on multiple sensors to find out the actual touch location, like a seismometer network is used to determine the surface coordinates and the depth of an earthquake. this is needed to determine what kind of feedback to provide, or what action to take.
then process the vibrations to see which ones might be touches from the user, like petting, etc, and which kind of touch they are--longer ones are petting, very short ones are just touches, harder short ones might be a pat, really hard short ones are an impact meaning it probably ran into something and needs to stop moving in that direction, etc.
then use the positional data tagged on these to tell where the touch was so it can determine what response to make. like an eartip touch means it tickles and so it flicks the ear or ducks it's head, a nose touch is a boop so it pulls it's head back, an eye-region touch makes it close it's eye, a paw touch may make it lift the paw and wave it at you or pull it back, depending on the kind of touch. a petting touch on the ear or face would make it lean the head into the touch, etc. a really hard one from the tip or bottom of the paw probably means it whacked the user or something and needs to pull the paw back to it's body and maybe "apologize". basically, do what a dog would do.
the imu sensor network is probably the most important thing to do, since general robotics has been done forever and is "easy", other than the mechanical way i'm doing it which isn't as common since i'm having all the motors in one external (big) box that is pulling cables to make it move, instead of haivng any of htem on the things that move, so the whole wolfy can still be squishy and relatively light, and not making a bunch of wierd motor noises inside. but anyway if the wolfy can';t sense touches and phsyical interactions at least similarly to a dog, then it can't react like a dog and it won't be a dog even if ti looks and moves like one. there are no sensors that can work under the fur to do this kind of thing except the accelerometers like an imu has, since those are incredibly sensitive and will detect all sorts of tiny little vibrations and can reprot which axis they are on, too
before imus were really cheap and common i was going to do it with microphones but that was a huge engineering problem because of electrical noise, induction in wires, filtering, etc., just to even get that turned into useful data i could then process for signals, and i couldn't do it. as soon as i realized it could be done with imus/accelerometors and saw their prices coming down so fast, and all the user-created projects with them, i could see this was the real solution to the problem.
if i can someday figure all that out, i'd also like to make a learning program (some form of ai as they call it nowadays) that lets the user teach it behaviors and responses like you can a dog. some things would gbe preprogrammed but overridable by teaching it differently if you wanted to. other things it wouldn't "know" at all, and it would learn what you like (respond positively to) and what you don't (respond negatively to). your responses could be verbal "no" for negatives, or verbal calling it's name or good dog or whatever for positive, and it would seek to get the most positives it can just like a dog will. it can also learn that since pets and touches are also positive, it should do things that get it more of those, unless you tell it to stop, which won't be a negative, but just a "dont do that right now" thing. that's probably far in the future since i don't even know how to do the imu stuff yet.
well, we'll have to see how it all works out, since i have the stuff i have, and have to use what i have for now. the money was already spent on those, so that's how i have to work for now. as i said beofre in either the rreply abvove or in the reply to your original pm i already have a tiny rpi to experiment with.
edit--this one
was about 12 bucks for the pair, now the page only sells one for almost 14 bucks.
Raspberry Pi Pico Development Board, Raspberry Pi RP2040 Dual-core ARM Cortex M0+ Processor, Running Up to 133 MHz, Support C/C++/Python, 2MB Quad SPI Flash Integrated with SPI/I2C/UART Interface 2pcs
if it turns out that it doesn't work this way, then i have ot do it a different way. i probably won't know for a long time how it's going to turn out, at the present rate i would guess years, but it's been decades getting to this point already just waiting for stuff to do this to be affordable for me, and learning enough to even begin.
if you know anyone that can and will directly help with creating this stuff then send them my way.
otherwise i'm going to keep plugging away at it as i have time and energy and patience.(i don't learn well by studying and reading up on things like other people do; i have to do things in tiny chunks and experiencing the results, seeing what happens, working with things. so i read up and study what i can, but i have very little ability to focus on those for very long, since i'm not interacting with anything. i can sit and just read a book tha'ts a story about people and things happening just fine, but when i start learning things i have to do something with them and i'm usually too tired to manage that, so it's usually just pure frustration to learn about things i can't do anything with at the time. also when i don't understand some part of what i'm learning and can't find something that properly explains it in a way i can grasp...those are one reason i wish i could work with people on this project because the cross-discussion can help me figure things out and it also makes me think about things.)
like talking about this with you, telling you how it needs to work, makes me think about things and i begin visualizing stuff in my head whcih is harder to do otherwise, since i already know how it needs to work and so i don't think about it further which means i jsut have the original visualizations that probably could be improved on but my brain doesn't do that because it doesn't need to. if that makes any sense (probably doesn't).
Last edited:
IMUs: I gather they're prone to drift (and famously noisy) so they might not be the best or only way to track skeletal positioning. Also the need for so many, while it attractively mimics nature, might be a hint that there are better ways in practice. Sensors are a boundless garden so keep an open mind about that aspect. Example: how about RF tags dotted around whose distances are read in turn by 3 readers in the body cavity, followed by triangulation taking account of known limb movement constraints (hence not needing 4)?
Many of the desired movements and senses are relative: an ear-flick only needs to be 'several' degrees, or are dynamic: severity of skin touch falls into one of a few categories from light to hard. That suggests low data rate, tailored solutions could be sufficient. Also times move on, compute is cheap and electrical noise in microphones is probably a thing of the past in something the size of a dog. (More in PM)
Design: It's useful you having described how you think about things. Your hands on learning approach is good for a large project because you can adjust course as it proceeds. Malleable development using an interpreter on bare h/w sounds a good fit.
Dev board: your Pico boards are a great start because
- RPF support and training materials are among the best
- They support the more capable full MP MicroPython - Python for microcontrollers
- The IO processor's quite quick and both these boards and the wifi version should be reliable
- There are lots of peripheral offerings for them
- Versions with more storage are cheap on AliExpress
- MP makes for a good hardware upgrade path e.g. to STM32
- There are several ways to speed up MP
How far have you got with them?
Many of the desired movements and senses are relative: an ear-flick only needs to be 'several' degrees, or are dynamic: severity of skin touch falls into one of a few categories from light to hard. That suggests low data rate, tailored solutions could be sufficient. Also times move on, compute is cheap and electrical noise in microphones is probably a thing of the past in something the size of a dog. (More in PM)
Design: It's useful you having described how you think about things. Your hands on learning approach is good for a large project because you can adjust course as it proceeds. Malleable development using an interpreter on bare h/w sounds a good fit.
Dev board: your Pico boards are a great start because
- RPF support and training materials are among the best
- They support the more capable full MP MicroPython - Python for microcontrollers
- The IO processor's quite quick and both these boards and the wifi version should be reliable
- There are lots of peripheral offerings for them
- Versions with more storage are cheap on AliExpress
- MP makes for a good hardware upgrade path e.g. to STM32
- There are several ways to speed up MP
How far have you got with them?
i have no idea how you would be able to make that work
can you point me to a working system that can tell exactly where an rf tag is and how it is moving in a space like this? that has schematics and code etc?
if not, it's just another thing i'd have to work out and design, and i'd much rather do what i already talked about because it would be far easier to do; i already understand hoiw it works and there is nothing to interfere with the imu readings like there would with rf stuff since there would be lots of rf noise from motors and power supplies and stuff to interfere with rfid or any other radio stuff
thats why i had to sklp the many touch sensors that might have been able to detect a touch; they;re not reliable under those conditions plus they interfere with each other hwne they're too close, and i have ot have them all over to make the touch zones overlap enough to have full coverage over the whole body.
also how would rfid tags be able to detect the touches on the fur?
unless you have a sensor design that isn't on the internet anywhere, there is no sensor to do that.
the only sensor that could do it, at least potentially, is the imu becuase the accelerometer can be sensitive enough to detect all sorts of vibrations; this is one way that mems microphones work.
as fqr as using mics to do it directly, as noted before it isn't practical because it requires too much audio engineering to prefilter everything and then to "see" all the signals in the noise. mems microphones might help bedause theyre alread y digital outputs but they're not as flexigble in data output choces as the imus
the imus / accleerometers are already digital and have internal stuff you can program as prefilters and data sorters to feed you the data you want out of it, or you can read the raw data if you need it, etc.
i've put lots and lots and lots of thtought into how this thing needs to work, and it has to do what i have stated here in the thread, plus other stuff i probably haven't written about yet, but if it can't do the stuff there is no point in making it at all
if all i get is the same crap that is nothing like a dog that already exists, why bother?
waht it comes down to is i don't need alternative ideas for what the wolfy will do....i need help implementing the required things it has to do, in the way that it has to do them.
if an idea for implementing it won't do what it needs to do, it's not a useful idea
i'm not interested in a partial implementation
that would be like building myself a cargo trike without any wheels; it wouldn't be useful at all
Tags were a poor example on my part, just to explain a basis for reducing sensor count. I wasn't clear how far your design's progressed and assumed that IMUs must be slow because accelerometers were when I used them - I fell into exactly that trap of forgetting how fast this stuff improves. I wish I knew someone who's worked with IMUs but not so far.
