My Advanced Realistic Humanoid Robots Project

[artbyrobot]

@wobbly

You just betrayed a huge misunderstanding of my system plans.  You think my windows 7 main loop will be polling a rotary encoder.  That is too low level for a windows 7 main pc.  That is the work of a microcontroller.  Microcontrollers interface with all sensors and aggregate findings and report to the main windows 7 pc their findings via usb.

[artbyrobot]

Robot update:

Ok so my solution to the issue I had of fishing line breaking when tested by hanging a 10lb dumbbell is finally here folks.

The solution is to sew a fishing hook's eye into the bone sleeve snugly with upholstery thread as a anchor point. Then I will draw my braided PE fishing line through this eye and back down. Instead of tying it off with a fancy knot which acts as a weak point or concentrated stress point, I will use a fishing crimp sleeve to crimp the rope off on itself. Similar to crimping two pieces of wire to eachother with a electrical crimp tube. Supposedly fishing crimp sleeves are used to avoid knot tying and offer even more integrity than a knot can while maintaining fishing line integrity more than a knot can. No weakness is introduced to the line like knots do. A side benefit is this crimp also protects the line from abrasion and acts as a physical standoff so the line isn't rubbing the bone sleeve as much which can cause micro abrasions and weaken it over time. I bought #2 and #3 fishing crimp sleeves which were around $6/100pcs on amazon.

[wobbly]

OK fine I give up, someone give me a bible.  I just want the world to make sense again.    ****ing hell.

[xrunner]

OK fine I give up, someone give me a bible.  I just want the world to make sense again. 

Well I suggest going fishing - have the OP send you all the unused fishing line and hooks you will need! 

[MK14]

I think most people around here would completely disagree with all of that.  Please correct me, people, if I'm wrong.

The main loop of an embedded system that uses Interrupts (sometimes many different types) needs to be written in such a way that Interrupts are not disruptive.  This is a well understood practice and widely adopted.  Once achieved, this allows very rapid response to external (or internal) stimulus.  It also means that polling is not needed - this saves valuable time in the main loop because polling is waiting for something to happen.  Repeatedly asking is it ready... no...is it ready... no...is it ready... yes... is wasteful of time and CPU resources.

Polling absolutely is burdensome, and should be avoided in time-critical systems such as the one you are contemplating.

There are a lot of different ways, of designing and building, complicated real-time embedded systems.

Although interrupts can be very tempting, and useful.  Things can also, be done, in a way, which minimizes their use.

(As I suspect you already know), interrupts are a mixed blessing.  A bit like junctions, on motorways (Freeways).  As useful as they are, they can also cause major headaches (bugs and other problems), such as the somewhat random timing delays they add to the system (jitter), and they can make it much harder to quickly test systems out.

E.g. Two specific interrupts, if they occur simultaneously, may crash the system.  But such an issue, may only occur, once a week (if you are lucky/unlucky).  Making testing harder, and reproducing such a problem, tricky.

Yes, their system could be designed to use a large number of interrupts.  But that would NOT be the only way, of solving the issues needed to create their Robot system.

EDIT:  Accidentally put wrong quote in, corrected it now.

[artbyrobot]

@MK14

your post appears to have come too close to taking Artbyrobot's side on this matter and so it has been flagged as suspicious.  It is common knowledge on this forum and thread that all posts written by Artbyrobot are stupid and ill-informed and so any attempt to pretend to agree with Artbyrobot will be treated with strong prejudice and scorn.

[wobbly]

@MK14, I completely agree.  Horses for courses as they say.

I was simply staggered at the OP's remark "[interrupts] are bad.  They should NEVER have been invented".

[artbyrobot]

@wobbly

You say "I was simply staggered at the OP's remark "[interrupts] are bad.  They should NEVER have been invented"." --- well when you are talking or thinking deeply on something important, I'm sure you don't like to be interrupted.  I suggest you treat your main while loop with the same respect and wait till it is giving some issue eye contact before speaking to it rather than interrupting what it is doing which is very disruptive and unpleasant.  Main while loops deserve more respect.

[xrunner]

I still don't think you are aware of the sheer difficulty of your project no matter how forcefully you write or try to "correct" people here who are educated in engineering far beyond your level. As I tell my friends when they try to gaslight me "Your Jedi mind tricks don't work on me" and yours aren't working on me either. And I don't want to discourage anyone from doing anything that they enjoy. However lets proceed ...

So going back in the thread I took a screen shot of the arm of your design, in which you try to mimic the bones of the human anatomy. But of course you can't make a simulated human "muscle" So you have fishing line pulleys for tendons and motors for muscles. You think this will be able to duplicate the human arm functionality. I also have shown you that this can be done already without using a reconstruction of human bones and the unneeded complexity of all that biological baggage. It isn't needed.

So here's what I want to see from you. Forget all the other things like these -

The rest of the robot body
AI / Brain / thinking
Sight
Hearing
Touch
and anything else besides the raw mechanical arm with all that you think it needs to operate like a real arm.

Build the arm and mount it on a test table that can hold it so it can move as it should. If that includes a few more motors and pulleys at the "shoulder" then build them. What you are going to do is write a program to do the following:

Set up a coffee cup full of water
Set up three other empty glasses
Write the following test program -

Program the arm with hand to pick up the coffee cup by the handle
Have it move to each empty container and pour out 1/3 of the liquid into each one
Bring the coffee cup back to the original position, while the hand turns the coffee cup upside down as it is moving (I can do this with one hand) and set it down on the table.

You will not use any AI or sight system but simply hard-code all the movements as a test trial of the mechanical baseline capability - can the arm do what I program it to do, or is the whole mess an unstable design that will not be acceptable for a robot?

You will do this by trial and error and making measurements of the distances and twists and turns the arm and fingers must make to complete the task, and using those as pre-determined data points in the program. Do you implement feedback in any form so you know that a certain part has actually moved? If so use it. Remember this is a test of the arm's function - not of anything at any higher level in the system as a whole.

I'll check back in, say, ten years and see how you are doing. My prediction however, is that you will not be able to get the arm to do this task because you will soon find the control issues will be to complex for you to handle and that the design (using human bone structure etc.) is too unstable for it to work. Only then will the sheer complexity of your task hit you really hard. But I hope you will have had fun trying and you will have learned a lot.

Now I'll start your next post for you because I am truly a learning system myself -

"I disagree ..."

[MK14]

I still don't think you are aware of the sheer difficulty of your project no matter how forcefully you write or try to "correct" people here who are educated in engineering far beyond your level. As I tell my friends when they try to gaslight me "Your Jedi mind tricks don't work on me" and yours aren't working on me either. And I don't want to discourage anyone from doing anything that they enjoy. However lets proceed ...

One possibility, that comes to mind.  At least for people, who are NOT the OP.
Is that technical people, tend to have a raised possibility of being, at least a little bit, autistic in various ways e.g. ADHD.

Which can affect that persons, ability, to properly access/perceive, the flow of time, how long complicated technical projects will take, and other practicalities, of massive projects, like the one in this thread (Robot).

But possible lack of (relevant) experience, can also worsen the accuracy, of peoples estimates on how long and how complicated, something will need to be.

The story goes (but it was probably true, but I might be remembering it partly wrong), a company decided to cheap out use a very recent graduate, to create their software.

The boss was delighted, when, 1 week into the software task (which MUST be completed, and ready, within 4 weeks of the original start date), the recent graduate, reckoned they had completed 90% of the software.

The remaining '10%' of the software, only took 11 more weeks to complete.

It is NOT easy, even for hardened and highly experienced experts, to accurately estimate how long some technical things, are going to take to complete.

IIRC, new software has a terrible reputation for being late.  Compared to most other industries and things (Projects).

[artbyrobot]

@xrunner

You say "I still don't think you are aware of the sheer difficulty of your project" --- I disagree.  In fact, I chose it precisely because it was so difficult and anything less difficult would be lame by comparison.  It is precisely the staggering difficulty that allows for the measure of a man and makes it fun.  It is the ultimate test of one's abilities.  For someone who seems to have boundless potential in their own sight, a proper challenge is needed to find out what ones limitations actually are.  I have not found many limitations so far because no challenge in life has been enough for me.  I need bigger challenges to find my limits.

You say "people here who are educated in engineering far beyond your level." --- each may be more educated vertically, but need not be more educated horizontally necessarily.  And when we move outside of your various niches of expertise, this is where lacking is often discovered.  To have width and depth of education in engineering is rare.  You assume if one has depth in one niche they are the expert on niches outside their domain as well it seems.  NOBODY on this forum is educated in engineering a humanoid robot that is a humanoid form factor and lifelike that moves realistic and talks realistic and has advanced AI and can play sports, do chores, make more robots, etc.  This doesn't exist so we are mostly all on even playing field although likely I'm the leader in this field as I've been studying this for over a decade where most of you are just considering it.


You say "this can be done already without using a reconstruction of human bones and the unneeded complexity of all that biological baggage. It isn't needed." --- it is needed if you want the same range of motion and degrees of freedom while retaining the structural integrity and durability of a human body as well as the ability to place, angle, and position the muscles in such a way that you have a proven reference point for everything which the human body already provides.  Once you step outside of this, you introduce many unknowns and can no longer use the human body for reference on all questions and gain from its wisdom built into it.  So then you hamper your progress and shoot yourself in the foot by NOT taking this proven design and just copying it to save all that trial and error and learning the hard way why your new design fails.

You say the control issues will be too complex for you to handle.  I disagree.  It really isn't complex at all.  You just create a rotating magnetic field using a dual h bridge to your bldc motor and then you measure the joint angle changes this outputs as the bldc motor winches the fishing line in which actuates the joint in question.  Easy as pie.  Literally children do this stuff.  The harder part is designing the robust downgearing system since I didn't want off the shelf noisy metal gears ruining everything by the noise.  And it seems I might be nearing success on that stuff.


You say "the design (using human bone structure etc.) is too unstable for it to work." --- I have no idea why you'd think that.  Sounds ridiculous.  Obviously human bone structure is not too unstable because just look at your own body.  It does work.  So its already proven.  We aren't seeing if the human body's skeletal and musculature systems are too unstable.  We already know they are stable enough.  That is not the question.

You say "Only then will the sheer complexity of your task hit you really hard." --- not sure why that would change my view of the complexity.  But maybe.  Doubt it though.

You say "But I hope you will have had fun trying and you will have learned a lot." -- well I appreciate that.  Those are kind words which are rare on this thread so I appreciate that.

You say "Now I'll start your next post for you because I am truly a learning system myself -"I disagree ..."" --- I guess you must have a magic ball in order to see into the future with such accuracy like this.  You were right!






@MK14

I'm not sure what autism has to do with any of it, and I don't know if I have some autism in some areas or not.  Both my wife and mom who are mental health professionals with degrees in it did not diagnose me as having autism despite knowing me for decades.  But anyways, I agree that estimating how long a project will take is hard if you never did that project before or anything similar.  It's practically impossible.  Factors like real life getting in the way and not knowing how much time you will have to put into it make it even more complicated and full of unknown variables.  Not knowing if you will outsource parts or use open source outside hardware also throws unknowns.  But xrunner's point was I don't realize how hard this project will be whereas your point is I don't know how long it will take.  I find these similar but distinct topics.  They may have overlap but aren't the same thing really.

[wobbly]

Regarding timescale and project cost estimation in Electronics and Software...

Off-topic anecdote time...

I started a new job in 2006 as a software engineer in SAAS and electronics company (telecoms sector).  My boss had a little velvet drawstring bag on his desk with some Dungeons and Dragons dice in it.  D4 up to D20 etc.  After a while I plucked up the courage to ask about the bag and if he was a keen D&D player.

"Ahh, I wondered when you'd ask about that", he said with a grin.  "No I don't play D&D any more, although I was a keen player a few years back.  These days I use the dice to pad out timescales.  I'll take my own best estimate for a project's duration and then roll a few of these dice and add that many weeks to the project."

I could tell he was absolutely serious.  He did have a very good sense of how long a project would realistically take and so the dice only added a bit of extra breathing room.

A few months later he passed on another bit of advice...

"If you ever have to estimate the cost of something, give it your best estimate and then multiply that by "e".

("e" being the base of natural logarithms, equal 2.718... an irrational number).

The reasoning being that if you send a quote to the customer and it's something like $1,000,000 dead-on then they are clearly going to think you just plucked a number out of the air without thought.  But saying "it'll cost you $1,013,417.33" apparently looks favourable with the beancounters.

I have no idea if it worked or not but we never went overschedule and the customers never baulked at our quotes for the 15 years I was there!

Not sure if this is commonplace but it sounds like something out of the MBA playbook to me.

[xrunner]

@xrunner

I disagree. 

Well I predicted that one 100% accurately. 

Your word salad wasn't tasteful though.

So can you do the task I suggested? When will we be able to see the arm on a test rig and do the things I suggested?

Another old but useful phrase that I use from time-to-time

"The proof is in the pudding"

[MK14]

But xrunner's point was I don't realize how hard this project will be whereas your point is I don't know how long it will take.  I find these similar but distinct topics.  They may have overlap but aren't the same thing really.

You are right, they are NOT the same thing.

But, usually, if something is extremely technically hard/difficult/tricky.  Then creating that part of the project, including any software it requires and getting it working well.  Can take considerably longer, to do.

E.g. If all you need to do, is flash an LED at some slow rate, to indicate something to a user.  It may not take very long to wire up and use an NE555 timer, or LED with built in flashing circuitry, or even writing some simple software, to flash the LED on and off, at a sensible rate.

But, if instead, it turns out you need to drive a large and complicated screen.  It may need a GUI, which could involve learning how to do it well.  Along with other challenges, in order to complete that part of the project.  Such as finding out how to program an FPGA, to interface between your hardware and the screens, rare/complicated interface (because the screen of the right dimensions, shape, and brightness (for outdoor daylight operation)/clarity for your Robot, might have this rare interface type, built into it, rather than more common/standard ones, such as HDMI).

In other words.  Hard technical problems, can take a (very) long time to solve.

[artbyrobot]

@wobbly

Cool story wobbly but I don't think any of it applies in this case.



@xrunner

You say "So can you do the task I suggested?" --- no.  I would not waste time hard coding movements other than finger forward and backward joint by joint just ensuring everything moves right and testing speeds.  To integrate all the joints into complex movement patterns with full inverse kinematics and motion profiles harmonizing all of it will be the work of the AI.  Hard coding that is a waste of time.

You say "When will we be able to see the arm on a test rig and do the things I suggested?" --- I at one point on my journey did plan to mount the arm to a 5 gallon bucket and just have webcam eyes on a pole sticking out the bucket and this would be the test rig.  But then everyone started calling it the bucket bot and it became a meme.  I did not like this.  I want people to visualize the final realistic humanoid robot so I won't build a test rig at all.  When the arm is ready, I will mount it to the torso skeleton and actuate it more from there.  The torso is very involved with the arm anyways as it contains the liquid cooling and air cooling pumps, the batteries, the main windows 7 PC, and even has some of the motors that I just couldn't fit into the arm but are needed for the arm.  So no test rig.  The torso is too interdependent and so the torso will be the test rig.  As far as when.  I have zero clue.  Maybe 7 or 8 years I hope.

[xrunner]

You say "So can you do the task I suggested?" --- no.  I would not waste time hard coding movements other than finger forward and backward joint by joint just ensuring everything moves right and testing speeds.  To integrate all the joints into complex movement patterns with full inverse kinematics and motion profiles harmonizing all of it will be the work of the AI.  Hard coding that is a waste of time.

So you are going to build the entire very complex arm with fishing line pulleys and motors and then just proceed with the rest of the project. All this without knowing what mechanical issues that the arm has which would have been revealed with thorough bench testing. No sane engineer would do that.

The AI cannot fix the internal mechanical issues that the arm will have (and it will have many that you cannot foresee). It can only work with the mechanical device it is presented with. Any mechanical internal design flaws are out of it's scope of change. Those are YOUR problem as the designer of the arm. And you blatantly state you will not test a complete arm before proceeding with even more complex parts of the design.

IMHO - You are very clearly incapable of doing a project of this magnitude. This realization will hit you sooner or later, like Co. Kurtz said in Apacolypse Now "Like a diamond ... a diamond bullet shot right through my forehead". It will eventually hit you that hard, and you will slowly fade away from it and do other things.

I have my own suspicions of why you are doing all this but I better be quiet about those here. 

[artbyrobot]

@xrunner

You say "So you are going to build the entire very complex arm with fishing line pulleys and motors and then just proceed with the rest of the project." --- No, I said I will test the full ranges of motion back and forth of each joint.  I will also test the speeds so do these slow and fast.  I can even test with weights applied to test load bearing capacities.  I can even test to failure to know limits.  What I will not do is a complicated movement sequence of the whole arm plotted by hand hard coded which will not add to my knowledge of the function of the arm in my opinion in any way compared to the tests I already pointed out. 

You say "The AI cannot fix the internal mechanical issues that the arm will have (and it will have many that you cannot foresee)." --- I didn't say it could.  However, any looseness or bouncyness rather than rigid and tight precise movements the AI can accomodate for.  Humans move this way just fine unlike industrial machines.  But AI can account for this.

You say "you blatantly state you will not test a complete arm before proceeding with even more complex parts of the design." --- I disagree.  I will test each individual joint of the arm as I complete that joint one at a time.  But to test alot of it the arm has to mount to the ribcage first.

You say "You are very clearly incapable of doing a project of this magnitude." --- I disagree.  In fact, the only possible way that's true is if my time runs out aka I die.  And if I die too soon, then I'll just finish it in the afterlife if God lets me continue working on it and I don't see why He would not.  So either way it will get done IMO.  At that point I'd have zero time restriction.

You say "I have my own suspicions of why you are doing all this but I better be quiet about those here." --- I was doing it before I decided to share it on some forums for feedback so no, it's not just for feedback or donations or w/e.  I actually like robotics and have wanted to do this for years because of robot movies and stuff.  They are cool.  Also, it's not a sex bot if that's the other angle you are going with.  I think that's sinful.

[KerimF]

Let us recall that “Necessity is the mother of inventions/innovations”.

We may say that the true level of one’s intelligence shows up only whenever he needs really to do/have something new (new to him, if not to the world). And not having it would be out of question for him, unless he dies before finishing it.

In reverse, without having real interest in doing something, one’s failure (in doing it) is possible, almost always, no matter how intelligent he might be.

I personally lived both cases several times.

[wobbly]

And if I die too soon, then I'll just finish it in the afterlife if God lets me continue working on it and I don't see why He would not.

I wish you had said this at the beginning.  That changes everything!  Now we have an infinitely long timescale
Fair enough, this opens all sorts of avenues hitherto unexplored.  Bravo.

Edit: This also suggests that the budget is now unlimited too, correct?

[Mechatrommer]

@Mechatrommer

You say "i would love to see motors mechanism of arm and fingers alone i suspect that will be super bulky." --- it really is super bulky.  Which is why I had to spread them out throughout the forearm, upper arm, and even into the sides and back of the robots torso just to fit it all.  Like I literally have some finger actuation motors mounted onto the sides of the spine since there were too many motors to even fit them all in the arm.

i would love to see what motor model used that you want to placed spreaded in forearm etc, to conform with anatomically correctness, i suspect those motors will be very underpowered or either very slow if you want to implement gear down as well to increase torque/force etc..

in the early day i tried to built 2 wheels robot (3rd wheel is just dummy castor) and run straight through closed loop encoder reading read by arduino like board and sent via USB to WinXP atom machine for processing and motor command sent by the WinXP machine back to arduino and then motor driver.. the outcome is a little bit funny, i didnt have time to polish it. i also drew and plan a humanoid robot with arm and leg but not as detail as yours. i figured my actuator/motor problem is, human muscle contains 2 type or muscle, fast muscle and slow (but strong) muscle. combined we can get fast and strong limb/arm.. and i havent found a mechanical/electrical motor that has such 2 type (speed but weak + slow but strong) system and also that compact enough i can fit into arm/leg joint (elbow or knee) very close to anatomically correct size. until the motor is developed i will keep my dream on hold. thats why my interest in actuators/motors you are going to use.

hint: dont get easily agitated by the "dream killers"/naysayers around here, thread like this is guaranteed 100% will attract those kind of people, including myself sometime. i'm not saying you nor they are wrong, but my advice is treat them as reminder or things to consider, and dont talk much to prove you are right unless you have physical proof to show, keep working on it 90% and 10% to show/prove it here... since to my eye your objectives are too daunting almost impossible to achieve by myself at least. some pictures of fishing hook tucked into black fiber taped things wont cut it, i prefer some video to show it in motion.

and use "quote" button to quote's people word to reply, thats much neater imho.. ymmv.

And if I die too soon, then I'll just finish it in the afterlife if God lets me continue working on it and I don't see why He would not.

thats the spirit! i thought about it too, but rethinking about it, imho in afterlife 99% of us will regret because we dont work hard enough to find the truth about the purpose of life and the genesis of creation, because the truth is too buried deep inside many of false claims. and those who succeed (read as in heaven), will not give a rat arse about fishing hook tucked to a black goo fiber tape, they will simply ask from the ultimate creator, hey my lord, just teach me how to design and build a real biological muscle and bones and skin etc ymmv.

ps: sorry broken english i'm in hurry no time grammer check...

[Manul]

@Manul

You say "Coding: better don't" --- I disagree.  I am a highly accomplished coder and can seemingly code anything I put my mind to.  Nothing is too hard for me.  I have a natural affinity for it.  I excell at it.  I also really enjoy it quite a lot.

I watched parts of your coding videos, some old, some new. There's nothing there. It's just messing around - at best. On the other hand, your seem to have amazing drawing skills.

I will try to explain what I mean in words you should understand. You know that it's a sin to waste your talents. God gave you artistic abilities, yet you throw them away and insist on messing around with things which are clearly not your cup of tea. Have you considered that it might indeed be evil, who deceives you into doing so? I tell you this not to insult you or anything, but because it's genuinely sad for me to see when people waste talent or generally don't realize how off-track they are. Similar to how talented people waste themselves in alcohol or drugs.

[Mechatrommer]

i dont consider coder that uses ready made codes/library as accomplished/hardcore coder, real coder codes/designs their own AI engine no offense, i admit we dont have enough lifetime to do all the things.

[SteveThackery]

I still think you would be better off, using proper real-time embedded operating system(s).

Of course he would, but I don't think he knows enough about OSs to understand the differences.  Windows is a GENERAL PURPOSE OS, whereas any half-way decent bit of automation uses a REAL TIME OS.

Windows (especially W7) is - in my opinion - a fine OS, but there is little about it that is aimed at real-time system architectures. It helps if you throw enough processing power at it to overcome most of the real-time limitations, but it has poor error handling and recovery, for instance.  OK, so I'm using a car analogy, but would you be happy for it to blue-screen-and-reboot half way through overtaking another car?

@artbyrobot is correct to sing the praises of W7, but a serious developer would be looking at a custom Linux build.  Or maybe better, a custom ANDROID build.  (See what I did there?)

[wobbly]

@artbyrobot is correct to sing the praises of W7, but a serious developer would be looking at a custom Linux build.  Or maybe better, a custom ANDROID build.  (See what I did there?)

I do see what you did there.

I think Linux has a similar set of problems, it's still a general purpose OS.  True, a custom build (something like LFS) might yield a workable platform.  But why tie yourself to a pre-existing platform?  Not sure if Github works in Heaven anyway (probably does, perhaps they get the nightly builds?).

The project is so esoteric it would be best to just develop a new OS from bare metal.

[MK14]

I still think you would be better off, using proper real-time embedded operating system(s).

but I don't think he knows enough about OSs to understand the differences.

Well, if I understand their (OP) background correctly (which I probably learned from their YouTube channel).  They use to be a computer programmer, although I don't know the specific details.

You're right, they would be much better off, using a proper real-time operating system, meant for such tasks.

But it wouldn't be the first embedded system, which has been created around a Windows operating system.

I've met at least one person, who thinks that if the Windows, is suitably tamed (although this was a much older windows version, they were talking about), by removing/disabling unwanted things, and running the minimum of unnecessary services and things.  It can actually perform rather well.

Some people (most probably), seem to really like Windows, and are surprisingly (at least to me), reluctant to change to something else.  Whether that is for embedded systems, or their normal desktop computer.

Better the devil you know than the devil you don't.

Although the things I keep on hearing about Microsoft's antics, amaze me, at how so many people, just happily carry on using it.