Newbie here: Building InMoov on a very tight budget ... recommendations to keep the cost down.

[Hernandez Remigio]

Hi I am just starting out on my InMoov build.  I am very very tight on budget and would like to build my InMoov at the least cost as possible.  I live in the Philippines and usually buy my electronics on-line at AliExpress.com (Chinese suppliers).

1.  I have seen some builders using inexpensive standard servos like MG-996R with corresponding adapters instead of the more expensive HS-805BB giant servo.  I would assume it could work otherwise why would others use it.  Going this route would save me a lot of money for the servos.  Is this a feasible alternative at least for the head movements since they seem to have the least torque requirement in comparison to shoulder/elbow movements, etc.

2.  Instead of the nervo boards that is recommended to connect the servos,  can I use off-the-shelf arduino controller shields that can be bought on-line?  I already have some of these shield in my previous arduino projects.  I know it would be easier to use the nervo boards but I am on a very limited budget and buying cheap china controllers and shields and shipping them for free using postal service to my home here in the Philippines is the least costly option for me.

3.  I can source a second hand Kinect Xbox 360 here in my country for less than US$20.  Are there other interfaces required to make it work with a computer?

I am intent on building my InMoov but I would like to build it at the least cost as possible.  

Thanks in advance your advice and recommendations.

Rio

[Perry S]

1. I have used lower price servos. They are constantly evolving since Gael originally designed the bot. Look for similar specs from HK

2. Go with the Nervo! It is a low cost board specifically designed for this bot with support. I can not imagine saving money trying to implement your own.

3. Not sure about that but sounds like you should worry about printing basic bot components before buying a kinect, if you end up even wanting one.

[Hernandez Remigio]

Hi Perry,

Thanks for the quick reply and i appreciate your comments and suggestions.

I realize that cost is relative to the individual and what is considered low cost to you may be expensive to me and it depends on the location and economics on where we all live.

As an example,  the 63 Euros or US$73 cost of the the Nervo board + components will already enable me to buy several Chinese made Arduino UNOs or Megas with corresponding sensor shields and also some PCA 9685. Of course all of these bought from chinese suppliers thru Aliexpress (free shipping).  In Aliexpress,  10 MG-996Rs cost around $33,  the PCA 9685 is $2 only.  So as you can see,  for me,  the Nervo board is expensive since I can already build a small and simple robotics project on $73 budget.

When I read that the approximate cost for a InMoov is around $1700,  this figure alone would have been enough to sound warning bells and tell me not to embark on this project as I will not be able to afford such an expense.  In my country,  $1700 is around Php90,000 which is roughly 10 months salary for the average Filipino daily wage earner (Php400/day).   Luckily my salary is a lot higher than a daily laborer but still the cost of $1700 is prohibitive for me.

Despite the prohibitive cost,  after reading through some of the workarounds from some InMoov builders to use alternative and lower cost solution such as replacing HS-805BB giant servos with standard MG-996R using 3D printed adapters,  using PCA9685 instead of Nervo boards, etc.  I am hopeful that I will be able to build my InMoov at a much lower cost than initially anticipated.  

I realize that these alternative solutions may have lower component reliability and durability compared to the more expensive recommended ones.   I am not building a mission critical robot to be used for life support systems or medical procedures. My InMoov will purely be for recreational use to entertain my family, friends and myself.  Downtime due to component failures is acceptable to me and I have the time and patience to repair or replacement them when needed.  So I can afford to use cheap servos at the expense of reliability and  durability.   I believe function wise they will be the same.   I will simply replace the damaged ones since these Chinese made components are relatively inexpensive and I can afford to buy more replacement parts.

Thanks for your support and I am looking forward to being a contributing member of the InMoov community assuming I will be successful in building mine.

More power and success to Gael and all the others that have contributed to the success of  InMoov.

Rio

[wilco van toorn]

The MG-996R will not work in the bicep, shoulder, torso and stomach. maybe the hands but thats about it. There is a reason the big ones are in the parts that do heavy lifting. To lift an arm it needs the torque and strength. You dont have to do a complete inmoov at once, You could start with the head on a stand, Buy the right servos, learn about the printing, and then after that the programming. Without a nervoboard you can use it, just pin out the arduino mega, gael has old pcb layouts here on the site, search for power distribution. But an entire inmoov, you will want to have the nervo boards. And if you can buy the components very cheap, you only need the pcb for 27 euros without shipment.

I understand its very much money, And its not only for you expensive, but for many, even in europe. 

Remember, you will fail, a lot and often when printing an inmoov. But the best advice, learn from your failures, analyze them and understand why it went wrong. 

If you first do the head, you only need 3 small servos, 1 regular and a big hs805bb, or the hk15338, way cheaper and for the most part works good. even in the bicep, but run it slower and morfe carefull. When building the head you can save up some money to save for the torso parts. When finished with the torso you can save for the arms and so on. Most build there inmoovs in 2-3 or some even 6 years. 1 bit at a time so it is expensive but it is a journey and buyiing everything at once is wrong and never the way to go. There are always changes in inmoov, so first do the head, and learn from that. it will take you 3-6 months to understand the programming and printing it all.

Good luck.

Wilco

Op vrijdag 7 september 2018 16:10:00 UTC+2 schreef Hernandez Remigio:

[Hernandez Remigio]

Hi Wilco,

Thanks for the advice.  I will take it slow and easy :-)  I have been printing and making small robotics projects for about a year so am pretty mature in printing stuff and playing with electronics.  Its the new programming tools that I am still starting to learn.  Its been almost 3 decades since I programmed a computer back in the days when mainframes and PC-DOS was still in their prime.  Low level programming on Motorola and Intel chips was my thing then.  Now I need to learn new languages like C++ and learn to use MRL.   Anyway I will take it slow and learn from you guys.  Thanks again and more power to InMoov and the MRL guys.

[wilco van toorn]

MyRobotLab is python scripting. And printing an inmoov is like starting printing all over again. It is a challange. Good Luck Sir.

Op zaterdag 8 september 2018 16:09:05 UTC+2 schreef Hernandez Remigio:

[CheesyRex !!!]

Hi Perry, 

I was interested to know how you went with the PCA9685 boards, are they worth looking at or do they require too much time to reconfigure the existing programming to work with them.
I'm doing my research before I start, so i can plan out my movements a bit better and I'm wondering if the PCA9685 might be the cost saving factor I need.

Thank you for your time.
Brian

[Mark Seaton]

I too attempted to build without the nervo boards to start with. It can be done, it is just very time consuming and fiddly.

You basically just need to run all the servo wires to the arduino and also sully power, lots of wires, lots of soldering, lots of confusion.

I ended up getting the nervo board as I was getting frustrated with all the fiddly wiring.

In Australia, we have another problem, The Au dollar is not so good, so the nervo board ends up A$110 and you need 2.

Also the large servos are expensive, the cheapest ones I could find were A$55 each. I have purchased about 6 so far, but have just found a slightly cheaper, smaller one for $45 which I am testing, it may work for some parts but not others. 

I have been building this for over a year and have most of the 3d printing done and assembled, I have the head done and 1 hand and arm with 2 spare servos to start on the next arm, so as you can see, i am also waiting until I have the money to purchase the parts.

I have a kinect controller, but have not hooked it up as I don't know the answer to your question either.

[CheesyRex !!!]

Thank you for the advice Mark, also an Aussie here so exactly why I am trying to find the most cost effective option.
I've spent 2 weeks researching servo motors and believe I've found most of an alternate range that the community have given the go ahead for.
I've purchased a couple of the Servo driver boards to have a look at, I'm looking into doing this as my major project for my Degree. But with study comes a lack of funds.

I'm looking at substituting:

MG996R servos in for a number of locations,

JX PDI-6225MG-300 servos for some of the stronger motors.
If I need to I can redesign parts to hold the new motors.

I'm going to be starting with the arm and head first but will hopefully have the majority of the parts printed and assembled (even if not wired up) by November to make him look pretty to show off.

Just have to finish off my 3d printer enclosures and I want to put together an acetone fume smoothing enclosure.

thanks again

Brian

[Perry S]

Hi Brian and Mark,

Somebody asked about the PCA9685 boards but I do not use them. It may have been somebody else above that suggested them.

As a suggestion you can run the bot with one Nervo and Mega with a lot of functionality still. At least you could back when I used those components. I ran mine for almost a year I think before I got my second set. The only thing that the second set really ran back then was the fingers. I had such trouble with the fingers working that I disabled them most of that time until I designed something that worked for me. Not saying better, just different.

Cheers Aussies!

[Ray Edgley]

Hello Brian,

I have used the PCA9685 in the head, one in each arm and one on the back.

It has the advantage of not needing as many wires running from the back to each arm or up the neck into the head.

MRL does support the PCA9685 with a library that works very well.

However the Inmoov scripts do not support the PCA9685

Inmoov2 is currently under development, but don't expect to see it released with Nixie due out soon.

Inmoov2 is rumored to be able to support a setup using the PCA9685.

If you do go down the path of the PCA9685, you will be creating most of what the Inmoov scripts do from scratch.

If you love to program, then go for it, but if writing lots of code sounds a little to daunting, then stick with the Arduinos and nervo boards.

Ray