Ti-Tek’s, Echo v1.0 Prototype is a desktop scale, mechatronic platform in the form of a humanoid torso. The prototype is a wirelessly controlled, fully self-contained unit that interacts with physical environment via two “free to control” arms.
For more details on the project and to provide comments on our current and ongoing projects visit https://ti-tek.com.au
What I ask a lot around here, and probably a lot of people would like to know, is how much did this cost you? Or, if somebody wants to build exactly the same robot, how much approximately complete hardware would cost? Except that, you probably spent a lot on man hours, software licences, etc.
Nice project!
Thanks for sharing with us!
Since it is a torso prototype, does that mean the next step is adding this on a humanoid robot?
Looking forward to reading your developments on this!
See ya!
It was built in my garage, so I am very glad to hear it looks professional!
It cost roughly $600(Australian) plus a ps4 controller which I already had. From a quick look around at the parts at RobotShop I think it could easily cost under $400US.
You would obviously need a 3D printer, but i doesn’t need to be a high end printer. You don’t need a large print bed it would just need to be at least 60mm deep. I haven’t tried printing the parts in anything other then ABS plastic, some of the parts snap into place but I think PLA would be flexible enough if you prefer to build with that.
As for the software, yeah the design suite’s for various 3D modeling software can get pretty pricey, but most offer free trails and AutoDesk even offers free versions of software for people who are enrolled in certain educational courses (these have restrictions).
The actual build time wasn’t substantial, (design was a different story). For all the parts, printing and cleaning would take around one day. Assembly and Electrical (depending on skill of the builder), lets say it could be a great project over one weekend.
It was a great project is super fun to operate! I am so happy it all ended up working out. If people are interested, I am looking at producing proper instructional videos on the build and getting documents together. Trying to recover some of the costs through donations at our site (ti-tek.com.au). At this stage if you donate to the project we will give you access to all the code and print files for this robot. This is in the interest of funding the next projects which are aimed at upgrades for the robot. Once the cost has been recovered we will just release all source and files for free to everyone.
Yes that’s absolutely right! The following projects that I am now planning will be upgrades and add-on’s to the existing torso. Looking to make the prototype modular, so that we can easily upgrade the arms into any different tool and add additional add-on’s. After that the projects will be; a head to feed vision to a phone that is also linked to the gyros so it can pan and tilt maybe even audio, and a mobility add-on. This could be wheeled or tracked, but I am very keen to have a go at bipedal legs with balancing. It’s all very exciting!
At this moment each servo is controlled individually by the controller. Which, as I see you have clued onto, is extremely difficult to precisely manipulate. Looking to develop the program to easily control the displacement of the end point of each arm, however, I suspect that the math for that solution would quickly exceed the capabilities of the 2560 microcontroller. Could possible change out the math with more of a list instead but ultimately we want to upgrade to a CPU to enable future developments. How do you think we should go about doing this?
Certainly the microprocessor would not support the calculations, especially because it involves investment of matrices in kinematics and possibly recursive methods if you are looking for a path planner.
The first would be to model the arm and get direct kinetics (Knowing the configuration of the joints, finding the point in the space where the manipulator is) and inverse (knowing the point in the space where the manipulator is, finding the combination of positions in the joints that will reach that point).
To do this you can check the following repositories that already have indications to calculate them:
Do not forget that there are restrictions on the movements that must be considered in the code. Placing a maximum and minimum in the movements of the joints will help that the arm does not want to make movements or reach unwanted positions.
To know the position of the manipulator you will need either a camera to perform position estimation or a combination of sensors to estimate position (but it would be complicated given their sizes).
I hope this gives you an idea and we will see progress of that project!
Wow, amazing project! I’ve always wanted to build a humanoid robot myself. Are you planning on automating the movement of the arms to serve a specific purpose? Like, being a part of a production line or even making food?
I’m looking forward to v2.0
What is stopping you from building one? The project was super fun. Trying out different things and always learning! It is only manually controlled right now and I have just been using it to hold wires and things while I’m soldering ect. I am working on a few upgrades to help automate it. Hope to add them to the project as soon as possible!
Hey @wMquade. I have just been busy recently but I will hopefully start building one in the near future. I’m sure Echo is a good personal assistant Can’t wait for v2.0.
@wMquade This is pretty nice! Can you give any indication of the grip strength, load capacity, and repeatability/precision? As I understand it, these are typically the main points of differentiation between these arms built on self-contained servos, and larger steppers.
Also, are the model files for the 3D printed parts available on thingiverse or something like that? The provided link seems to lead only to a paywalled member site.
Can’t wait for v2.0.