We are Dimitris and Joan, students of the master’s degree in Automatic Control and Robotics at Universitat Politecnica de Catalunya (UPC), Barcelona. We chose as our final project to build a poppy-robot but with some modifications.
First of all we decided to replace the -expensive- dynamixel motors with slightly cheaper motors. Our approach is similar to the “supermodified” solution( link). The idea is to replace the inner parts of the motor and integrate a microcontroller (arduino) the driver of the motor and the magnetic encoder, inside the motor case. We also have to fit the new motors to the 3D-printed parts of Poppy.
The second part of the project is to build a platform with only the legs of the Poppy in order to study autonomous biped locomotion.
We created this post to introduce ourselves. We are going to create separate posts in the corresponding categories in the forum. We would like to have any kind of feedback (especially the negative ) and of course, any advice or ideas are very welcomed.
Hello @poppy_esaii, and welcome to Dimitris and Joan into the poppy forum!
In the poppy team we start to work into the developpement of open-sources and low cost motors for robots. Supermodified is a really great project, we are really interested about your background!
Can you give us some precision about your plans for motors? You just want to buy some RC servos and hack it or you want to do more?
I think @Thot have the same kind of project with biped locomotion…
The idea is to buy RC servos and “hack” it, even though the only part of the servo we use is the … dc motor. Originally we had the idea to buy just dc motors and built the case and the electronics but due to restricted time (and courses) we decided to just modify the usual RC servos.
If not tomorrow, the day after tomorrow I will open a post in the “electronics” category (I hope it is the correct one) with more details. Just for curiosity, we ended up to but two type of servos both from Hi-tec. The replacement of MX-24 is the HS-7954SH ( ~90 euros) and the replacement of MX-64 is the HS-7950TH (~100 euros). I know they are not that cheap but the options in RC servos for high torque motors are not so many (remember we want to study the biped locomotion as well). For the encoder we are using the AS5145 (12bit) and driver the VNH5180A-E, from ST. The implementation is done and right now I am working on the prototype of the pcb’s (double layer with toner transfer ain’t that easy ). As soon as we will have a complete first version we will share it.
What about your work? Are you trying to built a servo motor from scratch? That would be really interesting.
Welcome ! Great project we will for sure follow it. Thanks for planning to share it actively on the forum.
As @Nicolas said, we are working on new open source actuator (from scratch but very similar to cheap RC + supermodified) so we are very interested by any feedback you can provide on this solution.
BTW, if you fork the project, please use the last release available here:
Welcome on the forum ! We have very close projects indeed as I have built only legs of Poppy to study dynamic walking in all the directions. I had the opportunity to buy Dynamixel servos directly so I do not deal with the issues about new actuators but all the work done here by you and @Nicolas interests me a lot to build cheaper robots for show.
My main question here is what kind of servomotor is needed to walk (and which kind of walk) ? Huuuge question.
Hi @Thot. I am not sure I understand what you mean with “what kind of servomotors”. I do understand your huge question though!.
The control of the walking will be energy based and not zmp based (I think your approach is also not zmp). As for details, we are still researching how exactly we will try to implement it, as we keep finding different approaches. Right now, we are focusing to finish the modification on the motors and afterwards we will “dive” into control (which I am looking forward to it, personally ).
How are you doing with your project. Any progress?
“what kind of servomotor?” I mean, which specifications (torque, maximal speed, reversibility…)
Cool if you are doing energy based walking. I am on the same way (At this moment, I mix a bit a lot of approach and see what happens on the real robot).
My nightmare for the moment is "SENSORS " . I should have spent more time on sensors at the beginning of the project. For instance, if you want an estimation of kinetic energy, you must have a good gyroscopic sensor and a good estimation of the height of the center of mass…
) and of course, any advice or ideas are very welcomed.
). As soon as we will have a complete first version we will share it.
" . I should have spent more time on sensors at the beginning of the project. For instance, if you want an estimation of kinetic energy, you must have a good gyroscopic sensor and a good estimation of the height of the center of mass…