I designed and printed a new flexiforce adapter. With this one, the sensor is fixed at a perfect position even if you move the cable. Also it is between two metal parts. Here you can see it.
I also had built a new box to get everything in a fixed place and see how it would fit. Here is the front of the box.
And here the back of the box with all electronics. You just need to plug in one USB cable and one Ethernet-cable for the raspberry-pi and power it up.
With this setup is absolutely easy to install the wire-cables for one robotic finger through these bowden cables. Plug in the cables and connect to rviz. ready to go
Hello all,
i started a kickstarter project to build a ready-to-production prototype of the Humanoid Robotic Hand. I hope you can support me and spread the message, or pledge something on my kickstarter site.
Thanks alot
Woohoo, I got the honor of being the first backer 
Your campaign seems ambitious but I wish you all the best! Keep up the good work.
thank you @dbolkensteyn for your support. Have a nice day
Looks like i fell asleep for a short while, but so does the robotics industry. I thought i could use rotation and linear motion to simply move a robotic finger and i tried and it works nice. For linear motion i use linear drawer slides. Next would be to build something to hold servos to move it.
Here are some videos.
https://www.youtube.com/watch?v=CF0qYk1lGtM
https://www.youtube.com/watch?v=gGd50u3dBYI
hi,
someone knows a cheap current control servo ? Not the servos that start at 250 euro or more.
Or someone knows of “something” that could be used with a normal servo but makes a current control servo out of it ? some circuit board i connect infront of the servo or something i dont know of ?
thanks in advance
simpler is better. Setup for one finger.
Someone knows cheap vr-gloves or other gloves to get haptic or force/touch or joint-degree values ? thanks
This bowden outer shell is even smaller. Inner diameter is 0.6mm , outer diameter is 1.2mm. My 0.5mm steel wire fits perfect. And the connection between finger and “rest” is not bigger anymore (in diameter) then the finger itself.
Next thing is still getting a way that the finger could “sense”, “feel” by itself with force-sensors or anything to train a ML model. Perhaps this looks good for simulation Google AI Blog: Speeding Up Reinforcement Learning with a New Physics Simulation Engine .
new hardware control setup, as always simpler is better. I use a aluminium T-shape to mount the n20 geared motors and the flexiforce sensors together. The biggest thing now is the rpi4 with its hats and the power supplies.
Next steps are building a simulation with it and then run the model on real hw and realtime ROS 2.
