Current hand motion sensors suck, we are going to make it better and free.
Now that we are gathering data to train the machine learning side of things we found that the current somatosensory glove by hiwonder is lacking in many areas. We are now designing a more robust hand movement glove to better train the OpenMuscle band. Also we are upgrading the OpenMuscle PCBs to the next version and increasing the sensor count by a factor of two.
The data is in and the detection mechanism works even with ill-fitting around the forearm. We will be presenting the work so far at the local maker space!
The new sensor cells will have 4 radially pointed sensors and 1 lateral placement to better create a deterministic system for the network suite. We are researching the right training models for the machine learning and developing the GUI daily.
We have also ordered better clevis pins and springs to fine-tune the current bracelets.
There is a lot of information being published on the github and on this website so please be patient while we get all of the tutorials and materials put together. If we aren’t fast enough for you, consider helping the project.
Above you will find the super rough draft of the hand sensor we are building. It consists of 6 hall effect sensors and two PCBs. Each finger will have a ring that a magnet is placed in and the pcb will rest on the back of the hand.
Since hall effect sensors detect electromagnetic fields there will be no “moving parts” in our sensor just the hand itself that is moving.
The first prototype will only aim at sensing the first 10-20% of finger movement but will corelate more accurately for our purposes than those that we have tried previously.
So very soon we will be releasing Open Hand – the open source wireless hand somatosensory apparatus!
Check back for more news!