Im an archer as well an Arduino enthusiast , thus I came up with a challenge and I would like everyone input:
problem: I struggle to keep my draw elbow (the arm that holds the bow string, in my case my right arm) to the proper height (height distance from my side) and extension (properly aligned to my shoulders).
idea: to have a sensor installed to a elbow protector (cloth material) and that sensor would help me to keep the proper position.
I was thinking using an accelerometer, however what I have read may not be as precise, any inputs?
I also, was thinking using the new Arduino boards running TensorFlow and TinyML , but Im quite not sure what sensor to use.
Another way, would be using 3 cameras and ML - one capturing my arms, from the right hand side, another from the back and another from the top. In this way, I would be able to understand the 3 axis of my elbow position (height, alignment to my shoulders and extension).
Anyways, it sounds lots of fun to merge two hobbies and Im humble enough to ask for comments and ideias from this wonderful forum.
Thanks you all in advance and please collaborate with ideas.
An accellerometer can't report absolute position. If you have the money for cameras and AI, that will work. With a huge and complex development effort... so there is no easy way. If there was, you would see hundreds of similar consumer devices doing similar but different jobs.
I know that's not a comforting answer, but if you want to get some satisfaction from this hobby, it's best to aim for achievable goals.
Thanks aarg for your comments. I do believe we can collectively find a solution for this problem. Im quite positive there are solutions available. As for consumer products, I found the following that seems very interesting:
If you can use this as basis for our idea, should be possible.
It would be easy to duplicate the Mantis X8 sensor, (it is almost certainly just a 6DOF IMU and a Bluetooth transmitter, like the Arduino Nano 33 BLE Sense).
Analyzing the data and making sense of those data may require sophisticated programming. The product video gave me little clue what that analysis is actually doing, but tilt angles are trivial to calculate and display.
Sure, but you didn't feed back at all about the issue I raised. The product you linked to, apparently does detect movement patterns, which an accelerometer can easily manage. You said, "to the proper height" which is a position measurement. As I mentioned, an accelerometer is very crude at estimating position. So I would like you to comment. Do you want to relax or modify your requirements, or add additional devices (such as those that I alluded to)?
The cameras would work but you didn't feed back about that either... what kinds of solutions are coming up from the research that you must be doing?
Dear sir, using a mirror can help, however if I twist my head to look at my shoulder, I have to break my form, thus isn't the best tool. Nevertheless, thanks for the input
Delta_G - right on the money, that's the proper way, also let's not forget the top view, in which I must keep the arms (extended arm), shoulders and elbow aligned. It's really a 3 dimensional issue. Thanks for the comment
aargh - sorry I was on the run and my reply wasn't proper:
the product I sent the link is one tool designed for aiming, and you correct doesn't solve the elbow problem, nevertheless the solution does help once my "form" is proper.
As for the 'form" issue in hands, since accelerometers don't help us regarding shoulder position, I was thinking back on the three cameras. I saw a project used for industrial safety purposes, in which "lines" are draw in a certain area, for example if the operator cross that line, it triggers an alarm, perhaps we could use the same principle here and draw lines where the elbow should be positioned, in the three cameras and perhaps some kind of feed back to me (the archer) so I can focus in positioning my shoulder at the proper height - camera from the right side as well camera from the back and from the top two lines to keep the alignment . maybe we could record few shots and use the images to have a proper feedback.
After some research, I have to agree with you that 3D positioning is not available at this moment.
Thanks for the comments, it really opened other frontiers of thinking
Thanks for the reply. I think you can get somewhere with a combined sensor approach, and certainly if you have multiple sensors that have both accellerometers and IMU (which are available), you can put together a close enough approximation of what you are after from the angles as pointed out in reply #7, that maybe you can omit the 3D cameras (that is where high costs would come in).
A position calculated with inverse kinematics wouldn't be extremely accurate, but maybe you don't need a greater degree of accuracy.