Hey guys, I’m doing this mini project where I’m trying to figure out where It calculates the speed of the racket on impact with the ball (sensor is placed on the beam of the racket). I don’t have much experience using this particular sensor but is my problem basically turning acceleration into position and then doing some maths to find the speed? I’ve looked at quaternions and feel like the maths behind that is beyond me and need a bit of guidance in where to start off?
I’m using an Arduino nano.
For the short term the drift associated with low-quality accelerometers isn't a problem. If you can figure out when to re-zero your velocity value is possible to measure orientation and movement over short periods with
some degree of success. Tennis players don't stand still for long or during a volley however, so it sounds tricky.
Measuring the impact forces somehow would also give useful information, perhaps strain guages on the racquet above the handle would be useful - the change in velocity across each impact event provides information on
the momentum exchange, whether or not the velocity is accurate in an absolute sense.
Could be interesting, lots of experimentation may be needed to get good quantitative readings.
If a big budget project the first step would be to smother a racket in lots of sensors and connect lots of data
while also collecting high resolution video from multiple cameras!
Then you have a dataset to experiment with various approaches.
MarkT:
For the short term the drift associated with low-quality accelerometers isn't a problem. If you can figure out when to re-zero your velocity value is possible to measure orientation and movement over short periods with
some degree of success. Tennis players don't stand still for long or during a volley however, so it sounds tricky.
Measuring the impact forces somehow would also give useful information, perhaps strain guages on the racquet above the handle would be useful - the change in velocity across each impact event provides information on
the momentum exchange, whether or not the velocity is accurate in an absolute sense.
Could be interesting, lots of experimentation may be needed to get good quantitative readings.
If a big budget project the first step would be to smother a racket in lots of sensors and connect lots of data
while also collecting high resolution video from multiple cameras!
Then you have a dataset to experiment with various approaches.
One approach I’m thinking of to find the speed of ball on impact is to convert the raw sensor-data (2048LSB/g to actuall G-force value) but I want to convert this to actual km/h or m/sec values AND I want only to convert the highest detected force. So, it pretty much doesn't matter if the highest value is measured on the Z, Y or X axis, I only need the strongest forced to be converted
The difficult part is how to compare all 3 axis-values and use the highest one for display? It's for measuring force and speed of the sensor? Maybe afterwards I could somehow find out the racket head speed with the values I have?
I don’t really have a big budget for this and so far I’m only using the Arduino nano and the Mpu6050. I would’ve thought that would be enough for this mini project.
Stupid question, but How do the videos help btw?
Also is project too complicated with a sensor like the mpu6050?
I'm not sure the IMU itself can measure the velocity of the ball at impact.
One option worth entertaining is to measure the motion of the racquet (as you're doing), then at impact, use dynamic equations (possibly conservation of momentum) with the mass/inertia of the racquet and ball known, to 'calculate' the kinematics of the ball. Might need to build a mock-up test rig, where the racquet is swung 'consistently' and capture the motion with a couple of GoPros at the highest possible frame rate. With such info, you can then 'calibrate' your calculation method. Sounds like a neat project.
Which part of the racket? This will matter a lot as the racket is normally rotated during a swing. The two ends of it will be moving at vastly different speeds, depending on how it's being swung. The tip of the racket may move really fast one direction while the end of the handle moves slowly in the opposite direction, and the hands are basically at a standstill, just rotating the thing. So also the speed the ball sees the racket at, depends on where exactly on the racket the ball hits.
A very complex measurement. For starters you'd need multiple accelerometers on different parts of the racket to be able to follow the motion of the whole racket through space. Those sensors also have to be able to withstand the force of the ball hitting the racket, and measure this shock accurately. This hit is a very strong force that lasts a very short time, making it very difficult to accurately measure and integrate over to keep track of the speed of the racket itself.
What may work better is some form of high resolution positioning system, that can keep track of the position in space over time of different parts of the racket. The least expensive solution for this is probably a few bright markers and high speed cameras from different directions filming the action. Those images can then be used to calculate the speed/rotation/position of the racket over time.
wvmarle:
Which part of the racket? This will matter a lot as the racket is normally rotated during a swing. The two ends of it will be moving at vastly different speeds, depending on how it's being swung. The tip of the racket may move really fast one direction while the end of the handle moves slowly in the opposite direction, and the hands are basically at a standstill, just rotating the thing. So also the speed the ball sees the racket at, depends on where exactly on the racket the ball hits.
A very complex measurement. For starters you'd need multiple accelerometers on different parts of the racket to be able to follow the motion of the whole racket through space. Those sensors also have to be able to withstand the force of the ball hitting the racket, and measure this shock accurately. This hit is a very strong force that lasts a very short time, making it very difficult to accurately measure and integrate over to keep track of the speed of the racket itself.
What may work better is some form of high resolution positioning system, that can keep track of the position in space over time of different parts of the racket. The least expensive solution for this is probably a few bright markers and high speed cameras from different directions filming the action. Those images can then be used to calculate the speed/rotation/position of the racket over time.
I want to only measure the speed of the racket head. I've had a look online and I see sensors on the market that only use one sensor, so it's very doable and just complex to do I assume. The racket head speed and the ball speed is just the 2 things i'm planning on, because I wanted to add the no. of rpm on the ball after impact and some kind of sweet spot detection. I didn't realise, finding the speed on the racket and ball was this complex aha.
I'm sure the mpu-6050 can withstand the impact, as i'll be putting the sensor on the beam of the racket.
The sensors on the market may not be accurately measuring anything, just displaying a number that looks
plausible. Measurement is harder to do than just showing a number (which the average punter is unable to verify).
They might just be measuring centripetal acceleration, which is a very crude guide to speed indeed.