profiting of the winter months and learning about AI, i want to release for the first time - here in exclusivity - one of the projects i've been working on. 

Using a 15.5$ device called M5 Stack AtomS3, 0.8x0.8x0.5 inches in size, 21x21x13mm, that has a sophisticated IMU (Inertial Measurement Unit, three acceleration measurements and 3 gyroscope measurements , 200 times a second) inside, i developed a sensor for our boats that measures the waves Period, their Height and Significant Height (1/3 of the largest waves), records the attitude variations of the boat, records the jerks (e.g. sudden accelerations typical of slamming), and also records over time the possible additional shocks from neighbouring boats while mooring (in conjunction with my other project...stay tuned!). Accidentally, it also provides an approximate sea state according to the Beaufort scale...

The code is opensource and available here, 

https://github.com/pzucchel/boat-and-wave-AHRS" rel="nofollow - https://github.com/pzucchel/boat-and-wave-AHRS

including detailed documentation on the way to integrate it on your Wifi, allowing any other systems (e.g. my other project...) to collect and store the data 24/24 7/7 365/365 via NMEA0183. 

the code has been extensively tested by simulation, being the mathematics quite complex as you can see, and I've also explored other techniques to abandon them after some analysis  (read: Fast Fourier Transform), that may be resurrected by real data .

I will hopefully improve it further with real-world experience (i want to have vocal alarms triggering limit situations, in particular slamming: i still remember my bow thruster batteries collapsing in the middle of the atlantic...) 

I can provide further info on my boat installation, and maybe a simpler "ready to go" firmware for those not familiar with the arduino IDE...

admittedly , the data are broadcasted: so you need some other system (for example , signalk) to collect the data .

Any limitations? for sure! i am not detecting (yet) the waves direction, so i can't compensate for the doppler effect introduced by the boat motion: so the wave period is the APPARENT wave period and not the one indicated by the meteo services. But the boat slams according to the APPARENT wave period, so it's physically significant...

hi Leon!

it is indeed one of the locations I am going to test. it depends a bit on the purpose...slamming is probably best detected at the bow , waves at the center of gravity. Given the limited cost, I could also put two : the phase of movement may give additional information , in particular the wave direction. but this can be done in a system ...that I am actually building already ! fusion of sensor information with significant computing power can provide something very new in a boat .

going to the other exotic ideas , I am thinking to implement one to be integrated in the tricolor light at the top of the mast ! 

what can be difficult to realise is that such a small device has also significant computing power .I am actually running a madgwick filter (and more ) that integrates 3 acceleration measurements with 3 gyroscope measurements, and you can actually get the precise trajectory of where the device is, 50 times a second (I experimented up to 200Hz) ....so in a sense some locations provide better data ,but if the boat is rigid you can know the movement of any other part! maybe even measure the rigidity of the hull ?

to get some idea of where the technology is today  , see this :

http://www.cs.ndsu.nodak.edu/~siludwig/Publish/papers/SPIE20181.pdf" rel="nofollow - http://www.cs.ndsu.nodak.edu/~siludwig/Publish/papers/SPIE20181.pdf

i already have a running plugin for signalk collecting the data, and integrating them into the signalk format . the signalk data are stored in influxdb for permanent storage , for analysis in grafana. while the alarm and control part is done in home assistant ...

Have you thought about collaborating with someone like B&G to put this on a display?  Just as wind direction and speed are graphed over time, you could do the same with wave period and other information.    Might be "one more differentiator" for one of the biggies to use.  Suggest you consider a patent if you might.  Otherwise the opensource is going to feed someone else cheap..  :-)