I helped some LSU students build a 4 foot long boat with Pixhawk in rover mode.
After of hours of tuning and tens of hours testing…
Here is our latest test run.
2 Likes
Great grid going there! any pictures of the boat? You’ve always had good hulls.
This was a student project… I don’t claim this hull at all… LOL
They built it at my house and I knew what ever they came up with would work at slow speeds…
working speed is 1.3 mph
In this pic, a student and I took a ride on the S-Boat to rescue the DAB (Data Acquisition Boat)
after steering servo failed.
Wow, that’s not small at all. Let the students know they’re doing a good job, I like the innovation!
Looks great. Delighted the tuning went so well - very nice work.
Thanks, Grant.
Great Job!! can you post a little bit about the tuning Process? What were the problems? What changes were necessary?
Thank you so much!
I will get the student with the task of Navigation to post what he did… I was not always there…
But the 18" cat that I tuned was hour after hour of riding around on the boat trying things…
Reading what each param did and changing values and see what happens…
I can say by turning off all EKF things got far better with speed and navigation. The EKF is just not required
for slow boats, from what I found…
Eddie
The Student wrote this, it seems very good to me…
Over the course of 8 hours, the boat was left in auto mode so that it would follow the same path, but the steering parameters were continuously modified according to the following rules:
- If the boat was not turning tightly enough, the proportional gain was increased by 0.1.
- If the boat was exhibiting a large amount of oscillation (2-3 oscillations before converging to its target heading line), the proportional gain was decreased by 0.1.
- If the boat was exhibiting a small amount of oscillation (1-2 oscillations before converging to its target heading line), the derivative gain was increased by 0.1.
- If the boat was exhibiting any high frequency oscillation – that is, oscillation in the steering servo itself that could not be seen in the course of the boat, but could be seen in the PWM signal sent to the steering servo (and also heard in the movement of the steering servo) – the derivative gain was decreased by 0.1.
- If the boat was exhibiting error lasting longer than its lower-frequency oscillations, its integral gain was increased (this never happened, as the default value for integral gain was adequate) by 0.1.
- If the boat was exhibiting lower-frequency oscillations that were not reduced by a reduction in the proportional gain, the integral gain was decreased by 0.1.
- The period was changed to match the temporal length of the oscillations in the path of the boat.
- If the boat began to turn too early after reaching a waypoint, the waypoint radius was decreased by 0.1m.
- If the boat begin to turn too late after reaching a waypoint, the waypoint radius was increased by 0.1m.
- If the boat turned too sharply after reaching a waypoint, the turning radius was increased by 0.1m.
- If the boat turned too widely after reaching a waypoint, the turning radius was increased by 0.1m.
- After any change, the course of the boat was observed for at least 30 seconds, and for at least 2 turns.
The final PID parameters are shown below:
Proportional gain 21
Integral gain 0.2
Derivative gain 0.8
Period 30 seconds
Waypoint radius 3.5 meters
Turning radius 3 meters
3 Likes
Thank you so much!
Whenever our boat is ready here, these will be good informations!
Cheers!
can you put on your .param here ? i wounder how to turning off all EKF things ??
Hi @admired,
I don’t think a param file from 2016 is going to help. Maybe open a new thread and detail what issues you’re facing?
why ,turing off EKF will worked ?