This pool test focused on 3 core tasks: testing a new yaw to cv object function, finalizing sonar to be merged, and gathering more acoustics data.
New Yaw To CV Object:
Roughly half the pool test was spent debugging the logic of this new task. This function is an experiment to learn and prepare future tasks that will use more async logic. Most of the bugs we found were related to inconsistencies between the step and send features of task planning. For future reference, the send function not only allows you to send data to a task, but also steps through the task with the new data. We found that the new yaw2cv object task works significantly faster, but we experienced a pretty large bug we were un-able to solve. The move to local sub-task appeared to be moving objects in the global reference frame instead of the local one. After the pool test, we believe we found a bug that prevented dynamically sent poses from being transformed into the local reference frame.
Sonar:
We also confirmed our fixes for Sonar from last pool test solved the issues. We found that Sonar accurately turned to face the wall with no geometry issues. The only remaining known bug is that if sonar doesn’t see any wall object, it freezes up the task planner rather than relinquishing control.
Acoustics:
Last pool test, we took acoustics data without the thrusters on while someone holding the robot stationary. This pool test, all our data is taken when the thrusters are turned on and the robot submerged at a constant depth. We’re still working on integrating logic 2 into the acoustics code, but we were able to take some data to compare what logic 1 data looks like with our new logic 2 data.
First, we analyzed the height of pings at varying distance away from the robot. In most tests, there was one hydrophone whose ping signal was different compared to the other three, sometimes higher sometimes lower. When we placed the hydrophone directly under the robot, hydrophone peaks were around 6.5 V. At 2 ft away they were 6-6.1V, at 10ft away 6.2- 6.5 V, at 20ft away 4.9 V -5.2 V, at 30ft 6-6.2 V and at 40 ft 4.9-5.2 V. We’re unsure sure why the decrease isn’t linear, and need to investigate why the ping signal drops off in magnitude at 20 ft. We also need to investigate why one hydrophones’ peak amplitude is offset the other ones.
We also got some data on determining which side of the robot the pinger was on while thrusters were creating noise. This data will allow us to check how accurate our software is in determining which hydrophone got the pinger first.
Goals for Next Time:
I’m really excited for the next few pool tests because I think they’ll be very productive. We’re getting close to merging many parallel projects. At the next pool test we’ll hopefully finish up our new yaw2cv object and prequalify for competition.