You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: docs/Micromouse/AmbientSensorRecording.md
+1-1Lines changed: 1 addition & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -7,7 +7,7 @@ File writes are VERY slow, string manipulation is VERY slow due to how strings a
7
7
8
8
Anyway, after a LOT of different experiments trying to work out how much data and how fast I could get data out of the system the end result is that I simply have to store data in some sort of ring buffer and the dump it once I have time. Which is code for "When the mouse is stopped".
9
9
10
-
I know the topic comes up a fair bit on the monthly calls. I have to say I have been having quite a bit of success in getting the AI to help carry out what I call "Donkey work". Some of its technical approaches are questionable at best and just down right wrong at worse but for the testing I have been doing it has been very quick to wrap up code in timing handlers so that call can be timed and output to the screen. A surprise use has been if you give the AI the method declaration on how data is going to be written and then give it the output it can process that data very well. Then it is able to compare one run to another as it has all the data in its memory and can provide analysis on what changes improve or worsen a system. On a similar note you can give it a whole log file and then the example bit that you want to process and it will write a function to pull out that data specifically from the log. If you go a bit more mad you can get that thrown into a python panda data frame for analysis.
10
+
I know the topic comes up a fair bit on the monthly calls. I have to say I have been having quite a bit of success in getting the AI to help carry out what I call "Donkey work". Some of its technical approaches are questionable at best and just down right wrong at worse but for the testing I have been doing it has been very quick to wrap up code in timing handlers so that call can be timed and output to the screen. A surprise use has been if you give the AI the method declaration on how data is going to be written and then give it the output it can process that data very well. Then it is able to compare one run to another as it has all the data in its memory and can provide analysis on what changes improve or worsen a system. On a similar note you can give it a whole log file and then the example bit that you want to process and it will write a function to pull out that data specifically from the log. If you go a bit more mad you can get that thrown into a python pandas data frame for analysis.
11
11
12
12
Back to the goal of getting sensor data. I have 4 sensors. So I want to capture the ambient reading and what I call my RAW reading which is the value once I turn on the LED - Ambient. My value range is 0 - 32000 from my ADC (we ignore that I believe this is technically upscaled by the OS from 12 to 16 bits). Regardless this gives 8 data points per system tick to capture.
13
13
Short version - impossible to capture all for any decent length of time.
0 commit comments