I used the development board to sample data in a clean environment and in a jar with some cheese inside. After that i successfully created the ai algorithm and I managed to successfully load the configuration into the bsec_config_state.ino. The board with 8 sensors and the flashed code is then able to give me the percentages right.
After that i loaded the configuration into my custom code for the Raspberry Pi Pico using the bsec188.8.131.52 library, but when asking for the gas estimates i always obtain gas_estimate_1 at 100%, no matter in which environment i put the sensor.
I followed the workflow of the integration guide:
The sample rate is set to scan, is there anyone that already faced this kind of issue even if on other devices?
Could we know what are BME688 dev kit software version and BME AI-Studio software version you used?
Hi BST Robin,
I am using the BME688 Dev Kit Software version 2.0.6 and the BME AI Studio version 2.0.0.
You should use BSEC184.108.40.206 as the release note mentioned it in BEM688 dev kit release package.
To my understandings by switching to the bsec220.127.116.11 in my implementation with a single sensor i should be able to get more accurate readings right? Because the devkit implementation of the sensing, based on the .ino example, works fine already.
In the image below there is the output generated by the adafruit sensor connected to a raspberry pi pico: During the first "parallel mode" it was exposed to the environment while during the second one it was inside a jar with some cheese (the same one used for the sampling).
As you can see the probability doesn't even flinch, and that's what makes me worry. As you can see in the image the version this time around is 18.104.22.168 (obtained through bsec_get_version, no cheating).
I am attaching the .h and .c files for the configuration too, if someone wants to try.