Overview

This article describes the process to build a Stellar Studio SDK-based AI validation firmware including the deployed model and the aiValidation stack.
The AI validation firmware is used by the stedgeai application ('validate --target stellar-e (or stellar-pg) -mode target' command) to perform the on-target validation.
After that the build process has been completed, using one of the compilers supported in the Stellar Studio Stellar-E or Stellar-PG SDK, the next step is to flash and run the AI validation firmware on the Stellar-E or Stellar-P or Stellar-G target.
The flashing of the Stellar-E or Stellar-PG AI validation firmware can be done using one of avaiable debugger/programmers available for the Stellar devices, like OpenOCD SW and StellarLink USB dongle (if the HW is not already present in the Stellar evaluation board).
For OpenOCD, it’s automatically added when installing Stellar Studio: under the openocd/docs folder you can find a guide on “How to use OpenOCD and StellarLINK in StellarStudio”.
You can also find under openocd/scripts some utilities and documents on how erase and flash Stellar programs.
For StellaLink USB dongle, you can find sw drivers under the openocd/drivers folder but you also refer to “StellarLINK in-circuit debugger/programmer for Stellar and SPC5 automotive microcontrollers” and “StellarLINK device drivers”.

Please refer to “User manual - Getting started with Stellar Studio AI plugin for Artificial Intelligence (AI)”, “AI commands” section to have a complete overview of the Stellar Studio AI plugin (based on the ST Edge AI Core) and of the full process to validate a model on the target.