Vision-based action planning for image-based condition correction

This project focused on developing and simulating a vision-based system for controlling automated actions, with the goal of gradually adapting a visual scene to a desired target state.

By analyzing the difference between the current image and the reference image, the system calculated a series of corrective robot actions to minimize visual deviations.

The entire solution was developed using a simulation-first approach, enabling independent further development without physical hardware.

Image processing served both for error quantification and action planning, thus facilitating rapid development and flexible integration in a hardware-independent environment.