Autonomous intelligent agent simulation

End-to-end Python pipeline: industrial part recognition, STRIPS planning, and A* maze navigation.

Python
Computer vision
KNN
K-Means
STRIPS
A*

Industrial fasteners sorted in boxes for vision classification

Stage 1 — Computer vision

Captured and segmented bolts, screws, nuts, and washers. Extracted morphological features and classified parts with KNN and K-Means after HSV segmentation and contour analysis.

Original images of industrial fastener classes

Segmented and cropped parts ready for classification

3D scatter plot of selected feature space by class — Feature-space separation before classifier training.

Stage 2 — STRIPS planning

Modeled part reorganization with STRIPS/PDDL actions to move objects between boxes until a target layout is reached. The planner generates valid action sequences from random initial configurations.

Stage 3 — A* navigation

Simulated transport of the agent through randomly generated 2D mazes using A* pathfinding, completing the perception-planning-navigation chain.

Solved maze with A* path highlighted — Successful path planning in a dynamic maze environment.