TruState Robotics · Education

Computer Vision Lab

Our robots navigate real homes by turning camera pixels into decisions. In this lab, you'll learn to do the same: teach a 2D cleaning robot to see across six levels — starting with what a pixel even is, ending with pattern recognition. Snap blocks together if you're new, or write real JavaScript if you're ready.

🧩 Block coding for beginners⌨️ JavaScript for intermediates🔀 Randomized worlds — solutions must generalize
Booting the robot's camera…

From this game to real computer vision

Every block you snapped together exists as a battle-tested function in OpenCV, the library behind most real vision systems — including the ones we prototype for TruState robots. Here's your translation table:

1 · Pixels & Brightnesscv2.imread / cv2.cvtColor — every image is an array of numbers
2 · Thresholdingcv2.threshold, Otsu's method — the simplest segmentation
3 · Color ChannelsHSV color space + cv2.inRange — color-based detection
4 · Blur & Convolutioncv2.GaussianBlur / medianBlur — and the core operation of CNNs
5 · Blob Detectioncv2.connectedComponentsWithStats / findContours — pixels become objects
6 · Template Matchingcv2.matchTemplate — the ancestor of modern object recognition

Next steps: install Python and OpenCV (pip install opencv-python), load a photo of your own floor, and rebuild Level 2 on a real image. Then come tell us how it went — we're always looking for people who can make robots see.

← Back to TruState Robotics