This lab works on phones, but writing code is much easier on a laptop.
Perception · 3D Perception and Point Clouds · P5-L2 · Intermediate
A robot drives up a ramp. Its obstacle detector worked on the flat factory floor, but on the ramp it reports a wall of obstacles in front of it: the road itself.
You will see why a height threshold cannot handle a slope, estimate the ground plane from the data with RANSAC, and clean up the ground removal.
You have passed every stage. Now connect it back to the theory and move on:
Plain Python with math (and slamkit for small matrix helpers: matmul, transpose, solve, inverse, identity). Images, point clouds and detections arrive as lists inside the dictionary D; an image is a list of rows, each row a list of pixels. There is no NumPy or OpenCV: you write the algorithm yourself, which is the point of the lab. Print with print().
Stage 1 is an experiment on the flat-ground assumption. Stage 2 writes a RANSAC plane fit. Stage 3 refits the plane by PCA and removes the ground with a tolerance that grows with range. Your code is graded on hidden scenes with different slopes, rolls and obstacles; the simulator keeps the true ground labels and the true plane private. It goes beyond the Sensors lab, which removes ground with a height threshold on flat floors.
Educational simplification: a simulated 20-ring LiDAR 1.2 m above a planar ground (slopes from −3° to 6°, rolls up to 1.5°) with three or four box obstacles, a 90° field of view and range noise of 2 cm plus 0.2 cm per metre. Real ground is not a perfect plane, and real sensors have 16 to 128 rings, motion distortion and dropouts. All data in this lab are simulated.
Get notified when new robotics labs are released. One email per new lab, nothing else.