This lab works on phones, but writing code is much easier on a laptop.
Perception · Object Detection and Semantic Understanding · P6-L3 · Intermediate
A warehouse robot stops for every box the detector reports above 0.8. At dusk it drives straight past crates it never reported, and in the morning it stops for shadows.
You will see what the threshold trades, choose it from the cost of each error, and find where the misses concentrate.
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 explores the threshold. Stage 2 chooses it by minimising a cost. Stage 3 breaks the misses down by object size. The predictions are produced by a simulated detector with a stated error model, not by a trained neural network. Your code is graded on hidden scenes.
Educational simplification and honesty note: predictions come from a simulator that turns known ground truth into detector-like output with an explicit error model (detection probability falling with small area, occlusion and dusk; box jitter; wrong classes; duplicates; false alarms), then through non-maximum suppression; no neural network is run. Costs are given numbers, not a safety analysis. All data in this lab are simulated.
Get notified when new robotics labs are released. One email per new lab, nothing else.