From cleaning blocks to mapping them in 3D

In Issue #4, the first cleanups had just run and the scan system had grown teeth — full-screen AR camera, coverage fog, guided routes. A month on, we crossed the pilot’s first real milestone, and the scan archive started turning into something we didn’t have a word for yet. Now we do: a digital twin.

Before-and-after photos were the proof of concept. Before-and-after 3D is the proof.

20 of 100 blocks cleaned. $2,000 paid out to North Lawndale residents.

One hundred dollars per block, paid directly to the person who did the work on their own street. That’s 20% of the pilot goal — done, documented, and paid — with 80 blocks and 80 paid residents to go.

A scan is no longer a photo with a pin. It’s a measured walk through real space.

Every scan session now captures three streams at once: LiDAR depth, camera poses, and 1Hz GPS — uploading in chunks to cloud storage as the user walks.

158 scan sessions captured so far. Each one is a stretch of North Lawndale measured twice — once by the camera, once by the satellite.

The reconstruction pipeline fuses all that depth into per-block 3D — and it matches the street to the meter.

The ARKit camera poses anchoring each model have been verified metric-accurate against GPS. What comes out the other end is a block-by-block three-dimensional model of the neighborhood, built from residents’ walks.

Let’s be clear about what this is for. It’s not a game and it’s not a toy rendering. It’s accountability infrastructure. When a block gets cleaned, nobody has to take our word for it — we can show the block before and after, in three dimensions, measured the same way both times. A sponsor’s $100 buys a cleanup. The twin proves it.

A litter-classifier model is in training — partly on our own blocks.

The model trains on TACO, the open trash-annotation dataset, plus field-labelled crops from our own scans. North Lawndale’s litter is specific, and a model trained partly on our own blocks will recognize it better than one trained on the internet’s. When it’s ready, identification moves on-device: the app names what it sees while you scan, no upload required.

iOS v1.0.8 is in TestFlight — and the honor wall is up.

Version 1.0.8 (build 181) carries the capture and upload work above. On the site, a new honor wall now recognizes our major sponsors — the organizations that have given $2,000 or more — starting with UIC. The people and institutions funding this work should be named where everyone can see them.

20 down, 80 to go. Here’s how the next block gets funded.

$100 = one cleanup, one paid resident, one before-and-after photo set in your inbox.
$2,000 = your name on the honor wall alongside UIC, a digital certificate, a 501(c)(3) tax receipt, a feature in this newsletter, and social-media recognition.

Next issue: more blocks, more sessions, and the first before-and-after you can orbit.

The twin is about to start doing the talking — a cleaned block you can look at from every angle, before and after, measured both times.

Dr. King’s apartment is six blocks away. The work continues.

— The team at Global Community Cleaners NFP