The driving digital twin, re-aimed at the sky. A Doppler-radar detect-and-avoid safety-cage flies beside a BVLOS courier, tracks every intruder, forecasts the closest point of approach, and re-plans by the rules of the air before a conflict arrives. Fully client-side, flown in the browser.

A planned drone route is the easy part. The hard part is everything else moving in the air near it. This tool takes the continuity-and-physics idea behind the driving digital twin and re-aims it at low-level BVLOS airspace: a Doppler-radar detect-and-avoid safety-cage that runs beside a self-flying courier, tracks every intruder, forecasts the closest point of approach, and re-plans by the rules of the air before a conflict arrives. It is fully client-side, so you fly it yourself in the browser, with no backend and no install.
Watch a BVLOS courier fly a looping route by itself while intruders, a crossing drone, a descending bird, a helicopter transit, fly their own tracks. The twin rebuilds the static world as an adaptive voxel octree, finds the movers with radar, and colors the course as it re-plans: grey to red when a conflict is fresh, back to blue once it settles. Drag to orbit, scroll to zoom, switch scenarios. A live panel scores the twin's coasting forecast against a memoryless snapshot, so you can watch the snapshot diverge while the twin stays bounded through signal loss.
This is the airspace sibling of the road-focused research build Digital Twin Perception, and it shares the thesis of Your Perception Stack Is a Snapshot Machine: perception is a world model that persists and accumulates evidence over time, not a stream of independent snapshots. Here that argument is re-aimed from the road to the sky.
It is a research rig on a synthetic scenario, built on public data, real DGM terrain and OpenStreetMap obstacles, and on public standards. It is not a product and not a certified detect-and-avoid system.

A self-hosted time-tracking and resource-planning tool for startups and small teams. Live timer, budget-vs-actual analytics, team sentiment, all on your own server. Free community edition plus a Pro version.

Does your perception carry physics forward, or judge each frame alone? Pick a scenario, toggle a per-frame stack against a continuity-gated world model, and watch whether the road user stays tracked through occlusion. Then see where your stack lands and get an evidence-gap report.

Flat object lists drop what they have not seen before. This demo abstracts each detection up a semantic taxonomy to the most specific level it can safely justify, and cross-checks it with an independent segmentation path.