Enter the flyiverse.

A CARLA car controller built from an upstream-traced FlyWire circuit: a 30×30 grayscale camera input is mapped to cells and descending neurons drive steering, brakes, and throttle.

The boundary: The README explicitly says this is not a trained network; no driving behavior was independently tested.

An open simulation environment for fly biomechanics and control, with a rewritten FlyGym 2.x API announced in its README.

The boundary: This is an embodied-simulation toolkit; its README does not make it a whole-connectome dataset or validation result.

A digital-lifeform project built from a reported larval fruit-fly connectome and connected to a MuJoCo physics body.

The boundary: The repository describes a simulation built from larval wiring; source inspection does not validate biological behavior.

A browser game couples a FlyWire brain graph, MANC ventral nerve cord graph, and a flybody physics model through a modeled cross-specimen interface. Named-neuron stimulation feeds a hand-written gait; a separate trained walking-policy path bypasses the connectomes.

The boundary: The README explicitly documents cross-animal name matching, programmed gait, default kinematic assistance, and no quantitative whole-brain validation. Its separate trained walking policy is not connectome-driven.

A browser project combining a reported MaleCNS v1.0 whole-CNS simulation, physics body, arena, and connectome viewer.

The boundary: The README describes modeled dynamics and a physics body; the arena URL was checked only for HTTP reachability, not behavior.

A GPU LIF simulation couples a FlyWire v783 whole-brain graph to a NeuroMechFly/MuJoCo body with modeled sensory and motor layers, plasticity, and integration metrics.

The boundary: The repository's companion preprint and README report its own results; embodiment, sensory encoders, and integration metrics remain model layers.

A browser and Python laboratory runs a full FlyWire v783 LIF graph, drives selected sensory populations, and exposes reported circuit and lesion experiments with a kinematic body.

The boundary: Its README explicitly separates measured wiring from modeled transduction, point neurons, and kinematic body behavior.

A project for a simulated and DJI Tello flight controller based on FlyWire connectivity, LIF dynamics, optic-flow input, and decoded descending-neuron motor commands.

The boundary: README performance and flight claims are author-reported; it documents simplified flight physics and an arbitrary left/right proxy in its data.

A TypeScript project extracts a MaleCNS vision-to-wing subgraph, applies a fitted rate model, and couples rendered eye input to a cartoon flight body.

The boundary: The README calls the body and neuron dynamics cartoon/model layers and records unresolved flight limitations.

A Cocoa application runs a FlyWire whole-brain LIF network with C and Metal backends, exposes population inputs and outputs, and maps selected results to an animated fly.

The boundary: The README provides author-reported benchmarks; the animated behavior and sensory mappings are project layers over the connectome.

A research project couples a selected MANC flight or walking circuit to flybody physics at sub-millisecond steps and documents wiring-shuffle controls alongside the modeled layers.

The boundary: The README reports a single-model research result and explicitly documents no ground takeoff, weak walking, external calibration, and modeled electrical and muscle layers.