MOON / FIRST DAWN V3 / DOCUMENTATION

Explorer — drivable lunar rover

Status: included in this release; the public catalog advertises driving when the matching runtime is installed.

Explorer fulfills Corey's goal of a working first-person and third-person rover with convincing lunar physics and beautiful graphics. Previous work added rocket cameras; no drivable rover existed when this branch began. The simulator and Guide action-taking remain separate future work.

The first vehicle is a researched, manufactured Explorer exploration rover with six large compliant mesh wheels, independent suspension, a low chassis, panoramic camera dome, headlights and a recovery mechanism. It uses the existing world, resources and permissions. Research requires two continuously powered Minds for its driving benefits. No free vehicle or account inventory grants in the public game.

Use Rapier rigid-body dynamics at a fixed time step, lunar gravity 1.62 m/s², suspension contacts, traction limits, braking, airborne motion, landings and obstacle collision. Render wheel travel, steering and rotation from physics. Keep terrain collision independent of visual LOD. Driving must feel immediate while authoritative server state bounds movement; observers see the same rover. Factory ticks and ordinary worker routing remain separate.

Controls: keyboard throttle/brake/steer, first/chase camera switch, mouse look, headlights, park/exit, recovery. UI explains research/build/drive and current support or battery restrictions. Disconnect, tab blur, paused colony and lost support must stop accepting throttle and brake safely. Exiting cannot leave a held accelerator. Only the owner may take control; concurrent tabs must not fight. Persist the rover's latest physical pose and energy without corrupting world transactions.

Validation must cover real public API authorization, stale/reordered input, control timeout, support loss, restart, fabrication cost and discovery; measured driving/turning/braking/jumps/landings, varying frame rates and terrain transitions; both camera views and actual rendered model/terrain. Run rendered checks on a CPU-only environment away from the tower GPU. Test multiplayer observation. Complete existing regression checks and protected staging/publication checks with fresh private backups before updating V3. Preserve V1/V2 and all current players/worlds. Record deployment, manual, devlog, board and cold-start handoff.

AI visual driving is a later extension of the same camera and bounded control interface. Do not claim that an autonomous vision driver or forward-simulation API ships with the human-driving feature unless it is actually implemented and verified.

Sources: Rapier vehicle controller, NASA lunar gravity, NASA superelastic tires. Soil contact and vehicle tuning are a game approximation, not an engineering certification.

Historical verification checkpoint — September 10

403CPU/API tests pass. New camera checks execute the real update path at15,30and60fps, verify first-person heading, close chase distance and immediate retraction before a wall. The actual exported GLB wheel joints coincide with physics and the fallback model; routing clearance now contains the animated mesh. The chase controller no longer competes with OrbitControls. These are numerical/geometry tests, not rendered browser verification. Actual first/chase and multiplayer screenshots, protected deployment and public validation remain required.

Camera-button follow-up: the rover HUD now uses data-trip-view so the main orbit/surface view binding cannot overwrite it. A CPU HTML check reproduced the collision before the change and confirms separate controls afterward; three camera/geometry tests and production build pass. This does not replace actual browser verification.

Rendered validation — September 10, 13:10 UTC

Integrated CPU first-person, chase and observer checks passed with28.77m actual keyboard-driven travel and no browser errors. The nearby-rover entry path now keeps the existing loaded terrain frame rather than disposing the landscape; a regression failed on the old entry behavior, then passed against a clean disposable server. The first-person screenshot shows continuous ground to the horizon. An earlier green attempt was invalid because its fixture failed to bind the occupied test port; it is not counted as a passing render.

Camera rotation and projection changes now invalidate the visible-terrain cache even when the camera position is unchanged. Numerical tests cover stationary turns, FOV changes and unchanged-view reuse. No tower GPU was used. Pending: merge current public logistics/Industry controls, final integrated regressions, fresh backup and protected deployment. No live rover is claimed yet.

Included with the driving release

All430 integrated CPU/API tests pass after merging the current logistics fixes, asset Priority controls and collapsible HUD. The runtime package now contains its exact pinned production dependencies, including Rapier, so deployment does not borrow a physics engine from another release. The final browser/production verification receipts are recorded in the project handoff; the paragraphs above retain earlier candidate checkpoints.

Explorer naming update — September 10

The rover is now named Explorer, including its chassis lettering. The existing field-trip document/model paths, explorer robot role, surface-mobility research ID and field-trip-v1 protocol remain stable for existing vehicles and clients. The launch release passed 430 CPU/API tests and CPU-rendered first-person, chase and second-player observation checks.

Handling and smoothness — September 10

Explorer now has a 520 kg simulated chassis (previously 420 kg), more suspension damping and steering that becomes gentler as speed rises. Powered cruising speed is about 32 km/h; downhill momentum still needs braking. Lunar gravity remains 1.62 m/s², with real airtime over larger drops. Ease off the throttle before tight turns.

The local driving view blends both position and orientation when reconciling the server, and interpolates chassis and wheel motion between fixed physics steps. Nearby collision terrain reuses overlapping height samples at unchanged 2 m resolution; visual terrain builds have a smaller per-frame budget while driving. These changes reduce measured CPU spikes and camera correction snaps; frame rate still depends on the device and scene. Refresh the game after this release to load matching driving physics.