Team NeuroNexus - Interplanetary Survival Guide: Martian Map
NASA Space Apps Challenge 2026 · branch main
Demo Videos
Official judging cut: 30-second subtitled demo.
Full demo: about 240 seconds.
Live links
The live app runs on free servers that sleep when idle. Opening the landing page wakes them; give it about a minute
before entering the console.
Setup first: SETUP_INSTRUCTIONS.md contains the complete installation, NASA data, Gemini API-key, backend/frontend startup, troubleshooting, and demo instructions.
Setup instructions:
https://github.com/kmmohaimenulhaque-code/MartianMap-NeuroNexus/blob/main/SETUP_INSTRUCTIONS.md
OUR CHALLENGE:
NASA has explored Mars robotically for decades, studying its extreme environment, mapping its surface, and collecting many types of data.
Far from home, the first astronauts to set foot on Mars will want the best map possible, with information on the details of their routes and destinations, updates on current conditions, and the data necessary to make safe and scientifically meaningful decisions.
Project Description
NeuroNexus is an interactive planetary mission-planning and scientific intelligence system for Mars. It combines authoritative Mars datasets, deterministic terrain analysis, multi-objective route generation, and transparent AI interpretation to support mission planning and exploration decisions.
The goal is not to declare a universally "safe" place on Mars. NeuroNexus provides transparent evidence and derived planning aids so explorers can compare terrain, environment, science opportunities, and operational constraints before deciding on a route or site.
Why the shortest route is a trap
The Gale to Escalante benchmark contains six deterministic candidates. The low-distance route is 1,464.72 km with 31,332 m of climb and a 94.3 risk proxy. A 480 m longer candidate cuts climb to 22,597 m and estimated EVA time to 623.74 h. The Pareto set makes that trade-off visible instead of hiding it inside one shortest-path answer.
Three scientific and architectural innovations
- Empirical scale-bias calibration: Curiosity end-of-drive IMU tilt points at Gale and out-of-sample Jezero show that 463 m MOLA underestimates p90 steep ground by 1.79x at Gale and 2.35x at Jezero. A 1 m HiRISE surface evaluated at a 3 m rover-scale baseline reproduces felt tilt at 0.99x at Gale and 0.92x at Jezero.
- 668-sol spatio-temporal survival matrix: Across Mars Year 34, the planner records WITHIN 0, REVIEW 570 and EXCEEDS 98 status outcomes as dust optical depth changes. It also identifies 941 sq km of Gale Potential Exploration Zones under the documented walkability, slope and datum filters.
- Epistemic humility and guarded 27-tool AI: The interface exposes a 7.3 m seam bias, 215-day RAD archive lag, unavailable thermal data when no reference is within radius, and sub-meter rock limits. Deterministic graph search owns route geometry; Mars Intelligence only explains structured results and links to 76 canonical sources.
Why this matters right now in 2026
- Orbital imaging triage today: HiRISE covers less than 2% of Mars. The scale-bias model and seam detector identify where coarse orbital maps may hide steep terrain before a stereo pair is tasked or a route crosses a CTX and MOLA boundary.
- Earth analog crews and Artemis lunar EVA rehearsal today: NASA CHAPEA, MDRS and HI-SEAS crews, plus Artemis III planners, can rehearse multi-objective EVA trade-offs by bridging coarse LOLA grids to 1 m LRO NAC strips under lighting and shadow constraints.
- Zero-install planetary science: A 60 FPS browser console with a 45 MB memory footprint unifies 76 NASA PDS, USGS SIM 3292, ASU THEMIS IR-PBT v2.0, MSL RAD and NASA Ames MGCM sources without a workstation GIS license.
What it does
- 2D Mars map + USGS scientific map: synchronized exploration views with 2,052 USGS/IAU named Martian features.
- Interactive 3D Mars: ArcGIS Maps SDK SceneView using Mars 2000 coordinates, public Mars terrain/imagery services, persistent annotations, USGS feature classes, geology, landing sites, rover locations, and terrain overlays.
- AI Route Designer: deterministic multi-objective route generation over bounded NASA MOLA terrain, with candidates for EVA, terrain, science, operational, distance, and terrain-risk objectives.
- Terrain intelligence: NASA MOLA elevation with derived slope, roughness, local relief, terrain burden, and a clearly labelled NeuroNexus walkability score.
- Environmental context: Mars weather/environment assessment, NASA Ames MY34 dust modelling, historical THEMIS thermal observations, and site-science provenance.
- Science + exploration context: ancient-habitability evidence, documented accessible-ice study regions, potential exploration zones, landing sites, rover traverses, and human-exploration context.
- Mission awareness: NASA/JPL near-Mars small-body close-approach tracking.
- USGS route snapshot + exports: route visualisation against USGS SIM 3292 geology plus JSON/CSV exports.
- Mars Intelligence: Google Gemini interprets structured NeuroNexus results and explains trade-offs; Gemini does not generate route geometry or alter deterministic numerical results.
Live demo
Two ways to run it, and they are not the same thing:
여기까지가 README 앞부분입니다. 나머지는 원문에서 읽을 수 있습니다. README 전문 보기 ↗
