Post by Pabak Pany

In 2026, India's Chandrayaan-2 found water ice buried under the Moon's south pole. Nobody built the tool to find exactly how deep. Or how much. Or where to send the rover. We did. My team and I spent weeks obsessing over one question: what does it actually take to go from a radar signal 384,000 km away to a mission plan a rover can follow? The answer became PRISM — Polar Radar Ice Scout & Mission Planner. Attaching our full idea submission deck below. Here's what's inside: Why 32 years of lunar ice research still couldn't answer the mission planner's actual question How we pretrained a Vision Transformer on 5TB of Arctic radar data to teach it what lunar ice looks like — before it saw a single Moon pixel How DFSAR's two frequencies together give you ice depth — something no published paper has computed How DiffSAR inverts Maxwell's equations to estimate ice depth in metres, not just "ice or no ice" How a Science-LLM turns a radar map into a Chandrayaan-4 rover mission brief with 10+ annotated stops And why the whole thing runs on hardware ISRO already owns, at Rs. 0 Built for Problem Statement 8 of the Bharatiya Antariksh Hackathon 2026, under the mentorship of Dr. Rishitosh K. Sinha and Dr. Rajiv Bharti from the Physical Research Laboratory, Ahmedabad — the same scientists whose groundbreaking Chandrayaan-2 DFSAR paper in npj Space Exploration (May 2026) this work is built directly on top of. Chandrayaan-2 saw the signal. PRISM maps the ice. Chandrayaan-4 will mine it. 🇮🇳 Team Axiom_ai | KIIT University, Bhubaneswar SOUMYAJIT GHOSH | Pabak Pany | Jeetain Saini #BAH2026 #BharatiyaAntarikshHackathon #ISRO #Chandrayaan #LunarExploration #MoonIce #SpaceTech #AIForSpace #MachineLearning #DFSAR #RemoteSensing #PlanetaryScience #IndiaInSpace #Hackathon #KIIT #SpaceExploration #LunarScience #DeepLearning #ISRU #Chandrayaan4