ORCA Marine EcOsystem Reasoning with Collaborative Agents

Background The marine ecosystem plays a vital role in supporting livelihoods, food security, biodiversity, maritime transportation, coastal resilience, and the blue economy. Every day, vast volumes of satellite Earth Observation and oceanographic data, including Sea Surface Temperature (SST), chlorophyll concentration, and weather forecasts, are generated by ISRO and other global agencies. Marine stakeholders such as fishermen, researchers, coastal authorities, disaster management agencies, and maritime operators rely on timely access to oceanographic and meteorological information for operational planning and decision-making. As the volume, diversity, and complexity of marine data continue to increase, there is a growing need for an intelligent conversational platform that enables users to interact naturally with marine information, ask questions, explore scenarios, and receive synthesized, evidence-based recommendations tailored to their context. Advances in Agentic AI, conversational intelligence, and geospatial technologies present an opportunity to fundamentally transform how marine information is accessed and utilized. By integrating satellite Earth Observation data with autonomous AI agents, intelligent conversational decision-support systems should be able to answer questions such as 'Where is the nearest Potential Fishing Zone today?', 'Is it safe to venture into the sea tomorrow morning?', or 'What are the tide, weather, and sea conditions near my fishing location?' and provide explainable, context-aware recommendations. Description Develop an Agentic AI-powered conversational platform that enables users to access, analyze, and reason over marine information using natural language. The platform should autonomously interpret user intent, decompose complex requests into executable tasks, coordinate multiple specialized AI agents, retrieve relevant marine and geospatial datasets, perform spatial-temporal reasoning, and synthesize actionable recommendations through a conversational interface. The solution should be capable of integrating information from multiple sources, including satellite Earth Observation products, GIS layers, weather services, oceanographic observations, and marine advisories available in the public domain. Typical user queries include: • Where is the nearest Potential Fishing Zone (PFZ) today? • Is it safe to venture into the sea tomorrow morning? • What are the tide, weather, and sea conditions near my fishing location? • Are there any lightning or cyclone alerts in my area? • Which regions show high chlorophyll concentration and favourable sea surface temperature? • What is the safest route for a fishing vessel considering weather and sea-state conditions? • Why has fish productivity declined in a particular coastal region? • Which fishing zones should be avoided due to hazardous marine conditions or geofencing restrictions? The platform should not merely retrieve information from individual datasets but intelligently correlate observations from multiple sources, explain the reasoning behind its recommendations, and present insights through conversational responses, maps, alerts, and interactive geospatial visualizations. Expected Solution Participants are expected to develop an Agentic AI-powered Marine Intelligence Platform that leverages collaborative AI agents, geospatial technologies, and satellite Earth Observation data to provide intelligent conversational decision support. The solution should demonstrate the core principles of Agentic AI, including autonomous planning, reasoning, tool selection, task execution, collaboration among specialized agents, and explainable decision-making. The platform should be capable of: • Understanding user intent expressed in natural language. • Automatically identifying the language of the user's query and responding in the same language, with emphasis on supporting Indian regional languages. • Supporting contextual, multi-turn conversations that enable users to refine queries and explore related scenarios. • Autonomously discovering, retrieving, and integrating relevant satellite, marine, meteorological, and geospatial datasets. • Performing spatial, temporal, and contextual reasoning by correlating observations from multiple heterogeneous data sources. • Generating explainable, evidence-based recommendations supported by maps, charts, geospatial visualizations, and marine advisories. • Enhancing fishermen safety through proactive alerts for adverse weather, high waves, lightning, cyclones, and other hazardous marine conditions. • Providing geofencing-based notifications when approaching international maritime boundaries, restricted waters, marine protected areas, ecologically sensitive zones, or other predefined operational boundaries. • Assisting with route optimization, safe navigation, and operational planning based on prevailing and forecast marine conditions. • Delivering reliable recommendations together with the supporting evidence and reasoning used to derive each response. Participants are encouraged to design a modular multi-agent architecture comprising specialized AI agents for planning, marine data discovery, weather intelligence, ocean analytics, geospatial reasoning, risk assessment, visualization, reporting, and user interaction. The architecture should demonstrate autonomous collaboration among agents to solve complex marine intelligence problems while providing an intuitive conversational experience.