Title
Eliminating Transboundary Haze via Agentic AI and Microalgae Bio-Refineries: The Eagor-Core Framework for Palm Oil Estate Transformation
Authors
Abstract
Transboundary haze in Southeast Asia persists because fire remains the cheapest way to clear and prepare land on tropical peat, while the health and climate costs of burning fall on others. This short paper asks whether an alternative revenue stream from existing palm oil mill infrastructure, verified and paid for through an automated monitoring system, could change that private calculus. We present Eagor-Core, a conceptual agentic artificial intelligence (AI) framework built on the Energy–AI Governance Orchestration (EAGOR) architecture. Eagor-Core converts palm oil mill effluent (POME) lagoons into microalgae cultivation units and links their output to zero-burn payments verified through a three-tier monitoring pipeline of in-ground sensors, autonomous drones, and satellite Earth observation. At the physical layer, we formalize multi-pond control as a decentralized partially observable Markov decision process (Dec-POMDP) with explicit sensor-drift and fouling states, and propose multi-agent reinforcement learning (MARL) under centralized training with decentralized execution. A bounded, volume-weighted health index (the Finamatrix vGRE) modulates a normalized multi-objective reward. At the governance layer, adaptive conformal inference provides prediction intervals whose width determines whether the system acts autonomously, falls back to a conservative policy, or enters a safe state and requests a signed human-in-the-loop (HITL) decision. We derive the minimum conditional payment that makes zero-burn compliance privately rational, and extend it to independent smallholders, who manage about 40% of Indonesia’s oil palm area, through a mill-level cooperative with conditional dividends. We also provide an assumption-based techno-economic scoping estimate for a 10-hectare estate, an evaluation protocol, and a three-stage research and rollout agenda. We report no simulation results, field validation, or multi-seed experiments; the techno-economic values are assumption-based scoping estimates intended to guide future work.
Keywords
agentic AI; Energy–AI Governance Orchestration (EAGOR); microalgae; palm oil mill effluent; Dec-POMDP; multi-agent reinforcement learning
Classification-JEL
Q13, Q16, Q42, Q53, Q56, O33, C61, C63
Pages
109-143
How to Cite
Chan, L. C. W. J., & Pan, S. H. (2026). Eliminating Transboundary Haze via Agentic AI and Microalgae Bio-Refineries: The Eagor-Core Framework for Palm Oil Estate Transformation. Annals of Artificial Intelligence, 2026(1), 109-143.