Calibration-First Reward-Component Auditing for Reinforcement Learning Control in Smart Greenhouses

[Submitted on 12 Jul 2026]

View PDF HTML (experimental)

Abstract:Greenhouse reinforcement learning can test climate-control ideas at a speed and scale that is difficult to achieve with crop experiments alone. For smart-greenhouse control, however, a single simulator return is not enough: a grower or control engineer also needs to know when the policy heats, enriches CO2, vents, manages humidity, deploys screens, or uses this http URL propose a reproducible calibration-first reward audit framework that keeps named greenhouse-control reward components comparable across simulator training, facility-adapted rollouts, logged Autonomous Greenhouse Challenge records, and actuator-rule distillation. In GreenLight-Gym, the framework decomposes the scalar reward into conditional temperature, CO2, humidity and vapor-pressure-deficit, screen, and actuation-proxy terms; adapts GreenLight to the second Autonomous Greenhouse Challenge logged climate traces; and scores the same components on logged greenhouse data.

Submission history

From: Yaojun Wang [view email]
[v1] Sun, 12 Jul 2026 07:37:54 UTC (528 KB)