Code, trained checkpoints and trial-level data for "Risk-Aware Greenhouse Robot Navigation with Deep Q-Networks and RRT*: Checkpoint Sensitivity and the Role of the Objective"

This record contains the code and data supporting the article "Risk-Aware Greenhouse Robot Navigation with Deep Q-Networks and RRT*: Checkpoint Sensitivity and the Role of the Objective" (N. Karakoç, İ. Ö. Bucak). It includes the simulation framework of the geothermal tomato greenhouse, the DQN training script, the RRT* and cost-aware RRT* (RRT*-CA) planners, the unified timing harness, the final-episode and best-window checkpoints of all ten trained DQN agents with their training logs, the trial-level results, and the analysis and figure scripts used in the article.