Experimental Data and Plotting Scripts for DRL-BiRRT*-RC Robotic Manipulator Path Planning

This dataset contains run-level experimental records, processed tables, and plotting scripts supporting the evaluation of DRL-BiRRT*-RC, a robotic manipulator path-planning framework that combines point-cloud-conditioned deep reinforcement learning guidance, a bidirectional RRT* backbone, and conditional RRT-Connect recovery. The package includes the main comparison involving 12 planners across 7 planning cases, component ablation experiments with 5 configurations, scene-adaptation experiments with 4 configurations, training-objective ablation experiments with 4 objectives across 10 cases, and a checkpoint sweep involving a baseline and 18 policy checkpoints. Each planner or configuration was evaluated using 50 paired seeds per case. Run-level records are provided under data/runs/, while processed per-case and overall tables are provided under data/processed/. The accompanying Python scripts regenerate the data-driven figures reported in the associated paper. Detailed aggregation and time-limit conventions are provided in README.md and experiment_protocol.yaml. Tabular data and package-specific metadata are licensed under the Creative Commons Attribution 4.0 International License. The plotting scripts are licensed under the BSD 3-Clause License.