Shortcut Trajectory Planning simplifies offline RL with efficient trajectory generation

The authors propose Shortcut Trajectory Planning (STP), an offline model-based reinforcement learning framework that uses shortcut models as efficient trajectory generators to address the high inference cost of diffusion-based planners and the instability of two-stage distillation.

• STP trains a conditional shortcut trajectory model in a single stage, supporting adjustable one-step and few-step inference through step-size conditioning.
• The framework selects candidate plans using a critic augmented with feasibility-aware correction.
• Across standard D4RL benchmarks including locomotion, navigation, manipulation, and dexterous control tasks, STP achieves strong performance while simplifying the training pipeline for fast generative planning.