DARIL: When Imitation Learning outperforms Reinforcement Learning in Surgical Action Planning

TL;DR

This study compares imitation learning and reinforcement learning for surgical action planning, finding that IL outperforms RL in accuracy and consistency, challenging common assumptions about RL's advantages in sequential decision tasks.

Contribution

The paper provides the first comprehensive comparison of IL versus RL in surgical action planning, demonstrating IL's superior performance in this domain.

Findings

IL achieves higher action recognition and prediction accuracy than RL.

RL approaches underperform significantly compared to IL in this task.

Distribution matching favors IL, questioning RL's assumed advantages in sequential decision making.

Abstract

Surgical action planning requires predicting future instrument-verb-target triplets for real-time assistance. While teleoperated robotic surgery provides natural expert demonstrations for imitation learning (IL), reinforcement learning (RL) could potentially discover superior strategies through self-exploration. We present the first comprehensive comparison of IL versus RL for surgical action planning on CholecT50. Our Dual-task Autoregressive Imitation Learning (DARIL) baseline achieves 34.6% action triplet recognition mAP and 33.6% next frame prediction mAP with smooth planning degradation to 29.2% at 10-second horizons. We evaluated three RL variants: world model-based RL, direct video RL, and inverse RL enhancement. Surprisingly, all RL approaches underperformed DARIL--world model RL dropped to 3.1% mAP at 10s while direct video RL achieved only 15.9%. Our analysis reveals that distribution matching on expert-annotated test sets systematically favors IL over potentially valid RL policies that differ from training demonstrations. This challenges assumptions about RL superiority in sequential decision making and provides crucial insights for surgical AI development.