cs.RONov 19, 2026

Hindsight Self-Distillation: Teaching Robots to Act from the Present

Authors: Haisen He, Yechi Liu, Tianle Shi, Haoyun Liu, Chuanzhang Meng, Tiancai Jiao, Yifan Bai, Xinyuan Chang, 

Organizations: Shenzhen University of Advanced Technology · Southern University of Science and Technology · Primotion · New Laboratory of Pattern Recognition (NLPR), Institute of Automation, Chinese Academy of Sciences · 

Abstract

A robot must choose an action before it can see the outcome. In a recorded demonstration, later camera frames show what happened after the robot moved. Hindsight Self-Distillation (HSD) uses these later frames to teach a policy what to extract from current images. During training, a teacher sees both the current and later frames, while a student sees the current frames. They share a visual encoder and an action predictor. Each branch summarizes its images with a small set of feature vectors. These visual summaries supply all visual input to the action predictor. The teacher learns from demonstrated actions. For each noisy action sequence, the student learns from either the demonstration or the teacher's denoising prediction. It also learns to match the teacher's visual summaries. We analyze which aspects of the teacher's predictions the student can recover from its inputs. At deployment, the student encodes the current images once and reuses its summaries at every denoising step. HSD achieves average success rates of 98.50% on LIBERO and 92.31% on RoboTwin 2.0.

Figures & tables

Appendix figures & tables1 asset

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList