mlsanityData to 14 Oct 2026

How to Mitigate the Distribution Shift Problem in Robotics Control: A Robust and Adaptive Approach Based on Offline to Online Imitation Learning

arXiv 2605.25414 ↗First posted 25 May 2026cs.ROcited by 0

On the map: Offline RL and imitation learning from data → Imitation learning methods

Abstract (arXiv)

Distribution shift in imitation learning refers to the problem that the agent cannot plan proper actions for a state that has not been visited during the training. This problem can be largely attributed to the inherently narrow state-action coverage provided by expert demonstrations over the full environment. In this paper, we propose a robust offline to adaptive online imitation learning framework that handles the distribution shift problem in a lifelong, multi-phase scheme. In the offline learning phase, we leverage supplementary demonstrations to broaden the state-action coverage of the policy by utilizing a discriminator to effectively train the policy with supplementary demonstrations, thereby enhancing the robustness of the policy to distribution shift. In the subsequent online inference phase, our framework detects the occurrence of distribution shift and conducts self-supervised imitation learning from online experiences to adapt the policy to the online environments. Through extensive evaluations in MuJoCo environments, we demonstrate that our method exhibits better robustness to distribution shift and better adaptation performance to online environments than the baseline algorithms, which indicates superior performance of our framework against the distribution shift.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Imitation learning methods, one of 13 directions in Offline RL and imitation learning from data. Builds on 23 of its 35 references that are in the map. First posted on arXiv 25 May 2026.

Builds on

23 of 35 references.

Closest ancestors · 6

Same direction, newest first

  1. 2024Stable-BC: Controlling Covariate Shift with Stable Behavior Cloning12 Aug
  2. 2023CCIL: Continuity-based Data Augmentation for Corrective Imitation Learning19 Oct
  3. 2023Improving Behavioural Cloning with Positive Unlabeled Learning27 Jan
  4. 2022Learning from Guided Play: Improving Exploration for Adversarial Imitation Learning with Simple Auxiliary Tasks30 Dec
  5. 2022Multi-Task Imitation Learning for Linear Dynamical Systems1 Dec
  6. 2022Discriminator-Weighted Offline Imitation Learning from Suboptimal Demonstrations20 Jul

Other references · 17

  1. 2015Trust Region Policy Optimizationcited by 2,580
  2. 2016Generative Adversarial Imitation Learningcited by 1,597
  3. 2020D4RL: Datasets for Deep Data-Driven Reinforcement Learningcited by 1,427
  4. 2016Least Squares Generative Adversarial Networkscited by 1,001
  5. 2017Towards Principled Methods for Training Generative Adversarial Networkscited by 464
  6. 2024Mobile ALOHA: Learning Bimanual Mobile Manipulation with Low-Cost Whole-Body Teleoperationcited by 391
  7. 2017Measuring Catastrophic Forgetting in Neural Networkscited by 360
  8. 2016Data-Efficient Off-Policy Policy Evaluation for Reinforcement Learningcited by 286
  9. 2017Positive-Unlabeled Learning with Non-Negative Risk Estimatorcited by 219
  10. 2017DART: Noise Injection for Robust Imitation Learningcited by 191
  11. 2017Safe Policy Improvement with Baseline Bootstrappingcited by 144
  12. 2016Class-prior Estimation for Learning from Positive and Unlabeled Datacited by 46
5 more
  1. 2025Mahalanobis++: Improving OOD Detection via Feature Normalizationcited by 32
  2. 2024SAFE-GIL: SAFEty Guided Imitation Learning for Robotic Systemscited by 14
  3. 2024A Likelihood Ratio-Based Approach to Segmenting Unknown Objectscited by 12
  4. 2024Beyond Imitation: A Life-long Policy Learning Framework for Path Tracking Control of Autonomous Drivingcited by 4
  5. 2025Dynamic Contrastive Skill Learning with State-Transition Based Skill Clustering and Dynamic Length Adjustmentcited by 3

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Balance Equation-based Distributionally Robust Offline Imitation Learning
  2. 2021Mitigating Covariate Shift in Imitation Learning via Offline Data Without Great Coverage
  3. 2023GO-DICE: Goal-Conditioned Option-Aware Offline Imitation Learning via Stationary Distribution Correction Estimation
  4. 2024Efficient Imitation Learning with Conservative World Models
  5. 2023Offline Imitation Learning with Suboptimal Demonstrations via Relaxed Distribution Matching
  6. 2025Robust Offline Imitation Learning Through State-level Trajectory Stitching
  7. 2025Distributionally Robust Imitation Learning: Layered Control Architecture for Certifiable Autonomy
  8. 2026Distributionally Robust and Safe Imitation Learning