mlsanityData to 7 Oct 2026

SAFARI: Safe and Active Robot Imitation Learning with Imagination

arXiv 2011.09586 ↗First posted 18 Nov 2020cs.RO cs.LGcited by 5

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

Abstract (arXiv)

One of the main issues in Imitation Learning is the erroneous behavior of an agent when facing out-of-distribution situations, not covered by the set of demonstrations given by the expert. In this work, we tackle this problem by introducing a novel active learning and control algorithm, SAFARI. During training, it allows an agent to request further human demonstrations when these out-of-distribution situations are met. At deployment, it combines model-free acting using behavioural cloning with model-based planning to reduce state-distribution shift, using future state reconstruction as a test for state familiarity. We empirically demonstrate how this method increases the performance on a set of manipulation tasks with respect to passive Imitation Learning, by gathering more informative demonstrations and by minimizing state-distribution shift at test time. We also show how this method enables the agent to autonomously predict failure rapidly and safely.

Key facts. Cited by 5 papers in the map (data through 7 Oct 2026). Placed in Imitation learning methods, one of 13 directions in Offline RL and imitation learning from data. Builds on 37 of its 49 references that are in the map. First posted on arXiv 18 Nov 2020.

Builds on

37 of 49 references.

Closest ancestors · 5

Same direction, newest first

  1. 2020Active Imitation Learning from Multiple Non-Deterministic Teachers: Formulation, Challenges, and Algorithms14 Jun
  2. 2019A Divergence Minimization Perspective on Imitation Learning Methods6 Nov
  3. 2018An Algorithmic Perspective on Imitation Learning16 Nov
  4. 2018Deep Imitative Models for Flexible Inference, Planning, and Control15 Oct
  5. 2018HG-DAgger: Interactive Imitation Learning with Human Experts5 Oct

Other references · 32

  1. 2015Dropout as a Bayesian Approximation: Representing Model Uncertainty in Deep Learningcited by 4,416
  2. 2016OpenAI Gymcited by 1,684
  3. 2016Generative Adversarial Imitation Learningcited by 1,576
  4. 2019Meta-World: A Benchmark and Evaluation for Multi-Task and Meta Reinforcement Learningcited by 1,168
  5. 2016End to End Learning for Self-Driving Carscited by 1,000
  6. 2018Exploration by Random Network Distillationcited by 966
  7. 2015End-to-End Training of Deep Visuomotor Policiescited by 847
  8. 2017Learning Robust Rewards with Adversarial Inverse Reinforcement Learningcited by 485
  9. 2017Time-Contrastive Networks: Self-Supervised Learning from Videocited by 453
  10. 2019Relay Policy Learning: Solving Long-Horizon Tasks via Imitation and Reinforcement Learningcited by 380
  11. 2019Learning Latent Plans from Playcited by 310
  12. 2017Neural Network Dynamics for Model-Based Deep Reinforcement Learning with Model-Free Fine-Tuningcited by 288
20 more
  1. 2017Deep Imitation Learning for Complex Manipulation Tasks from Virtual Reality Teleoperationcited by 286
  2. 2017One-Shot Imitation Learningcited by 232
  3. 2017One-Shot Visual Imitation Learning via Meta-Learningcited by 225
  4. 2017DART: Noise Injection for Robust Imitation Learningcited by 187
  5. 2015Deep Spatial Autoencoders for Visuomotor Learningcited by 149
  6. 2018One-Shot Imitation from Observing Humans via Domain-Adaptive Meta-Learningcited by 146
  7. 2020Can Autonomous Vehicles Identify, Recover From, and Adapt to Distribution Shifts?cited by 139
  8. 2017Trial without Error: Towards Safe Reinforcement Learning via Human Interventioncited by 102
  9. 2017Vision-Based Multi-Task Manipulation for Inexpensive Robots Using End-To-End Learning from Demonstrationcited by 100
  10. 2017Transferring End-to-End Visuomotor Control from Simulation to Real World for a Multi-Stage Taskcited by 66
  11. 2018Task-Embedded Control Networks for Few-Shot Imitation Learningcited by 65
  12. 2017Deep and Confident Prediction for Time Series at Ubercited by 50
  13. 2019Model-Predictive Policy Learning with Uncertainty Regularization for Driving in Dense Trafficcited by 50
  14. 2017A survey of robot learning from demonstrations for Human-Robot Collaborationcited by 38
  15. 2014Interactive Policy Learning through Confidence-Based Autonomycited by 29
  16. 2017DropoutDAgger: A Bayesian Approach to Safe Imitation Learningcited by 14
  17. 2019Model-based Behavioral Cloning with Future Image Similarity Learningcited by 8
  18. 2019Regularizing Trajectory Optimization with Denoising Autoencoderscited by 2
  19. 2017On the exact relationship between the denoising function and the data distributioncited by 1
  20. 2018Improving Model-Based Control and Active Exploration with Reconstruction Uncertainty Optimizationcited by 1

Built on by

Cited by year

Citing papers per year: 2021: 2; 2022: 2; 2023: 1.

5 papers.

  1. 2021Coarse-to-Fine Imitation Learning: Robot Manipulation from a Single Demonstrationcited by 92
  2. 2021Learning Multi-Stage Tasks with One Demonstration via Self-Replaycited by 21
  3. 2022Discriminator-Guided Model-Based Offline Imitation Learningcited by 17
  4. 2023Get Back Here: Robust Imitation by Return-to-Distribution Planningcited by 4
  5. 2022DcnnGrasp: Towards Accurate Grasp Pattern Recognition with Adaptive Regularizer Learningcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Robotic Manipulation via Imitation Learning: Taxonomy, Evolution, Benchmark, and Challenges
  2. 2024VITAL: Interactive Few-Shot Imitation Learning via Visual Human-in-the-Loop Corrections
  3. 2021Imitation Learning: Progress, Taxonomies and Challenges
  4. 2021Learning from Imperfect Demonstrations from Agents with Varying Dynamics
  5. 2026Difference-Aware Retrieval Policies for Imitation Learning
  6. 2025A Smooth Sea Never Made a Skilled SAILOR: Robust Imitation via Learning to Search
  7. 2026OSIL: Learning Offline Safe Imitation Policies with Safety Inferred from Non-preferred Trajectories
  8. 2025Robust Offline Imitation Learning Through State-level Trajectory Stitching