mlsanityData to 14 Oct 2026

Saliency-Aware Quantized Imitation Learning for Efficient Robotic Control

arXiv 2505.15304 ↗First posted 21 May 2025cs.ROcited by 12

On the map: Vision-language-action models → Efficient VLA inference and adaptation

also in 1 more
Abstract (arXiv)

Deep neural network (DNN)-based policy models, such as vision-language-action (VLA) models, excel at automating complex decision-making from multi-modal inputs. However, scaling these models greatly increases computational overhead, complicating deployment in resource-constrained settings like robot manipulation and autonomous driving. To address this, we propose Saliency-Aware Quantized Imitation Learning (SQIL), which combines quantization-aware training with a selective loss-weighting strategy for mission-critical states. By identifying these states via saliency scores and emphasizing them in the training loss, SQIL preserves decision fidelity under low-bit precision. We validate SQIL's generalization capability across extensive simulation benchmarks with environment variations, real-world tasks, and cross-domain tasks (self-driving, physics simulation), consistently recovering full-precision performance. Notably, a 4-bit weight-quantized VLA model for robotic manipulation achieves up to 2.5x speedup and 2.5x energy savings on an edge GPU with minimal accuracy loss. These results underline SQIL's potential for efficiently deploying large IL-based policy models on resource-limited devices.

Key facts. Cited by 12 papers in the map (data through 14 Oct 2026). Placed in Efficient VLA inference and adaptation, one of 13 directions in Vision-language-action models. Builds on 44 of its 60 references that are in the map. First posted on arXiv 21 May 2025.

Builds on

44 of 60 references.

References · 44

  1. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  2. 2014Very Deep Convolutional Networks for Large-Scale Image Recognitioncited by 15,364
  3. 2021LoRA: Low-Rank Adaptation of Large Language Modelscited by 14,380
  4. 2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,974
  5. 2020Scaling Laws for Neural Language Modelscited by 6,586
  6. 2023QLoRA: Efficient Finetuning of Quantized LLMscited by 3,418
  7. 2023Sigmoid Loss for Language Image Pre-Trainingcited by 3,109
  8. 2023RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Controlcited by 3,080
  9. 2024OpenVLA: An Open-Source Vision-Language-Action Modelcited by 2,887
  10. 2017CARLA: An Open Urban Driving Simulatorcited by 2,449
  11. 2024π0: A Vision-Language-Action Flow Model for General Robot Controlcited by 2,407
  12. 2016Xception: Deep Learning with Depthwise Separable Convolutionscited by 2,332
32 more
  1. 2023PaLM-E: An Embodied Multimodal Language Modelcited by 2,071
  2. 2022RT-1: Robotics Transformer for Real-World Control at Scalecited by 2,061
  3. 2022GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformerscited by 1,919
  4. 2017Quantization and Training of Neural Networks for Efficient Integer-Arithmetic-Only Inferencecited by 1,593
  5. 2020D4RL: Datasets for Deep Data-Driven Reinforcement Learningcited by 1,427
  6. 2023LIBERO: Benchmarking Knowledge Transfer for Lifelong Robot Learningcited by 1,410
  7. 2022SmoothQuant: Accurate and Efficient Post-Training Quantization for Large Language Modelscited by 1,386
  8. 2018DeepMind Control Suitecited by 956
  9. 2019Stabilizing Off-Policy Q-Learning via Bootstrapping Error Reductioncited by 840
  10. 2023BridgeData V2: A Dataset for Robot Learning at Scalecited by 657
  11. 2024QuaRot: Outlier-Free 4-Bit Inference in Rotated LLMscited by 634
  12. 2019Learned Step Size Quantizationcited by 628
  13. 2020Up or Down? Adaptive Rounding for Post-Training Quantizationcited by 621
  14. 2016DoReFa-Net: Training Low Bitwidth Convolutional Neural Networks with Low Bitwidth Gradientscited by 569
  15. 2018Quantizing deep convolutional networks for efficient inference: A whitepapercited by 521
  16. 2021BRECQ: Pushing the Limit of Post-Training Quantization by Block Reconstructioncited by 479
  17. 2018PACT: Parameterized Clipping Activation for Quantized Neural Networkscited by 467
  18. 2018An Algorithmic Perspective on Imitation Learningcited by 459
  19. 2019Exploring the Limitations of Behavior Cloning for Autonomous Drivingcited by 346
  20. 2024TinyVLA: Towards Fast, Data-Efficient Vision-Language-Action Models for Robotic Manipulationcited by 339
  21. 2019A Study of BFLOAT16 for Deep Learning Trainingcited by 232
  22. 2023LoftQ: LoRA-Fine-Tuning-Aware Quantization for Large Language Modelscited by 196
  23. 2021End-to-End Urban Driving by Imitating a Reinforcement Learning Coachcited by 194
  24. 2021PTQ4ViT: Post-training quantization for vision transformers with twin uniform quantizationcited by 179
  25. 2017Visualizing and Understanding Atari Agentscited by 141
  26. 2018CIRL: Controllable Imitative Reinforcement Learning for Vision-based Self-drivingcited by 97
  27. 2018NVIDIA Tensor Core Programmability, Performance & Precisioncited by 79
  28. 2017Towards Effective Low-bitwidth Convolutional Neural Networkscited by 78
  29. 2024LiDAR-PTQ: Post-Training Quantization for Point Cloud 3D Object Detectioncited by 29
  30. 2019QuaRL: Quantization for Fast and Environmentally Sustainable Reinforcement Learningcited by 19
  31. 2023Poster: Self-Supervised Quantization-Aware Knowledge Distillationcited by 11
  32. 2018Low Precision Policy Distillation with Application to Low-Power, Real-time Sensation-Cognition-Action Loop with Neuromorphic Computingcited by 2

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 1; 2026: 11.

12 papers.

  1. 2025A Survey on Efficient Vision-Language-Action Modelscited by 34
  2. 2026FASTER: Rethinking Real-Time Flow VLAscited by 21
  3. 2026HBVLA: Pushing 1-Bit Post-Training Quantization for Vision-Language-Action Modelscited by 2
  4. 2026Revisiting Parameter Redundancy in Vision-Language-Action Models: Insights from VLM-to-VLA Adaptationcited by 2
  5. 2026QuantWAMs: Calibrating at the Right Granularity for World Action Modelscited by 2
  6. 2026Device-Conditioned Neural Architecture Search for Efficient Robotic Manipulationcited by 0
  7. 2026APEX-RBD: Mixed-Precision Exploration Framework for Hardware-Efficient Robot Dynamics Accelerator Designcited by 0
  8. 2026Efficient Vision-Language-Action Management and Serving for Robot Factoriescited by 0
  9. 2026Predict Before You Deploy: Offline Prediction of Quantization-Induced Task Degradation for World Action Modelscited by 0
  10. 2026ActTune: Action-Aware Precision and GPU Operating-Point Adaptation for Energy-Efficient Vision-Language-Action Inferencecited by 0
  11. 2026Lamarck's Driving School: Discovering Autonomous Driving Training Strategies through Evolutionary Competitioncited by 0
  12. 2026Recovering Low-Bit Failures in Vision-Language-Action Policies via Quantization-Conditioned Reinforcement Learningcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2024Quantization-Aware Imitation-Learning for Resource-Efficient Robotic Control
  2. 2026QVLA: Not All Channels Are Equal in Vision-Language-Action Model's Quantization
  3. 2025SRPO: Self-Referential Policy Optimization for Vision-Language-Action Models
  4. 2024ViSaRL: Visual Reinforcement Learning Guided by Human Saliency
  5. 2025Large VLM-based Vision-Language-Action Models for Robotic Manipulation: A Survey
  6. 2026AtomVLA: Scalable Post-Training for Robotic Manipulation via Predictive Latent World Models
  7. 2025From Code to Action: Hierarchical Learning of Diffusion-VLM Policies
  8. 2026ARP: Enhancing Quantized Skill Abstractions via Visual Alignment and Iterative Refinement for Robotic Manipulation ↗