mlsanityData to 14 Oct 2026

DynaNav: Dynamic Feature and Layer Selection for Efficient Visual Navigation

arXiv 2509.21930 ↗First posted 26 Sep 2025cs.CV cs.ROcited by 0

Not on the map. Why?

Abstract (arXiv)

Visual navigation is essential for robotics and embodied AI. However, existing foundation models, particularly those with transformer decoders, suffer from high computational overhead and lack interpretability, limiting their deployment in resource-tight scenarios. To address this, we propose DynaNav, a Dynamic Visual Navigation framework that adapts feature and layer selection based on scene complexity. It employs a trainable hard feature selector for sparse operations, enhancing efficiency and interpretability. Additionally, we integrate feature selection into an early-exit mechanism, with Bayesian Optimization determining optimal exit thresholds to reduce computational cost. Extensive experiments in real-world-based datasets and simulated environments demonstrate the effectiveness of DynaNav. Compared to ViNT, DynaNav achieves a 2.26x reduction in FLOPs, 42.3% lower inference time, and 32.8% lower memory usage, while improving navigation performance across four public datasets.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Builds on 60 of its 73 references that are in the map. First posted on arXiv 26 Sep 2025.

Builds on

60 of 73 references.

References · 60

  1. 2017Attention Is All You Needcited by 60,361
  2. 2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,974
  3. 2023Visual Instruction Tuningcited by 7,869
  4. 2021Masked Autoencoders Are Scalable Vision Learnerscited by 7,403
  5. 2019EfficientNet: Rethinking Model Scaling for Convolutional Neural Networkscited by 6,442
  6. 2016Categorical Reparameterization with Gumbel-Softmaxcited by 2,739
  7. 2017CARLA: An Open Urban Driving Simulatorcited by 2,449
  8. 2013Estimating or Propagating Gradients Through Stochastic Neurons for Conditional Computationcited by 2,079
  9. 2022RT-1: Robotics Transformer for Real-World Control at Scalecited by 2,061
  10. 2017Deformable Convolutional Networkscited by 1,488
  11. 2018BDD100K: A Diverse Driving Dataset for Heterogeneous Multitask Learningcited by 1,288
  12. 2019Habitat: A Platform for Embodied AI Researchcited by 1,191
38 more
  1. 2021Tokens-to-Token ViT: Training Vision Transformers from Scratch on ImageNetcited by 1,101
  2. 2023OpenFlamingo: An Open-Source Framework for Training Large Autoregressive Vision-Language Modelscited by 565
  3. 2017End-to-end Driving via Conditional Imitation Learningcited by 485
  4. 2022LM-Nav: Robotic Navigation with Large Pre-Trained Models of Language, Vision, and Actioncited by 439
  5. 2021Dynamic Neural Networks: A Surveycited by 408
  6. 2016Target-driven Visual Navigation in Indoor Scenes using Deep Reinforcement Learningcited by 366
  7. 2020DeeBERT: Dynamic Early Exiting for Accelerating BERT Inferencecited by 332
  8. 2016Dynamic Filter Networkscited by 319
  9. 2022Confident Adaptive Language Modelingcited by 310
  10. 2024LayerSkip: Enabling Early Exit Inference and Self-Speculative Decodingcited by 307
  11. 2023NavGPT: Explicit Reasoning in Vision-and-Language Navigation with Large Language Modelscited by 280
  12. 2024Navigation World Modelscited by 272
  13. 2019RoboNet: Large-Scale Multi-Robot Learningcited by 270
  14. 2024Mixture-of-Depths: Dynamically allocating compute in transformer-based language modelscited by 258
  15. 2023NoMaD: Goal Masked Diffusion Policies for Navigation and Explorationcited by 248
  16. 2019Depth-Adaptive Transformercited by 236
  17. 2020FastBERT: a Self-distilling BERT with Adaptive Inference Timecited by 226
  18. 2023ViNT: A Foundation Model for Visual Navigationcited by 222
  19. 2018Semi-parametric Topological Memory for Navigationcited by 209
  20. 2022GNM: A General Navigation Model to Drive Any Robotcited by 171
  21. 2017Adaptive Neural Networks for Efficient Inferencecited by 163
  22. 2021Not All Images are Worth 16x16 Words: Dynamic Transformers for Efficient Image Recognitioncited by 144
  23. 2016Learning Modular Neural Network Policies for Multi-Task and Multi-Robot Transfercited by 127
  24. 2022Socially Compliant Navigation Dataset (SCAND): A Large-Scale Dataset of Demonstrations for Social Navigationcited by 127
  25. 2024DeeR-VLA: Dynamic Inference of Multimodal Large Language Models for Efficient Robot Executioncited by 117
  26. 2020Sim-to-Real Transfer for Vision-and-Language Navigationcited by 89
  27. 2023SkipDecode: Autoregressive Skip Decoding with Batching and Caching for Efficient LLM Inferencecited by 89
  28. 2020ViNG: Learning Open-World Navigation with Visual Goalscited by 80
  29. 2024Vision-and-Language Navigation Today and Tomorrow: A Survey in the Era of Foundation Modelscited by 79
  30. 2023Bridging Zero-shot Object Navigation and Foundation Models through Pixel-Guided Navigation Skillcited by 77
  31. 2017PRM-RL: Long-range Robotic Navigation Tasks by Combining Reinforcement Learning and Sampling-based Planningcited by 69
  32. 2023EE-LLM: Large-Scale Training and Inference of Early-Exit Large Language Models with 3D Parallelismcited by 69
  33. 2023SACSoN: Scalable Autonomous Control for Social Navigationcited by 61
  34. 2021FrameExit: Conditional Early Exiting for Efficient Video Recognitioncited by 59
  35. 2022ViKiNG: Vision-Based Kilometer-Scale Navigation with Geographic Hintscited by 58
  36. 2019Deep Visual MPC-Policy Learning for Navigationcited by 56
  37. 2013Low-Rank Approximations for Conditional Feedforward Computation in Deep Neural Networkscited by 44
  38. 2022You Need Multiple Exiting: Dynamic Early Exiting for Accelerating Unified Vision Language Modelcited by 43

50 most cited of 60 All 60 (JSON)

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026GemNav: Discrete-Token Visual Robot Navigation using a Multimodal Large Language Model
  2. 2025DyNaVLM: Zero-Shot Vision-Language Navigation System with Dynamic Viewpoints and Self-Refining Graph Memory
  3. 2026StepNav: Structured Trajectory Priors for Efficient and Multimodal Visual Navigation
  4. 2025ImagineNav++: Prompting Vision-Language Models as Embodied Navigator through Scene Imagination
  5. 2025FiLM-Nav: Efficient and Generalizable Navigation via VLM Fine-tuning
  6. 2025Dynam3D: Dynamic Layered 3D Tokens Empower VLM for Vision-and-Language Navigation
  7. 2026ImagiNav: Scalable Embodied Navigation via Generative Visual Prediction and Inverse Dynamics
  8. 2025HyPerNav: Hybrid Perception for Object-Oriented Navigation in Unknown Environment