mlsanityData to 14 Oct 2026

UniLCD: Unified Local-Cloud Decision-Making via Reinforcement Learning

arXiv 2409.11403 ↗First posted 17 Sep 2024cs.ROcited by 2

On the map: Deep reinforcement learning methods → RL for energy, networks and operations

Abstract (arXiv)

Embodied vision-based real-world systems, such as mobile robots, require a careful balance between energy consumption, compute latency, and safety constraints to optimize operation across dynamic tasks and contexts. As local computation tends to be restricted, offloading the computation, ie, to a remote server, can save local resources while providing access to high-quality predictions from powerful and large models. However, the resulting communication and latency overhead has led to limited usability of cloud models in dynamic, safety-critical, real-time settings. To effectively address this trade-off, we introduce UniLCD, a novel hybrid inference framework for enabling flexible local-cloud collaboration. By efficiently optimizing a flexible routing module via reinforcement learning and a suitable multi-task objective, UniLCD is specifically designed to support the multiple constraints of safety-critical end-to-end mobile systems. We validate the proposed approach using a challenging, crowded navigation task requiring frequent and timely switching between local and cloud operations. UniLCD demonstrates improved overall performance and efficiency, by over 35% compared to state-of-the-art baselines based on various split computing and early exit strategies.

Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Placed in RL for energy, networks and operations, one of 15 directions in Deep reinforcement learning methods. Builds on 55 of its 124 references that are in the map. First posted on arXiv 17 Sep 2024.

Builds on

55 of 124 references.

References · 55

  1. 2023GPT-4 Technical Reportcited by 18,396
  2. 2017Proximal Policy Optimization Algorithmscited by 15,215
  3. 2023LLaMA: Open and Efficient Foundation Language Modelscited by 14,132
  4. 2023Visual Instruction Tuningcited by 7,869
  5. 2018MobileNetV2: Inverted Residuals and Linear Bottleneckscited by 4,795
  6. 2023RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Controlcited by 3,080
  7. 2017CARLA: An Open Urban Driving Simulatorcited by 2,449
  8. 2023Video-LLaVA: Learning United Visual Representation by Alignment Before Projectioncited by 1,402
  9. 2020Designing Network Design Spacescited by 960
  10. 2021MobileViT: Light-weight, General-purpose, and Mobile-friendly Vision Transformercited by 642
  11. 2017End-to-end Driving via Conditional Imitation Learningcited by 485
  12. 2018HAQ: Hardware-Aware Automated Quantization with Mixed Precisioncited by 443
38 more
  1. 2022LM-Nav: Robotic Navigation with Large Pre-Trained Models of Language, Vision, and Actioncited by 439
  2. 2021Dynamic Neural Networks: A Surveycited by 408
  3. 2022Learning robust perceptive locomotion for quadrupedal robots in the wildcited by 382
  4. 2016Target-driven Visual Navigation in Indoor Scenes using Deep Reinforcement Learningcited by 366
  5. 2019Exploring the Limitations of Behavior Cloning for Autonomous Drivingcited by 346
  6. 2018Model compression via distillation and quantizationcited by 289
  7. 2024Reinforcement learningcited by 247
  8. 2019Model Pruning Enables Efficient Federated Learning on Edge Devicescited by 222
  9. 2018Crowd-Robot Interaction: Crowd-aware Robot Navigation with Attention-based Deep Reinforcement Learningcited by 175
  10. 2023Robot Parkour Learningcited by 149
  11. 2016Designing Energy-Efficient Convolutional Neural Networks using Energy-Aware Pruningcited by 134
  12. 2020SPINN: Synergistic Progressive Inference of Neural Networks over Device and Cloudcited by 112
  13. 2021Core Challenges of Social Robot Navigation: A Surveycited by 106
  14. 2021Reward (Mis)design for Autonomous Drivingcited by 104
  15. 2019Urban Driving with Conditional Imitation Learningcited by 86
  16. 2021Minimizing Energy Consumption Leads to the Emergence of Gaits in Legged Robotscited by 76
  17. 2021Adaptive Inference through Early-Exit Networks: Design, Challenges and Directionscited by 75
  18. 2017RegNet: Multimodal Sensor Registration Using Deep Neural Networkscited by 69
  19. 2024Learning from Teaching Regularization: Generalizable Correlations Should be Easy to Imitatecited by 47
  20. 2022You Need Multiple Exiting: Dynamic Early Exiting for Accelerating Unified Vision Language Modelcited by 43
  21. 2022SPDY: Accurate Pruning with Speedup Guaranteescited by 36
  22. 2024Fast yet Safe: Early-Exiting with Risk Controlcited by 30
  23. 2024Real-Time Pill Identification for the Visually Impaired Using Deep Learningcited by 21
  24. 2023Coaching a Teachable Studentcited by 19
  25. 2024Enhanced Detection Classification via Clustering SVM for Various Robot Collaboration Taskcited by 19
  26. 2022SelfD: Self-Learning Large-Scale Driving Policies From the Webcited by 18
  27. 2023XVO: Generalized Visual Odometry via Cross-Modal Self-Trainingcited by 18
  28. 2021AppealNet: An Efficient and Highly-Accurate Edge/Cloud Collaborative Architecture for DNN Inferencecited by 17
  29. 2022Cloud-Device Collaborative Adaptation to Continual Changing Environments in the Real-worldcited by 17
  30. 2023Window-Based Early-Exit Cascades for Uncertainty Estimation: When Deep Ensembles are More Efficient than Single Modelscited by 16
  31. 2020Cluster Pruning: An Efficient Filter Pruning Method for Edge AI Vision Applicationscited by 11
  32. 2020Fast Adaptive Task Offloading in Edge Computing based on Meta Reinforcement Learningcited by 11
  33. 2021Neural Process for Black-Box Model Optimization Under Bayesian Frameworkcited by 11
  34. 2023Learning to Drive Anywherecited by 10
  35. 2020NaviGAN: A Generative Approach for Socially Compliant Navigationcited by 7
  36. 2024Real-time system optimal traffic routing under uncertainties -- Can physics models boost reinforcement learning?cited by 7
  37. 2023RED: A Systematic Real-Time Scheduling Approach for Robotic Environmental Dynamicscited by 6
  38. 2019Cost-Driven Offloading for DNN-based Applications over Cloud, Edge and End Devicescited by 5

50 most cited of 55 All 55 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

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

2 papers.

  1. 2025Rethinking Inference Placement for Deep Learning across Edge and Cloud Platforms: A Multi-Objective Optimization Perspective and Future Directionscited by 0
  2. 2026Time-Aware Assistive Navigationcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Bridging On-Device and Cloud LLMs for Collaborative Reasoning: A Unified Methodology for Local Routing and Post-Training
  2. 2026Hera: Learning Long-Horizon Coordination for Device-Cloud Collaborative LLM Agents
  3. 2026Latency-Tolerant Cloud-Edge Collaborative Vision-Language-Action Models via Emergent Representational Specialization
  4. 2026Sibyl: An Efficient Small-large Model Collaboration Framework for Long-horizon Tasks
  5. 2025Cooperative Task Offloading through Asynchronous Deep Reinforcement Learning in Mobile Edge Computing for Future Networks
  6. 2026On-Device Deep Reinforcement Learning for Decentralized Task Offloading Performance trade-offs in the training process
  7. 2025Splitwise: Collaborative Edge-Cloud Inference for LLMs via Lyapunov-Assisted DRL
  8. 2026Speculative Policy Orchestration: A Latency-Resilient Framework for Cloud-Robotic Manipulation