Model Recovery at the Edge under Resource Constraints for Physical AI

Bin Xu, Ayan Banerjee, Sandeep K. S. Gupta
12/1/2025
cs.AI

No official code found. Best implementation: xu842251462/ECAI2025 (0 stars, MIT, last commit 2025-09-09).

Abstract

Model Recovery (MR) enables safe, explainable decision making in mission-critical autonomous systems (MCAS) by learning governing dynamical equations, but its deployment on edge devices is hindered by the iterative nature of neural ordinary differential equations (NODEs), which are inefficient on FPGAs. Memory and energy consumption are the main concerns when applying MR on edge devices for real-time operation. We propose MERINDA, a novel FPGA-accelerated MR framework that replaces iterative solvers with a parallelizable neural architecture equivalent to NODEs. MERINDA achieves nearly 11x lower DRAM usage and 2.2x faster runtime compared to mobile GPUs. Experiments reveal an inverse relationship between memory and energy at fixed accuracy, highlighting MERINDA's suitability for resource-constrained, real-time MCAS.

View on arXivView PDF

Code Implementations(1)

xu842251462/ECAI2025Community66%
MIT

Model Recovery at the Edge under Resource Constraints for Physical AI

00C, C++Aug 21, 20251 years agoMIT

Similar Papers

Enabling Physical AI at the Edge: Hardware-Accelerated Recovery of System Dynamics

83% match2025

Embedded FPGA Acceleration of Brain-Like Neural Networks: Online Learning to Scalable Inference

83% match2025

Hardware-aware Neural Architecture Search of Early Exiting Networks on Edge Accelerators

82% match2025

Edge Intelligence with Spiking Neural Networks

82% match2025

HYDRA: Hybrid Data Multiplexing and Run-time Layer Configurable DNN Accelerator

82% match2024

Resistive Memory-based Neural Differential Equation Solver for Score-based Diffusion Model

82% match1 repo62024

Cite this paper

@article{xu2025model,
  title  = {Model Recovery at the Edge under Resource Constraints for Physical AI},
  author = {Bin Xu and Ayan Banerjee and Sandeep K. S. Gupta},
  year   = {2025},
  eprint = {2512.02283},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2512.02283v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2512.02283.svg)](https://codeofpaper.com/paper/2512.02283)

Discussion