MotiMem: Motion-Aware Approximate Memory for Energy-Efficient Neural Perception in Autonomous Vehicles

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Que, Haohua, Liu, Mingkai, Xie, Jiayue, Gao, Haojia, Sun, Jiajun, Xu, Hongyi, Yao, Handong, Qiao, Fei
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908918899277824
author Que, Haohua
Liu, Mingkai
Xie, Jiayue
Gao, Haojia
Sun, Jiajun
Xu, Hongyi
Yao, Handong
Qiao, Fei
author_facet Que, Haohua
Liu, Mingkai
Xie, Jiayue
Gao, Haojia
Sun, Jiajun
Xu, Hongyi
Yao, Handong
Qiao, Fei
contents High-resolution sensors are critical for robust autonomous perception but impose a severe memory wall on battery-constrained electric vehicles. In these systems, data movement energy often outweighs computation. Traditional image compression is ill-suited as it is semantically blind and optimizes for storage rather than bus switching activity. We propose MotiMem, a hardware-software co-designed interface. Exploiting temporal coherence,MotiMem uses lightweight 2D Motion Propagation to dynamically identify Regions of Interest (RoI). Complementing this, a Hybrid Sparsity-Aware Coding scheme leverages adaptive inversion and truncation to induce bitlevel sparsity. Extensive experiments across nuScenes, Waymo, and KITTI with 16 detection models demonstrate that MotiMem reduces memory-interface dynamic energy by approximately 43 percent while retaining approximately 93 percent of the object detection accuracy, establishing a new Pareto frontier significantly superior to standard codecs like JPEG and WebP.
format Preprint
id arxiv_https___arxiv_org_abs_2603_27108
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle MotiMem: Motion-Aware Approximate Memory for Energy-Efficient Neural Perception in Autonomous Vehicles
Que, Haohua
Liu, Mingkai
Xie, Jiayue
Gao, Haojia
Sun, Jiajun
Xu, Hongyi
Yao, Handong
Qiao, Fei
Computer Vision and Pattern Recognition
High-resolution sensors are critical for robust autonomous perception but impose a severe memory wall on battery-constrained electric vehicles. In these systems, data movement energy often outweighs computation. Traditional image compression is ill-suited as it is semantically blind and optimizes for storage rather than bus switching activity. We propose MotiMem, a hardware-software co-designed interface. Exploiting temporal coherence,MotiMem uses lightweight 2D Motion Propagation to dynamically identify Regions of Interest (RoI). Complementing this, a Hybrid Sparsity-Aware Coding scheme leverages adaptive inversion and truncation to induce bitlevel sparsity. Extensive experiments across nuScenes, Waymo, and KITTI with 16 detection models demonstrate that MotiMem reduces memory-interface dynamic energy by approximately 43 percent while retaining approximately 93 percent of the object detection accuracy, establishing a new Pareto frontier significantly superior to standard codecs like JPEG and WebP.
title MotiMem: Motion-Aware Approximate Memory for Energy-Efficient Neural Perception in Autonomous Vehicles
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2603.27108