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

arXiv cs.CV / 3/31/2026

📰 NewsSignals & Early TrendsIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper introduces MotiMem, a motion-aware approximate memory interface designed to reduce data-movement energy in neural perception systems for battery-constrained autonomous vehicles.
It leverages temporal coherence using lightweight 2D motion propagation to identify Regions of Interest (RoI) dynamically, aiming to avoid unnecessary sensor data movement.
A hybrid sparsity-aware coding approach uses adaptive inversion and truncation to create bit-level sparsity, further lowering memory-interface dynamic energy.
Across nuScenes, Waymo, and KITTI using 16 detection models, MotiMem cuts memory-interface dynamic energy by about 43% while preserving roughly 93% of object detection accuracy.
The results claim a new, improved energy–accuracy Pareto frontier compared with standard, semantically blind codecs like JPEG and WebP.

Abstract

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.