UPAQ: A Framework for Real-Time and Energy-Efficient 3D Object Detection in Autonomous Vehicles

TL;DR

UPAQ is a framework that significantly improves the efficiency of 3D object detection in autonomous vehicles by applying pattern pruning and quantization, enabling faster, more energy-efficient processing on embedded platforms.

Contribution

The paper introduces UPAQ, a novel framework that combines semi-structured pattern pruning and quantization to enhance 3D object detection efficiency on resource-limited AV hardware.

Findings

Achieves up to 5.62x model compression

Boosts inference speed by up to 1.97x

Reduces energy consumption by up to 2.07x

Abstract

To enhance perception in autonomous vehicles (AVs), recent efforts are concentrating on 3D object detectors, which deliver more comprehensive predictions than traditional 2D object detectors, at the cost of increased memory footprint and computational resource usage. We present a novel framework called UPAQ, which leverages semi-structured pattern pruning and quantization to improve the efficiency of LiDAR point-cloud and camera-based 3D object detectors on resource-constrained embedded AV platforms. Experimental results on the Jetson Orin Nano embedded platform indicate that UPAQ achieves up to 5.62x and 5.13x model compression rates, up to 1.97x and 1.86x boost in inference speed, and up to 2.07x and 1.87x reduction in energy consumption compared to state-of-the-art model compression frameworks, on the Pointpillar and SMOKE models respectively.