Evaluation and Optimization of Rendering Techniques for Autonomous Driving Simulation

TL;DR

This paper explores the use of offline industrial rendering to improve scene quality in autonomous driving simulations, aiming to enhance training data for computer vision tasks.

Contribution

It introduces a pipeline combining game engine rendering, IQA validation, and offline processing to optimize scene quality for autonomous driving simulation.

Findings

Offline rendering improves scene quality for training

IQA algorithms effectively validate scene improvements

Pipeline enhances simulation realism for autonomous driving

Abstract

In order to meet the demand for higher scene rendering quality from some autonomous driving teams (such as those focused on CV), we have decided to use an offline simulation industrial rendering framework instead of real-time rendering in our autonomous driving simulator. Our plan is to generate lower-quality scenes using a game engine, extract them, and then use an IQA algorithm to validate the improvement in scene quality achieved through offline rendering. The improved scenes will then be used for training.