NeRF-To-Real Tester: Neural Radiance Fields as Test Image Generators for Vision of Autonomous Systems

TL;DR

This paper introduces N2R-Tester, a tool that uses Neural Radiance Fields to generate realistic test images for autonomous systems, improving testing for vision components like vSLAM and object detection.

Contribution

It presents a novel method for creating diverse, realistic test data using Neural Radiance Fields and integrates it into a metamorphic testing framework for autonomous vehicle vision systems.

Findings

Effective generation of diverse test images for autonomous systems

Improved testing of vision components like vSLAM and object detection

Versatile application across different autonomous vehicle platforms

Abstract

Autonomous inspection of infrastructure on land and in water is a quickly growing market, with applications including surveying constructions, monitoring plants, and tracking environmental changes in on- and off-shore wind energy farms. For Autonomous Underwater Vehicles and Unmanned Aerial Vehicles overfitting of controllers to simulation conditions fundamentally leads to poor performance in the operation environment. There is a pressing need for more diverse and realistic test data that accurately represents the challenges faced by these systems. We address the challenge of generating perception test data for autonomous systems by leveraging Neural Radiance Fields to generate realistic and diverse test images, and integrating them into a metamorphic testing framework for vision components such as vSLAM and object detection. Our tool, N2R-Tester, allows training models of custom scenes and rendering test images from perturbed positions. An experimental evaluation of N2R-Tester on eight different vision components in AUVs and UAVs demonstrates the efficacy and versatility of the approach.