Unsupervised Neural Sensor Models for Synthetic LiDAR Data Augmentation

TL;DR

This paper introduces two unsupervised neural sensor models using CycleGANs and Neural Style Transfer to generate realistic LiDAR data for augmentation, improving object detection performance in autonomous driving scenarios.

Contribution

The work presents novel unsupervised neural sensor models for LiDAR data augmentation, enabling realistic simulation without paired data, and provides a framework for evaluation using object detection tasks.

Findings

6% mAP improvement in object detection with augmented data

Neural sensor models produce more realistic LiDAR data

Framework enables effective evaluation of data augmentation methods

Abstract

Data scarcity is a bottleneck to machine learning-based perception modules, usually tackled by augmenting real data with synthetic data from simulators. Realistic models of the vehicle perception sensors are hard to formulate in closed form, and at the same time, they require the existence of paired data to be learned. In this work, we propose two unsupervised neural sensor models based on unpaired domain translations with CycleGANs and Neural Style Transfer techniques. We employ CARLA as the simulation environment to obtain simulated LiDAR point clouds, together with their annotations for data augmentation, and we use KITTI dataset as the real LiDAR dataset from which we learn the realistic sensor model mapping. Moreover, we provide a framework for data augmentation and evaluation of the developed sensor models, through extrinsic object detection task evaluation using YOLO network adapted to provide oriented bounding boxes for LiDAR Bird-eye-View projected point clouds. Evaluation is performed on unseen real LiDAR frames from KITTI dataset, with different amounts of simulated data augmentation using the two proposed approaches, showing improvement of 6% mAP for the object detection task, in favor of the augmenting LiDAR point clouds adapted with the proposed neural sensor models over the raw simulated LiDAR.