Learning to Simulate Realistic LiDARs

TL;DR

This paper presents a data-driven pipeline that learns to simulate realistic LiDAR sensor effects from real data, improving the fidelity of simulated LiDAR data for autonomous systems.

Contribution

It introduces a model that maps RGB images to LiDAR features, enhancing simulated point clouds to better match real sensor data.

Findings

Model captures realistic LiDAR effects like dropped points and high intensity returns.

Enhancing simulated data improves downstream vehicle segmentation performance.

The approach applies to different LiDAR sensors, demonstrating versatility.

Abstract

Simulating realistic sensors is a challenging part in data generation for autonomous systems, often involving carefully handcrafted sensor design, scene properties, and physics modeling. To alleviate this, we introduce a pipeline for data-driven simulation of a realistic LiDAR sensor. We propose a model that learns a mapping between RGB images and corresponding LiDAR features such as raydrop or per-point intensities directly from real datasets. We show that our model can learn to encode realistic effects such as dropped points on transparent surfaces or high intensity returns on reflective materials. When applied to naively raycasted point clouds provided by off-the-shelf simulator software, our model enhances the data by predicting intensities and removing points based on the scene's appearance to match a real LiDAR sensor. We use our technique to learn models of two distinct LiDAR sensors and use them to improve simulated LiDAR data accordingly. Through a sample task of vehicle segmentation, we show that enhancing simulated point clouds with our technique improves downstream task performance.