Offline reconstruction of missing vehicle trajectory data from 3D LIDAR

TL;DR

This paper introduces a novel method for reconstructing missing vehicle trajectory data from 3D LIDAR point clouds using calibrated microscopic traffic flow models, improving data completeness for autonomous vehicle applications.

Contribution

The paper proposes a new approach combining traffic flow models with LIDAR data to recover missing vehicle trajectories, calibrated and tested with real datasets.

Findings

Gipps' model provided the best trajectory recovery results.

Short gaps (<5s) are effectively recovered with linear regression.

Longer gaps are accurately reconstructed using the proposed traffic flow model method.

Abstract

LIDAR has become an important part of many autonomous vehicles with its advantages on distance measurement and obstacle detection. LIDAR produces point clouds which have important information about surrounding environment. In this paper, we collected trajectory data on a two lane urban road using a Velodyne VLP-16 Lidar. Due to dynamic nature of data collection and limited range of the sensor, some of these trajectories have missing points or gaps. In this paper, we propose a novel method for recovery of missing vehicle trajectory data points using microscopic traffic flow models. While short gaps (less than 5 seconds) can be recovered with simple linear regression, and longer gaps are recovered with the proposed method that makes use of car following models calibrated by assigning weights to known points based on proximity to the gaps. Newell's, Pipes, IDM and Gipps' car following models are calibrated and tested with the ground truth trajectory data from LIDAR and NGSIM I-80 dataset. Gipps' calibrated model yielded the best result.