Before we begin, you may bring a cup of coffee that can help us through the paper.

Overview

In this paper, the authors propose a neural network that preicts the vectorized map elements in BEV.

HD map construction is a crucial problem for autonomous driving. In this paper, the authors introduce an online map learning method that provide semantic and geometry priors to self-driving vehicles.

Preliminary

If you are not familiar with the followings, check out the blog posts.

• HD map
• Convolutional Neural Network
• PointPillar
• Instance embedding

Keywords

CNN, HD map, Lane detection, LiDAR, Instance Embedding, PointPillar, PointNet, DBSCAN, NMS

More to come…