New frontiers in data-driven autonomous driving
CVPR 2021 tutorial

[Site in construction]

Welcome!

This hands-on tutorial focused on the latest, data-driven approaches to prediction, planning, and simulation in self-driving, as well as their interplay with computer vision. You will learn basic concepts, state-of-the-art solutions and also build your own solutions. This tutorial is prepared by Lyft Level 5 team with the purpose of advancing state-of-the-art self-driving research.

Autonomy 2.0

Modern production self-driving systems used in the industry still rely excessively on hand-engineering, especially when it comes to planning and simulation. This is becoming a limiting factor in self-driving development. Autonomy 2.0 is a paradigm of using ML-first approaches for these components offering greater scalability, safety and comfort of self-driving cars.

Modern self-driving pipeline and the amount of ML employed in each component.
The goal of Autonomy 2.0 is to make the entire stack ML-first and data-driven.

Introduction

This hands-on tutorial focused on the latest, data-driven approaches to prediction, planning, and simulation in self-driving, as well as their interplay with computer vision. You will learn basic concepts, state-of-the-art solutions and also build your own solutions. This tutorial is prepared by Lyft Level 5 team with the purpose of advancing state-of-the-art self-driving research.

Autonomy 2.0

Modern production self-driving systems used in the industry still rely excessively on hand-engineering, especially when it comes to planning and simulation. This is becoming a limiting factor in self-driving development. Autonomy 2.0 is a paradigm of using ML-first approaches for these components offering greater scalability, safety and comfort of self-driving cars.

Datasets for training

To train and test the models you will need a large-scale dataset.

L5Kit: Software development kit

Part of this tutorial is ability to try the discussed topics and build their own ML prediction, planning and simulation modules for self-driving cars. This is done using L5Kit - a software development platform that can be found at l5kit.org.

ML Perception

Perception description

Large-scale mapping

Perception description

ML prediction

Prediction - ability to tell what will happen next is the first step of decision-making.

ML planning

Planning - ability to tell what the car should do

ML simulation

Simulation description

Deployment & road testing

Deployment description

Scaling data

Autonomy description

Benchmarks & competitions

Competition description

Organising team

Peter Ondruska

Hugo Grimmett

Christy Robertson

Ray Gao

Luca Del Pero

Vladimir Iglovikov

Yawei Ye

Qiangui Huang

Błażej Osiński

Long Chen

Luca Bergamini

Christian Perone

Oliver Scheel