Self-driving autonomous cars are coming to change the society we live in – probably revolutionise the way we travel in 21st century. The autonomous vehicles are being widely developed and tested and probably will become more dominant by 2040. Big companies such as Tesla, Google, Audi, Uber and Nissan, among others, are pouring millions of dollars to get the cars on the road. The rush to develop self-driving are also fueling lucrative deals for startups that specialise in AI and robotics.
But are you wondering how to get the resources to get started with the self-driving autonomous cars? Don’t worry, we’ve got you covered. Here we list down few free resources that might help you get started:
This online class is an introduction to the practice of deep learning through applied theme of building a self-driving car. It is designed for beginners and for those who are new to machine learning. The material for the course is free and open to everyone. The research group includes software engineers and research scientists and is backed by companies like Google, Amazon Alexa, Nvidia, among others.
If you are interested enroll for the classes, here are few steps involved to get started:
- Create an account on the site to stay up-to-date.
- Join slack channel(deep-mit.slack.com)
- Watch the lectures and guest talks
- Interact with the instructor Lex Fridman on Twitter, LinkedIn, Instagram, Facebook or just subscribe on YouTube.
- And if you have any question, you can check out their FAQ Google Doc.
This lengthy survey paper provides a state-of-the-art survey on Computer Vision for Autonomous Vehicles. This survey paper covers several specific topics, which includes recognition, reconstruction, motion estimation, tracking, scene understanding and end-to end learning. The paper also discuss open problems and research challenges. It also provides an interactive platform which allows to navigate topics and methods, and provides additional information and project links for each paper. The paper is available online in PDF format. The survey claims that it bridges the gap between robotics, intelligent vehicles, photogrammetry and computer vision by providing an thorough overview and comparison which includes works from all fields. If you want to get up to speed in the field quickly and you want everything in one spot then this survey paper might prove to be a great help.
Kevin Huges in his blog explains how he was able to train an AI to drive a virtual vehicle using the same technique Google uses for their self-driving cars. He also says that with 20 minutes of training data, his AI was able to drive the majority of the simplest course. Kevin Huges used Tensorflow to train an agent that can play MarioKart 64. Kevin Huges was a developer at Shopify. He also worked on Artificial Intelligence at Queen’s University.
Python plays Grand Theft Auto V:
In a video series Harrison Knisley shows how to create a self-driving car using Python to play Grand Theft Auto V. He says GTA V is a open, sand-box type of environment that can be controlled and it makes a great development area. You can also check out his code here.
The open source code by Udacity is written by hundreds of students from across the globe. In here, you will be able to contribute code to a real self-driving car that will run on the road. Like any open source project, this code base will require a certain amount of thoughtfulness. You can also form a team and take the Udacity open source code challenges by joining Udacity’s slack community and then join the #challenges channel in Udacity. Each Udacity challenge will contain prizes for the effective contribution, you will also get an exciting learning experience.
The online teaching startup, Udacity offers a four-month ‘nanodegree’ course, which teaches you the skills and techniques used by self-driving car teams at the most advanced technology companies in the world. The only prerequisites for students interested in the intro class are some programming experience like C++, Python and algebra. Though enrolling for the course is not free and will cost some money.
Try deep learning using MATLAB