Presentation: $200 Self-Driving Cars With RasPi and Tensorflow

Track: Hardware Frontiers: Changes Affecting Software Developers Today

Location: Pacific LMNO

Day of week:

Slides: Download Slides

Level: Beginner

Persona: Architect, CTO/CIO/Leadership, Data Engineering, Data Scientist, Developer, Developer, .NET, Developer, JVM, DevOps Engineer, Front-end Developer, General Software, ML Engineer, Mobile Developer, Security Professional, UX Designer

What You’ll Learn

  • How to use video and user inputs to clone human behavior.
  • That you want to go and build your own autonomous car!

Abstract

We (Will and Adam) will start by actually building and driving the $200 open source self driving Donkey Car. You’ll learn about the hardware components and software(python) that let it drive, capture data, and create autopilots.

Next, we’ll show you the autopilots that have been winning recent DIY Robocar races. This will give you an intuition about the constraints of self driving on cheap hardware and how to leverage cloud services to overcome them.

Lastly we’ll talk about where this project is going and how you and your kids, can help us get to the self driving future faster.

Buy car parts (see donkeycar.com for a list) before the conference, we’ll give you the 3D printed frame.

Question: 

What is your motivation for this talk?

Answer: 

William: I really want to make it easy for people to contribute to self-driving technology, because I want to never have to drive again.

Adam: It’s a fun project, it gives you intuition and introduction to machine learning, It’s a positive environment and a fun thing.

Question: 

Who should come to your talk?

Answer: 

Adam: Anyone who is interested in machine learning, people who are doing machine learning but don’t have intuition of the cause and effect of model building, Python programmers, people who like cars, people who have done RC (remote control) cars in the past and are ready for the next level.

William: Anyone who wants to try teaching machines rather than programming them.

Question: 

What can people come take away from this talk?

Answer: 

William: One thing would be how do you use a video and user input stream to train a neural network to “clone” a user's behavior.

Adam: I think everyone is going to want to come out of this wanting to build a car. It’s a cool project. People can take the off-the-shelf design and get it going in a few hours then improve it make it better.

Question: 

What keeps you up at night?

Answer: 

Will: One package in the car software will change that causes a bug that will take lots of time to find and fix. The things I enjoy staying up late thinking about about are, how do we a car that can be trained by someone that doesn’t know how to code. Right now our autopilots drive a bit like a disobedient 3 year old would. Now all we need is a little discipline.

Speaker: William Roscoe

Lead Engineer @Ceres Imaging

Will is a Lead Engineer at Ceres Imaging where he writes python to scale flight planning and image processing. In his spare time Will works toward replacing trains (cough BART cough) with platooning self driving busses. To this end he co-founded the Donkey Car project, to enable DIY hobbyists to accelerate self driving technology. You can find him at the monthly DIY Robocar races in Oakland.

Find William Roscoe at

Speaker: Adam Conway

Product Management and Marketing @Datacoral

Adam runs Product Management and Marketing at Datacoral by day, but by night makes robots, 3D Printers, CNC and satellites. Adam helps organize the DIYRobocars Races in Oakland with Chris Anderson and Will Roscoe. Previously Adam helped found Aerohive a now publicly traded WiFi company.

Find Adam Conway at

Similar Talks

Evolution of Edge @Netflix

Qcon

Engineering Leader @Netflix

Vasily Vlasov

Mistakes and Discoveries While Cultivating Ownership

Qcon

Engineering Manager @Netflix in Cloud Infrastructure

Aaron Blohowiak

Monitoring and Tracing @Netflix Streaming Data Infrastructure

Qcon

Architect & Engineer in Real Time Data Infrastructure Team @Netflix

Allen Wang

Future of Data Engineering

Qcon

Distinguished Engineer @WePay

Chris Riccomini

Coding without Complexity

Qcon

CEO/Cofounder @darklang

Ellen Chisa

Holistic EdTech & Diversity

Qcon

Holistic Tech Coach @unlockacademy

Antoine Patton