Track: Hardware Frontiers: Changes Affecting Software Developers Today

You Can and Should Make Hardware

Every industry will experience increasing pressure to collect data from, and maniuplate, the physical world: e-commerce, food, advertising, sales, manufacturing, agriculture, healthcare, labor management. This will require us to understand sensors and autonomy. We hear that making hardware is hard, but what if hardware isn’t hard? What if lean enterprise ideas worked just as well for hardware as they do for software, maybe even better?

When we accept this idea, we have the freedom to embrace failure and to release imperfect versions. This gives us the freedom to iterate like crazy - not only do we learn a ton about the problem space, we also grow our skills along the way.

Any device that will survive first contact with the market needs about 15 testable iterations before we really discover pmf, manufacturability and durability. So multiply the cycle time by 15 to get a good estimate of when we can be in the market. We really want to get this down to a week per releasable iteration, and there are ways to do this.

Taking Advantage of Advances in Mobile Hardware

As part of the engineering team at MZ we have been constantly updating the game engine that runs the highly successful Game of War: Fire Age, Mobile Strike, and Final Fantasy XV a New Empire mobile games. We have taken the engine from its 2D sprite-based roots and turned it into a full-fledged 3D engine with advanced rendering and animation features. We will present our unique solutions to the problems that arose while trying to create the best experience possible on new hardware without alienating customers with older, less-powerful devices.

Rethinking Deep Learning: Neural Compute Stick

The Movidius™ Neural Compute Stick (NCS) is a tiny fanless deep learning device that you can use to learn AI programming at the edge. NCS is powered by the same low power high-performance Movidius™ Vision Processing Unit (VPU) that can be found in millions of smart security cameras, gesture-controlled drones, industrial machine vision equipment, and more. The Movidius Neural Compute Stick enables rapid prototyping, validation and deployment of Deep Neural Network (DNN) inference applications at the edge. Its low-power VPU architecture enables an entirely new segment of AI applications that aren’t reliant on a connection to the cloud.

The NCS combined with Movidius™ Neural Compute SDK allows deep learning developers to profile, tune, and deploy Convolutional Neural Network (CNN) on low-power applications that require real-time inferencing. This talk explores this cutting-edge device an offers a glimpse into what the future holds for software developers diving into the space of deep learning.

Rethinking Applications for the NVM Era

Storage looks and feels like a block device hidden behind layers of abstractions - system calls, pagecaches, block device drivers and things we don't want to think about. This might change in the near future with the introduction of CPU accessible, byte addressable, persistent memory. How would your logging service, database or filesystem look in this world? Actually, would you need any of those or could a persistent C++ STL container be your storage service of choice?

TensorFlow: Pushing the ML Boundaries

In this talk, you will see how Google uses Machine Learning to address problems that were not solvable just a year ago. We will look at some of the latest models we’ve developed and where the opportunities & challenges are, for example how we can automize the machine learning process. We will also describe latest advancements in TensorFlow to deal with model complexity and how developers can get start building complex models today. Finally, we will talk about the need for compute power & scale, using hardware accelerators like the TPU.

$200 Self-Driving Cars With RasPi and Tensorflow

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.