Explore the challenges and solutions in designing efficient Augmented Reality (AR) systems in this conference talk from Strange Loop. Dive into the parallels between AR software design and distributed systems, examining how to balance complex computer vision algorithms with resource limitations on wearables and mobile devices. Learn about the importance of concurrent systems, real-time interactions, and caching mechanisms in AR development. Discover strategies for speeding up networking and computation, including the use of unreliable messages, actor models, and embedding SLAM (Simultaneous Localization and Mapping) techniques. Gain insights into the interdisciplinary nature of AR, the fundamental human limits in perception, and how to overcome hardware constraints through innovative software engineering approaches.
Overview
Syllabus
Intro
Problems in Computer Science
Why is AR so hard
AR is an interdisciplinary field
Understanding the world
Running in realtime
Point cloud
Augmented Reality
Developer Experience
Putting it all together
Infinity Heart Problems
RealTime Computation
Beyond Moores Law
Holmes Law
Fundamental Human Limit
Systems Design
Systems Design Principle for Distributed System
Speeding Up Networking
Unreliable Messages
Ideal Solution
Speeding Up Computation
Slam Computation Pipeline
Embedding Slam
Actor Models
Restructure
Credits
Thanks
Were also hiring
Taught by
Strange Loop Conference
