View this email in your browser


It’s been a fruitful and busy start to 2019 for the Uber Visualization team. In this sixth edition of our newsletter, we highlight what we’ve been up to in the New Year, from launching a new open source autonomous visualization system to releasing new cool updates for, and H3 frameworks. We also discuss what’s coming up next in our open source portfolio, as well as invite you to join Uber, Mapbox and Cruise Automation at our next Visualization Night for an evening of presentations and conversation about autonomous vehicles visualization systems.

Latest Updates

Autonomous Visualization System (AVS) Launch
We’ve officially launched AVS, our web-based 3D visualization toolkit for building applications from self-driving and robotics data. This announcement represents the culmination of 2 years of collaboration between the Advanced Technologies Group (ATG) and Visualization teams at Uber. Read more in our Featured Story section below, as well as our recent article on the Uber Engineering Blog’s New Sharing Capabilities
New capabilities have been unlocked, enabling users to share their map data analysis with others quickly.

Sharing Maps Powered by Dropbox
We’re introducing Dropbox integration to the website to allow users to quickly create shareable links in their maps. The map data and configuration are stored on users’ Dropbox accounts, which allows users to manage their visualization data. Dropbox Integration

Exporting as Single HTML File
Thanks to the library UMD module build, we launched a new feature which allows users to export current map into a single HTML file. The newly created file will have all the necessary JavaScript libraries and code to load and display users’ data with no programming knowledge required. This allows users to embed the map in any application, or simply save their maps locally for viewing later.

d9f2f9fe-b0cd-461b-a272-f0833ad5d638.gif Plugin for Tableau
The team is also working on allowing to be consumed by popular analytics tools such as Jupyter Notebook and Tableau. With these new functionalities, users will be able to import data from DataFrame, setting up map configurations and sharing their maps without leaving Jupyter Notebook and Tableau. To design the extension for Tableau, we’re working with Tableau Zen master Chris Demartini and Allan Walker from Mapbox. These features are in the final development stage and will be released in April 2019. Be on the lookout for an upcoming meetup at Mapbox during which we'll highlight these features! Extension for Tableau for Jupyter Notebook Python Package

Manifold Updates
Manifold is a model-agnostic visual debugging tool for machine learning developed by the team. It makes use of in-browser computation capacity of TensorFlow.js, and allows interactive dataset dissecting based on ML model prediction outcomes, helping ML practitioners to identify problematic data slices or feature values that contribute to a sub-ideal model performance, and thereby facilitate their model iteration process. Currently, the team is working on an external release of Manifold, in response to the numerous requests we received from the ML community since we published its method on Uber Engineering Blog. 6.4 Release
We released version 6.4 of and, the last milestone planned for the 6.x series visualization frameworks. The release includes enhancements to many core layers. The Text Layer now supports better control of the font rendering quality, including font weight and raster size, as well as crisper font rendering at all zoom levels using the Signed Distance Field font rendering technique. The Scatterplot Layer now supports drawing both strokes and fills, and outline width can be controlled per-instance. The Contour Layer now supports rendering IsoBands in addition to IsoLines. We added support for flattened binary data to both Path and Polygon Layers. In addition to layer enhancements, this release included bug fixes and feature improvements in’s Orthographic View for InfoVis use cases. See the full list of release highlights in the What’s New and Upgrade Guide for more information.

H3 is an open source library for easily dividing the globe into a hierarchical grid of hexagons. We released H3 version 3.4.0, including support for drawing grid-based lines and getting the set of base cells (allowing efficient iteration of all indexes).


Featured Story: Better Understanding the Self-Driving World with Autonomous Visualization System (AVS)

We recently open sourced our Autonomous Visualization System (AVS), a toolkit for describing and visualizing autonomous vehicle perception, motion, and planning data. 

AVS includes two main components: XVIZ and XVIZ is the data layer for AVS and includes a formal but flexible protocol specification for representing data generated from autonomous systems. XVIZ is optimized for data primitives like LiDAR point clouds, camera images, object bounds, trajectories, vehicle speed over time, and predicted plans. XVIZ provides a stream-oriented view of a scene changing over time and a declarative user interface display system.

0c5aa31f-c683-4885-abc1-06b3692f0084.gif is the web-based component toolkit built on top of for creating visualization applications for autonomous data in the XVIZ format. consists of react components for processing, rendering, interacting with, and controlling data in an autonomous visualization application. It includes components for 3D viewports, camera and playback controls, as well as charts and tables for plotting and visualizing data.

The Uber Advanced Technology Group and Visualization teams partnered to build AVS after recognizing the challenges around mastering complex computer graphics and data visualization techniques. Through this collaboration, our teams designed a suite of effective tooling solutions to easily and intuitively render autonomous data, as well as a new open standard for autonomous visualizations. By sharing AVS with the broader autonomous community, we hope that others in the autonomy and robotics industries will unlock greater advancements with these tools to deliver safer, more efficient transportation solutions for everyone.

Interested working together? Reach out to us at with feedback and questions!

Maps in the Wild

be410f4c-f241-4099-9c54-213aa46d37fb.pngHaiti, admin level 4, with population and clinics and hospitals

UNICEF’s Magic Box has adopted internally to visualize poverty, population vulnerabilities, and the spread of disease around the world. Since then, they have been gradually connecting datasets like administrative boundaries, populations, and health facilities from You can check out an interactive and public version of this data here and read more by following Mike Fabrikant's articles on Magic Box in his blog.


Impressed by this piece of digital artwork? It’s a 3D visualization of the Statue of Liberty, rendered in by Allan Walker from Mapbox!


Here is yet another cool piece of art from Allan -- a visualization of Hurricane Harvey, with radar data from NOAA NCEI.

Share your data visualizations with us! Tweet @ubereng with #keplergl or #deckgl! 

Want to share feedback, discuss possible collaborations, or ask a question about our frameworks? Contact us at!

Things to Read

The New York Times published this list of their best visual stories and graphics of the year 2018. 

How Every Member Got to Congress, a very elaborate and innovative flow chart, will definitely figure on The Times’ 2019 list. We learn that over a third of Congress have been lawyers and nearly 20 percent have served in the U.S. Military. 

In The rising Western skyline, The Washington Post explores the striking transformations in Western cityscapes such as Denver and Seattle with striking graphics. 

RJ Andrews published his highly-anticipated treatise on visualization, Info We Trust: How to Inspire the World with Data. The book is a striking book, with over 300 hand-made illustrations, and an invitation to reflect on the nature and power of visualization.

External Engagement

We were invited to share our experience building Manifold, Uber’s visual debugging tool for machine learning, at the TensorFlow Developer Summit on March 7. During her presentation, Uber Data Visualization’s Lezhi Li talked about how TF.js helps the development of visual analytics tools like Manifold. Check out a video recording of her presentation to learn more.


Uber’s Travis Gorkin delivered a keynote at the Linux Foundation Open Source Leadership Summit on how Uber’s data visualization products are being used to explore self-driving, geospatial, and business data in the context of urban mobility.


And it’s time for another Uber Vis Night! Join us on March 27 from 6 - 8 p.m. at Uber’s SF office on 555 Market for a night of talks delivered by Uber, Mapbox, and Cruise Automation about visualization and mapping for the autonomous world. RSVP here as spots are limited!


Thanks for reading! 

The Uber Visualization Team

Copyright © Uber Technologies Inc., All rights reserved.
You are receiving this email because you are subscribed to Uber visualization news. 

Our mailing address is:
Uber Technologies
1455 Market St
San Francisco, CA 94103-1331

Want to change how you receive these emails?
You can update your preferences or unsubscribe from this list.