Get best of the week

Uber Open Source Unmanned Transport Data Visualization

image

Uber hopes to create a standard visualization system for the work of engineers in the development of unmanned vehicles based on an open version of its system.

While Uber does not hide its ambitions for unmanned vehicles, the travel company is quietly moving forward in developing new technologies for the industry. The latter is a new, open version of the unmanned vehicle visualization system (AVS), which will allow developers and engineers to share data on unmanned vehicles in an understandable and standardized way.

“Understanding what unmanned vehicles see when navigating in an urban environment is necessary to develop systems that will make them work safely.”, - Uber Xiaoji Chen, Joseph Lizi, Tim Wojtashek and Abhishek Gupta engineers write on their blog. “And just like we use standard street signs and road infrastructure to help drivers, unmanned vehicle developers will be well equipped with a standard visualization platform that will present input from sensors, classify images, display traffic information and use all other methods used to create an accurate image of the nearest space. ”

With the new AVS, Uber offers engineers a web-based tool for creating applications for the analysis of perception, movement and data processing for autonomous vehicles. Using open-source tools, Uber wants to provide developers with an autonomous and standardized environment that eliminates the need for developers to create their own visualization software for autonomous vehicles. “Using abstract visualization in AVS, developers can focus on developing basic autonomy capabilities for transport systems, remote assistance, mapping, and simulation ,” wrote a team of Uber engineers.

Autonomous driving is not just a challenge for automakers. Large technology companies such as Google, Microsoft and Nvidia, as well as various academic institutions and startups, solve various aspects of this task. Visualization tools that display what autonomous cars “see” around them are especially important to ensure the safe operation of unmanned vehicles. Thanks to more sophisticated touch technologies and other hardware solutions that are constantly being introduced, an ever-developing ecosystem for engineers is being created, since visualization becomes not just a data reproduction, but also a simulation environment and a tool for mapping, image collection, data labeling and much more. This in itself creates an entire infrastructure built around providing tools,necessary for engineers to complete these tasks.

But in all these processes, according to the Uber Engineering team, there is a noticeable lack of standards. “The lack of a visualization standard has led engineers to build their own tools based on off-the-shelf technologies and frameworks to quickly provide solutions,” the Uber blog says. “However, in our experience, these attempts to develop tools around disparate and ready-to-use components lead to the creation of poorly supported, inflexible and generally not complete systems that do not allow a solid foundation for the platform to be formed.”

AVS operates in two layers. The first level, XVIZ, is a data layer for processing information flows from various sensors in an autonomous car, for example, a point cloud from car lidars. The second layer, streetscape.gl, takes all the data from XVIZ and converts it into visual streams in the form of 3D graphics, charts, tables and videos, depending on the user's preferences.

Opening the AVS source code, Uber said that they want to not only simplify access for developers, but also attract third-party developers to add new features and contribute to the platform. The Uber Engineering team said it hopes that AVS will eventually expand to areas other than autonomous vehicles, as well as other mobility-related areas such as urban investment, geospatial analysis, and advanced mapping, among others. “We believe that an open data and tools strategy can help governments, developers, researchers, and the industry as a whole speed up the process of creating a more intelligent transport ecosystem in the future.”


image

About ITELMA
- automotive . 2500 , 650 .

, , . ( 30, ), -, -, - (DSP-) .

, . , , , . , automotive. , , .

Read more useful articles:

More articles:


All Articles