This recent Thursday Baidu made the announcement of the release of ApolloScape, a platform that has been billed as the world’s largest open-source dataset for autonomous driving technology.

Talking of Apollo, a flexible architecture with high performance is basically an open autonomous driving platform which supports fully autonomous driving capabilities.

ApolloScape made its release under autonomous driving platform Apollo by Baidu, and according to them is hoped to become “the Android of the auto industry.” Apollo gives access to the developers with a complete set of service solutions as well as open-source codes. The open sourced data of ApolloScape now also provides developers with a base for building self-driving vehicles.

When compared to any other open-source autonomous driving dataset, the data volume of ApolloScape is 10 times greater and that includes Kitti and CityScapes. This data can be utilized for road networks, simulation scenes, perception, as well as to enable autonomous driving vehicles to be trained in more complex environments like weather and traffic conditions. ApolloScape also defines with pixel-by-pixel semantic segmentation technique 26 different semantic items  The goal of image segmentation is to make their analyzing easier.

The ApolloScape dataset will thus result in a great save of a huge amount of time for all researchers and developers on real-world sensor data collection.

Going further, ApolloScape can also successfully simulate the complex scenario of dozens of vehicles that are driving on the same road. It is as of now one of the most advanced intelligent driving simulation technologies that are available to help autonomous driving developers:

• Effectively examine and optimize forecasting.

• Decision making as well as path planning.

According to a report by Rand Corporation, there would be a requirement of a fleet of 100 vehicles that will drive nonstop for 500 years in order to accumulate real road testing data that is sufficient to conclude a 20 percent advantage for autonomous vehicles over human drivers.

Ahead of data, ApolloScape, in order to create a simulation platform that aligns with real-world experience will also facilitate advanced research on cutting-edge simulation technology.

Apollo has also further announced it has joined a top-tier research alliance investigating state-of-the-art technologies in computer vision and machine learning for automotive applications: The Berkeley DeepDrive (BDD) Industry Consortium.

Known to have Housed at the University of California, Berkeley and led by Faculty Director of the California Program for Advanced Transportation Technology (PATH), Professor Trevor Darrell, the BDD consortium has attracted big tech names as partners, that including Qualcomm, NVIDIA, Ford and General Motors.

The main research focus of BDD is in the fields of artificial intelligence that are known as deep reinforcement learning, clockwork FCNs and cross-modal transfer learning.

Baidu also released a large number of automated driving open data sets for the Apollo platform at a press conference. The high-quality real data is said to be an indispensable "raw material" for autonomous driving developers. However, only a few companies have the ability to develop as well as maintain an automated driving platform that regularly not only calibrates and collects vast amounts of new data but is also suitable for the same.

The Vice president of Baidu and Head of Baidu Research Institute, Haifeng Wang, told, “The partnership to ramp up the innovation of theoretical research, applied technology, and commercial applications will incorporate the industrial resources of Apollo and Berkeley’s top academic team.”

Apollo Open Platform and BDD at CVPR 2018 (IEEE International Conference on Computer Vision and Pattern Recognition) will jointly conduct a Workshop on Autonomous Driving this June in Salt Lake City where based on ApolloScape they will organize task competitions.

For More Information: GitHub

Official Link: Apollo