Apollo provides open training data sets, which include laser point cloud obstacles, traffic light detection and the Road Hackers system. Going forward, Apollo will continue to add additional autonomous driving data.
The Traffic Light Detection feature provides the image data for common vertical traffic lights. All data is collected during daylight in 1080P resolution.
Current data covers days that are: Sunny, Cloudy, Foggy.
The Road Hackers system provides two types of data:
• Street View Images that are front-of-vehicle images
• Vehicle Motion Status that includes the current speed and track curvature of the vehicle
Apollo provides Open Demonstration Data, which includes sensor demonstration data, calibration demonstration data, end-to-end data, and EGO localization demonstration data.
In the future, Apollo will offer additional data for on-board modules.
The Calibration Demonstration Data set provides:
• A data package collected with the calibration data recorder.
• The raw packets from HDL-64ES3.
• Relative motion from GNSS/INS and its corresponding MD5 checksum.
End-to-End Data provides the original sensor data as input and the vehicle’s control instructions as output.
Currently this means that input data consists of images, and output is the vehicle’s control instructions, such as: Steering wheel angle, Acceleration, Braking.
EGO Localization Demonstration Data provides:
• Environmental information
• Related location values
• Consecutive video frames
• Deep neural network modeling
Using this data, developers can learn the mechanisms of the positioning module.
Apollo provides the computing capability and annotation services that match the open data.
We are committed to empowering all levels of developers through targeted support on the Apollo Open Data Platform. Apollo invites you to join us!Join Now
We are committed to empowering all levels of developers through targeted support on the Apollo Open Data Platform. Apollo invites you to join us!