Coupling Apollo with the CommonRoad Motion Planning Framework​

Open-source software platforms are becoming more important for researchers in the field of autonomous driving. One of the leading platforms for autonomous driving is Apollo, whose purpose is accelerating the testing and deployment of autonomous vehicles. However, Apollo’s complex software structure hampers an easy integration of software modules, especially for motion planning. Moreover, the standalone version of Apollo requires users to upload their algorithms to a cloud platform to allow testing in diverse scenarios, which is unacceptable for many companies. In contrast, the open-source CommonRoad benchmark suite contains diverse testing scenarios in different settings, such as highways, urban environments, dense traffic, and settings where interaction with bicyclists and pedestrians is particularly important. In addition, CommonRoad provides a motion planning framework in Python, which enables rapid prototyping of motion planners, along with additional tools, such as an efficient drivability checker, a map format converter, and interfaces to the traffic simulators SUMO and CARLA. In this work, we introduce a Python API between the planning module of the Apollo platform and the CommonRoad software framework, which bridges the gap between rapid prototyping for planning algorithms and their validation through real test drives. We demonstrate our interface with two scenarios.     «

Open-source software platforms are becoming more important for researchers in the field of autonomous driving. One of the leading platforms for autonomous driving is Apollo, whose purpose is accelerating the testing and deployment of autonomous vehicles. However, Apollo’s complex software structure hampers an easy integration of software modules, especially for motion planning. Moreover, the standalone version of Apollo requires users to upload their algorithms to a cloud platform to allow testi...     »