The collection of Composable Benchmarks for Motion Planning on Roads (CommonRoad) provides researchers with a means of evaluating and comparing their motion planners. A benchmark consists of a scenario with a planning problem, a vehicle dynamics model, vehicle parameters, and a cost function composing a unique ID. Along with the benchmarks, we provide several tools for motion planning.

Main features:

  • CommonRoad Input-Output: Methods to read, write and visualize CommonRoad scenarios and planning problems.
  • CommonRoad Scenario Designer: Conversion maps from OpenStreetMap and OpenDRIVE and graphical user interface to edit maps and scenarios manually and to populate the maps with traffic using the traffic simulator SUMO.
  • CommonRoad Drivability Checker: Check the drivability of a trajectory, check for collisions, or to perform transformations into curvilinear coordinate systems.
  • CommonRoad Route Planner: High-level guidance for motion-planning algorithms and for defining reference paths.
  • CommonRoad-Search: Uninformed & informed search algorithms with motion primitives to solve motion planning problems.
  • CommonRoad-Reach: Methods for computing reachable sets and extracting driving corridors for automated vehicles in dynamic traffic scenarios.
  • CommonRoad-RL: Solve motion planning problems on CommonRoad using reinforcement learning methods, currently based on Stable Baselines.
  • CommonRoad-Geometric: Graph-based autonomous driving research, fully customizable data processing pipeline for extracting PyTorch-based graph datasets from traffic scenarios.
  • CommonRoad-CriMe: Unified notations, vehicle models, and coordinate systems for criticality measures of autonomous vehicles.
  • CommonRoad-OpenSCENARIO-Converter: Automatic traffic scenario conversion between OpenSCENARIO and CommonRoad.

For more details, please refer to the CommonRoad website.