Toolbox for Map Conversion and Scenario Creation for Autonomous Vehicles.
-
Updated
Aug 27, 2026 - Python
Toolbox for Map Conversion and Scenario Creation for Autonomous Vehicles.
Interface between CommonRoad and the traffic simulator CARLA.
Deep-learning model for selection of high-level corridors in motion planning
A trajectory planning framework integrating risk-reachable sets and safe reinforcement learning for enhanced autonomous driving safety.
Multi-vehicle autonomous valet parking with CARLA, CommonRoad, occupancy-grid conflict resolution, Hybrid A*, and V2I-style ZeroMQ coordination.
Developed an autonomous agent (a car) capable of safely executing maneuvers in highway environments surrounded by other vehicles. The scenarios were based on CommonRoad benchmarks, requiring the agent to operate effectively in a partially observable environment. The project focused on ensuring safe navigation and decision-making.
To associate your repository with the commonroad topic, visit your repo's landing page and select "manage topics."