Abstract
Testing research approaches for autonomous and connected driving is often challenging as there are many real-world factors involved. With this work, we present V-TRAD, a visually-augmented testbed for investigating path tracking and planing algorithms, analyzing real-time sensors and evaluating driving scenarios in a unified physical-virtual environment. As one of our main contributions, we provide custom miniaturized connected vehicles and an extendable software stack that allows complex simulations and versatile embedded in-situ visualization. The testbed features a motion capture and projector system and a hand-held pointing device. To demonstrate the functionality, versatility and potential of V-TRAD, we implemented three scenarios that visualize path trajectories, use live LiDAR sensor data for creating environmental maps and utilize agents to manage autonomous driving strategies in a highway scenario. We make V-TRAD open source and provide additional building instructions to support researchers, practitioners and students to test their approaches in an easy and realistic way.
Related Publications
V-TRAD:
A Visually-Augmented Testbed for Research in Autonomous Driving
| Category | Robotics |
|---|---|
| Project date | 2025 |
| Collaborators | Paul Auerbach |
| Technologies and Methods |
Autonomous Driving Hybrid Testbed In-Situ Viz Reinforcement Learning Environmental Maps ROS2 LIDAR Simulation Sim2Real Path Tracking Visual Analytics |