HiveMind · Autonomous Highway Sandbox
Research in progress
- My role
- Training & Simulation Lead
- Timeframe
- August 2026 – Present
- Work context
- Directed research · Three-person team
- Tools & skills
- SUMO · TraCI · PettingZoo / TorchRL · OpenUSD · NVIDIA Omniverse · ovrtx · Isaac Sim
Cooperative traffic research
Since August 2026, I have been collaborating with two teammates on a multi-agent autonomous-vehicle research simulation to study cooperative traffic behavior, with advice from an NVIDIA engineer. The project is in progress, using SUMO, TraCI, PettingZoo/TorchRL, OpenUSD, NVIDIA Omniverse, ovrtx, and Isaac Sim.
Simulation and scene pipeline
We are developing repeatable traffic baselines and reinforcement-learning workflows for connected vehicles to share state and coordinate speed, lane selection, routing, and fleet behavior.
In parallel, we are developing an OpenUSD data and scene pipeline to synchronize timestamped SUMO vehicle states with interactive highway environments for visualization and validation.
Libraries and skills
Our work spans multi-agent simulation, reinforcement-learning workflows, OpenUSD scene composition, and timestamped data synchronization. The stack includes SUMO, TraCI, PettingZoo/TorchRL, OpenUSD, NVIDIA Omniverse, ovrtx, and Isaac Sim.
Development includes vehicle physics, control, simulated LiDAR sensing, and safety-validation workflows.