Skip to content

Future Work

This architecture has the potential to be extended into additional autonomous driving capabilities beyond the current AVP implementation. Below are some possible directions for future expansion.

Autonomous Valet Parking (AVP) — Implemented

This architecture was already extended to develop a complete multi-vehicle AVP implementation, using namespaced Autoware stacks, Zenoh bridging, and parking goal execution.

See Multi-Vehicle AVP.

Adaptive Cruise Control (ACC)

Enable vehicles to maintain safe distances from others by automatically adjusting their speed based on a detected lead vehicle.

Cooperative Maneuvers

Implement coordinated behaviors such as merging, overtaking, and intersection handling, where vehicles communicate to negotiate priority and avoid conflicts.

Occlusion Handling

Integrate strategies to detect and address occluded objects, using multiple viewpoints or predictive modeling to reduce blind spots and improve safety in complex environments.

Distributed Image Stitching & Sensor Fusion

Enable multiple vehicles or infrastructure-mounted cameras to share and combine visual or LiDAR data in real-time, creating a unified, high-coverage perception view that enhances detection accuracy and environmental awareness.

Additional Directions

  • Improved network robustness and transport tuning, including exploration of alternative Zenoh configurations, transport backends, and message scheduling strategies to reduce timing variability under multi-host operation.
  • High-bandwidth sensor distribution, such as sharing camera or LiDAR streams across hosts to enable distributed perception and infrastructure-assisted sensing.
  • Cloud or edge-assisted execution, where selected components are offloaded to external compute resources to reduce per-host load and improve scalability.
  • Deployment on embedded robotic platforms, enabling validation under real-world wireless conditions and hardware constraints.
  • Scalability beyond three hosts, using dedicated networking infrastructure or more homogeneous compute platforms to evaluate behavior under higher vehicle counts.