feat: Semantic sidewalk segmentation — TensorRT on Orin (Phase 3) #72

Closed
opened 2026-03-01 01:01:45 -05:00 by sl-jetson · 0 comments
Collaborator

Summary

Train and deploy a semantic segmentation model to distinguish sidewalk vs road vs grass vs obstacle. Runs on Orin Nano Super (67 TOPS).

Deliverables

  • sidewalk_segmentation_node — subscribes camera images, publishes segmentation mask
  • TensorRT FP16 optimized model (DeepLabV3+ or BiSeNetV2)
  • Costmap integration — feed segmentation into Nav2 as traversability layer
  • Training pipeline doc (dataset: Cityscapes/Mapillary subset + fine-tune)

Context

LIDAR sees obstacles but not surface type. Camera segmentation tells Nav2 to prefer sidewalk over road, avoid grass.

## Summary Train and deploy a semantic segmentation model to distinguish sidewalk vs road vs grass vs obstacle. Runs on Orin Nano Super (67 TOPS). ## Deliverables - sidewalk_segmentation_node — subscribes camera images, publishes segmentation mask - TensorRT FP16 optimized model (DeepLabV3+ or BiSeNetV2) - Costmap integration — feed segmentation into Nav2 as traversability layer - Training pipeline doc (dataset: Cityscapes/Mapillary subset + fine-tune) ## Context LIDAR sees obstacles but not surface type. Camera segmentation tells Nav2 to prefer sidewalk over road, avoid grass.
sl-jetson self-assigned this 2026-03-01 01:01:45 -05:00
Sign in to join this conversation.
No Label
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: seb/saltylab-firmware#72
No description provided.