[P1] Object detection + labeling — identify and name common objects #468

Closed
opened 2026-03-05 12:06:59 -05:00 by sl-jetson · 0 comments
Collaborator

Goal

Identify objects in the environment using RealSense RGB + YOLO for contextual awareness.

Requirements

  • ROS2 node running YOLOv8n on Orin (TensorRT FP16)
  • Detect COCO 80 classes: chairs, tables, dogs, cats, cars, bicycles, etc
  • Publish /saltybot/detected_objects (DetectedObjectArray.msg: class, confidence, bbox, distance)
  • Distance estimation from RealSense depth at bbox center
  • Spatial memory: remember where objects were seen (link to obstacle map)
  • Voice query: "whats in front of you?" → TTS list of visible objects
  • 15+ fps on Orin alongside person tracking
  • Configurable: minimum confidence, classes of interest filter
## Goal Identify objects in the environment using RealSense RGB + YOLO for contextual awareness. ## Requirements - ROS2 node running YOLOv8n on Orin (TensorRT FP16) - Detect COCO 80 classes: chairs, tables, dogs, cats, cars, bicycles, etc - Publish /saltybot/detected_objects (DetectedObjectArray.msg: class, confidence, bbox, distance) - Distance estimation from RealSense depth at bbox center - Spatial memory: remember where objects were seen (link to obstacle map) - Voice query: "whats in front of you?" → TTS list of visible objects - 15+ fps on Orin alongside person tracking - Configurable: minimum confidence, classes of interest filter
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#468
No description provided.