Aetec Vision Lab aetec.net

Aetec · Cloudflare-ready static web apps

Adaptive navigation, AI object detection, 3D recording, and reconstruction workflow demos.

Use a phone camera and available mobile sensors for live prototyping, then upgrade the same UI toward robot-mounted cameras, depth cameras, LiDAR, wheel odometry, and ROS/AMR middleware.

Object detection · motion sensing · adaptive side-following · suspicious / defect review · 3D capture · reconstruction QA

Web App 1

Phone / AMR camera adaptive navigation demo

Turns the phone into a robot vision sensor that can follow a side reference object, react to nearby obstacles, and suggest fast LEFT / RIGHT / STOP steering cues for demo navigation.

  • Camera selector and environment camera preference
  • DeviceMotion / DeviceOrientation dashboard
  • COCO-SSD browser object detector
  • Fast steering arrow for LEFT, HOLD, RIGHT, or STOP
  • Manual stereo-depth calculator for two USB cameras
Launch Adaptive Navigation Vision
Web App 2

AI object detection / suspicious object detection / defect review

For a phone or robot-mounted camera: object boxes, baseline comparison, color/texture anomaly score, heatmap overlay, meeting-room demo object detection, people / mood cue detection, and suspicious / defect review workflow.

  • Capture a known-normal baseline view
  • Compare current view against baseline for visual anomaly
  • Flag suspicious objects, unknown items, phone-like electronics, and unusual visual changes
  • Export JSON detection / review report
Launch AI Object Detection
Web App 3

3D scene capture / recording workflow

Use an iPhone or robot camera to record video, collect overlapping keyframes, monitor scan quality, preview capture coverage, export the dataset ZIP, and load a processed GLB/GLTF model back into the browser.

  • Visible red REC indicator, timer, and keyframe counter
  • Phone orientation / rough motion metadata
  • Sharpness, lighting, overlap, and motion advice
  • Texture-based map preview and ZIP export
  • GLB/GLTF model viewer for the final map output
Launch 3D Scene Capture
Web App 4

3D workflow planner

Plan the scan before recording: choose meeting-room loop, object orbit, equipment-zone pass, or corridor mapping; estimate frames, passes, overlap, and reconstruction route.

  • Capture scenario and output target planner
  • Recommended keyframes, passes, and overlap
  • Photogrammetry route selection
  • Printable demo checklist and JSON plan export
Launch 3D Workflow Planner
Web App 5

Reconstruction workbench

After recording, load metadata.json, check dataset readiness, document processing settings, and import the finished GLB/GLTF model for browser review.

  • metadata.json readiness scoring
  • Frame / yaw / quality / resolution checks
  • Processing profile notes
  • Reconstruction plan JSON export
  • GLB / GLTF / USDZ model review
Launch Reconstruction Workbench

Architecture direction

This version is a browser-safe demo, not a robot safety controller. For real AMR deployment, keep collision avoidance, braking, E-stop, bumper/LiDAR safety fields, and speed limits on the robot controller/PLC layer. Use this UI for perception visualization, dataset capture, labeling, scan planning, reconstruction QA, and engineering validation.