PROJECTPerception systems
Computer Vision & Robotics Perception.
Real-time pose estimation for autonomous mobile robot docking, paired with a vision-based safety interlock.
01 /
Project overview
01
Develop a C++ perception pipeline for AMR docking using OpenCV’s solvePnP on multi-marker ArUco grids for real-time 6-DoF pose estimation under heavy occlusion.
02
Fuse pose estimates with a 6-state Kalman filter to eliminate tracking jitter.
03
Build a vision-based safety interlock using MOG2 background subtraction and centroid tracking, cutting the processed pixel area by over 60%.
02 /
Technical documentation
Project documentation will be added here.
Preview a local filePRIVATE PREVIEW
Try an image, PDF, presentation, STL model, or CAD file. Your file stays in this browser. This is a preview only; it is not uploaded or published.
Maximum size: 50 MB. PowerPoint and CAD files open in a compatible app.