Work

A selection of open-source work across 3D computer vision, robotics, and machine learning, from visual SLAM and deep image matching to LiDAR perception and state estimation.

• Neuracore

  Physical-AI infrastructure for robot learning: data capture, training, and deployment for embodied agents.

  Code
• 

  SuperSLAM

  Open-source framework for deep-learning-based visual SLAM.

  Code
• 

  Deep Image Matching using Local Feature Transformer

  Pose-estimation pipeline enabling 3D reconstruction using the Local Feature Transformer (LoFTR).

  Code
• 

  State Estimation of a Drone by Visual-Inertial Sensor Fusion

  Optimal state estimation using extended & unscented Kalman filters, implemented in MATLAB.

  Code
• 

  Fast Plane Extraction from a 3D Point Cloud

  Fast, parallel implementation of the RANSAC algorithm to segment a plane in a point cloud.

  Code