← Back to Work

MV-NDVI

Ongoing

Multi-View NDVI • Custom Sensor Rig • Precision Agriculture

Overview

MV-NDVI is a research project exploring the extension of traditional Normalized Difference Vegetation Index (NDVI) into three dimensions. By capturing multi-spectral imagery from multiple viewpoints, we aim to create volumetric plant health representations that go beyond conventional 2D vegetation analysis.

Custom Sensor Rig

We designed and built a handheld multi-camera acquisition system specifically for this project. The rig integrates four synchronized cameras and a LiDAR sensor for robust pose estimation.

Cameras

  • Monochrome + Red bandpass (660nm)
  • Monochrome + NIR bandpass
  • RGB color camera

Pose Estimation

  • Livox LiDAR
  • Real-time SLAM
  • Multi-sensor fusion

Data Collection

We collected our own dataset in a vegetable garden, capturing multi-spectral imagery of various plants at different growth stages. The controlled environment allows us to validate our results against ground-truth plant health measurements.

Publication

This work will first be presented as a poster presentation at ICRA Field Robotics Workshop 2026, followed by a full paper publication.

Read Paper (Coming Soon) →