Aus Kognitive Robotik

Wechseln zu: Navigation, Suche

A small METRIK subproject concerning mapping of urban vegetation.

Workflow and tools


image select

Need a tool (python / opencv) that

  1. loads video
  1. checks if h.264
    1. if yes: load all keyframes
    2. if no: check frame for quality: blur, select if no blur
  1. emit list or manipulate directory directly to prepare image
  2. selection as input to photogrammetry tools


Agisoft: currently testing Agisoft Photoscan for volumetric assesment

  • works nicely apart from certain GUI issues

Visual Structure from Motion (vsfm): currently testing vsfm

  • preliminary results


  • evaluate

SVO: Semidirect Visual Odometry

from Robotics and Perception Group Uni Zürich

Motion parallax based depth estimation

  • see prototype in urban-vegetation/scripts/


Canon Hack Development Kit (CHDK)

take full control of a Canon Powershot device

Aerial systems

Currently investigating two kinds of flying systems for carrying the camera

Large system

10-inch sized larger quadrotor: can easily carry a nice camera but generates a lot of downwind that makes the leaves wiggle a lot. Maybe try a slight angle on the camera.

Using PPZ, Lisa/M 2.0, first autonomous flights made in 2014

Small system

5-inch sized small quadrotor: is much nicer for flying in constrained spaces and much less downwind. Very limited payload.

Using Naze32 with SummerGames2.5:

  • uses mavlink telemetry protocol
  • check configuration options with qgroundcontrol
  • telemetry and GPS works fine
  • check GPS position hold


Lisein et al.: A Photogrammetric Workflow for the Creation of a Forest Canopy Height Model from Small Unmanned Aerial System Imagery

Berni et al.: Remote sensing of vegetation from UAV platforms using lightweight multispectral and thermal imaging sensors

Dandois & Ellis: High spatial resolution three-dimensional mapping of vegetation spectral dynamics using computer vision

Links and related stuff