ecto_pcl

ecto_pcl is another collection of ecto modules and tools that make ones life much more rosey when developing perception based ecto graphs.

Dependencies

• cmake
• git
• python
• boost
• ecto https://github.com/plasmodic/ecto

• pcl 1.0 (non ros version, standalone)

  http://www.pointclouds.org/downloads/linux.html

  Quoted here for convenience:

  sudo add-apt-repository ppa:v-launchpad-jochen-sprickerhof-de/pcl
  sudo apt-get update
  sudo apt-get install libpcl-dev
  

Style

If you are using eclipse, you may use the style file available here:

https://github.com/plasmodic/ecto/raw/master/misc/ecto.style.xml

0.4.5 (2016-04-24)

• add dependency on proj to fix PCL compilation on kinetic
• remove useless dependency on ecto_ros
• Contributors: Vincent Rabaud

0.4.4 (2016-04-24)

• compile on Xenial
• fix running examples for doc
• Contributors: Vincent Rabaud

0.4.3 (2015-09-01)

• SACSegmentationFromNormals: support input indices
• SACSegmentation: support input indices
• fix SACSegmentation cell
• ConvexHull: add input indices
• remove useless dependency
• clean tests

• Added new parameter to ExtractIndices - Added parameter to keep the filtered cloud organized, i.e., coordinates of 'removed' points are set to 'NaN'

• add proper test dependency
• add proper nosetest
• clean extensions
• Contributors: Michael Görner, Sven Albrecht, Vincent Rabaud

0.4.2 (2014-12-21)

• Use the new header with PCL 1.7.0+
• Always find pcl_conversions Should be okay since it's a package dependency.
• Find pcl_conversions if building against PCL 1.7.0+ It gets included depending on the PCL version in src/ros/pcl_bridge.cpp
• fix sample with newer API
• Contributors: Vincent Rabaud, pgorczak

0.4.1 (2014-10-15)

• Merge pull request #34 from v4hn/pass-through-indices Add PassThrough for index extraction
• Merge pull request #33 from v4hn/euclidean-cluster-extraction-indices Allow EuclideanClusterExtraction to work on input indices
• Add PassThrough for index extraction This module behaves similarly to PassThrough with two changes: - It also accepts an optional set of input indices of the input cloud It's probably not good to add this to PassThrough as well, because it breaks whatever[:] >> passthrough[:]
  • It produces PointIndices of the original cloud instead of a new cloud
• Allow EuclideanClusterExtraction to work on input indices This allows for a pipeline which outputs Clusters as sets of indices of the _original (organized) input cloud from the depth image camera. That way 2d bounding boxes in the input image can be more efficiently computed.
• Merge pull request #32 from v4hn/pcl-cell-return-values pcl_cell: don't ignore the return value of the wrapped cell
• pcl_cell: don't ignore the return value of the wrapped cell This makes sure ecto actually stops processing in case on of the cells produces an error and returns something else than ecto::OK.
• Merge pull request #31 from v4hn/convex-hull-explicit-dimensionality ConvexHull: add parameter to explicitly set dimensionality of the data/hull
• ConvexHull: add parameter to explicitly set dimensionality of the data/hull I had qhull 2012 via pcl 1.7.2 complaining over here that the input is, indeed, _not three dimensional when computing the hull of a detected table. Turned out pcl is always told to reconstruct a 3d-hull.
• Contributors: Michael Görner, Vincent Rabaud, v4hn

0.4.0 (2014-04-04)

• depend on pcl_conversions which is lighter
• drop Fuerte support
• Contributors: Vincent Rabaud

0.3.14 (2013-08-28 19:12:21 -0800)

• fix dependencies with PCL