Commits on Dec 5, 2019

1. refactor(doTransform): 10-100x faster transform of pointcloud …

   It seems like generating an Eigen::Vector3f and transforming it for each iteration is very inefficient. In our case, this mod took transform-time of a point-cloud from 90-100ms to 1-2ms.
   
   Verified to give same result as old doTransform using:
   
   ```
   sensor_msgs::PointCloud2ConstIterator<float> x_old(cloud1, "x");
   sensor_msgs::PointCloud2ConstIterator<float> y_old(cloud1, "y");
   sensor_msgs::PointCloud2ConstIterator<float> z_old(cloud1, "z");
   
   sensor_msgs::PointCloud2ConstIterator<float> x_new(cloud2, "x");
   sensor_msgs::PointCloud2ConstIterator<float> y_new(cloud2, "y");
   sensor_msgs::PointCloud2ConstIterator<float> z_new(cloud2, "z");
   
   std::vector<double> compare_vector;
   for (; x_new != x_new.end(); ++x_new, ++y_new, ++z_new, ++x_old, ++y_old, ++z_old) {
   compare_vector.push_back(*x_new - *x_old);
   compare_vector.push_back(*y_new - *y_old);
   compare_vector.push_back(*z_new - *z_old);
   }
   double max_diff = *max_element(compare_vector.begin(), compare_vector.end());
   double min_diff = *min_element(compare_vector.begin(), compare_vector.end());
   
   ROS_INFO("Biggest differences: %f, %f", max_diff, min_diff);
   ```
   
   Not sure how/where to put this test-code in a proper test, so I'll leave it here until I get some feedback.

   

   steinmn committed Dec 5, 2019
   Configuration menu

   • View commit details
   • Copy full SHA for 63802d0
   • Browse repository at this point

   Copy the full SHA

   63802d0 View commit details

   Browse the repository at this point in the history