Executes and combines the various SLAM packages to produce useful input and output data. This includes coordinating the various nodes in order to produce pickled and ros bags of output data, and stereo rectified images.

Example Usage

#######

display a folder of images rosrun dataset displayFolder directoryToImages rosrun dataset displayFolder directoryToImages --debayer True --delayTime 66

debayer images before displaying with BG2RGB wait 66 ms before displaying a new folder

1. begin the base nodes. This includes the dataset server, bumblebee configuration, and bumblebee Rectificaiton Server. Edit the baseLaunch.xml file settings, notably the dataset folder where the raw captured images reside, and the stereo output configuration folder. Add the tag publishTransform = True for this first bag. This will allow a transform tree to be published and recorded.

roslaunch dataset baseLaunch.xml

OR

roslaunch dataset baseLaunch.xml publishTransform:= True

2. Create a rosbag of bayer encoded left and right images. This saves both the tf transform chain topic, and /dataset/currentImage raw bayer topic to a rosbag rawImageBag.bag To play it back,rosrun rosbag play -r 0.5 rawImageBag.bag, this runs the bag at half the rate it was originally published. Run baseLaunch before execution with the publish transform option. Change the args parameter in the CreateDatasetBag.xml to change the minimum number of Images

roslaunch dataset CreateDatasetBag.xml

run baseLaunch with no publish transform parameter

First create a rosbag of stereo images.

roslaunch dataset CreateStereoBag.xml rosplay

rosrun dataset unpackBumblebeeData /media/ryan/EXTRA/DATA3/01-01-1970--02-02-06.rawData

first run

roslaunch dataset baseFeatures.xml

then to execute individual detector extractions

rosrun dataset extractLoopFeatures /home/ryan/DATA3 A_1 --max_images 10

OR args dataset:= DATA3 -> the folder name where to search relative to root_dir for the dataset

default_bag_dir := Bags -> search for stereo bags @ root_dir/dataset/default_bag_dir default_out_dir := Features -> save all output data @ /media/ryan/EXTRA/output/default_out_dir detectorName := FAST -> detector Name (SURF,FAST,ORB,AKAZE,BRISK) loopNumber := 1 maxImages := 15 root_dir := /home/ryan -> where to find the dataset track_name := A ->track sequence within a single video

roslaunch dataset extractSingleLoopFeatures.xml

To execute in looped mode where a dataset has been split up into individual loops, provide a metaData file as shown in github project gpsData

1. execute a minLaunch.xml where it provides only the stereo processing node and tf transform chain.

roslaunch dataset minLaunch.xml BumblebeeConfigurationFolder:=DIRECTORY_TO_CALIBRATION publishTransform:=True

2. once these have started up, start recording the stereo output topics and saving them to bags, give a track and loop number to select which set of images to use

roslaunch dataset CreateStereoLoop.xml ################ ###to get libviso and loop bags

1. execute configuration script ->lib viso parameters must be adjusted manually in the visoConfig.xml file in viso_extractor

roslaunch dataset visoLoopLaunch.xml

ensure that the DATAFolder/Bags folder has been CreateDatasetBag ->adjust output folders in xml

roslaunch dataset CreateStereoLoop.xml

###################### ##extracting features from looped data

1. initilaize the front_end feature node for feature extraction rosrun front_end feature_node

rosrun dataset extractLoopFeatures /home/ryan/DATA3 A_1 --max_images 50

2)Begin extracting features from a rosbag that contains stereo rectified Images

rosrun dataset extractOperatingCurves /home/ryan/DATA3 A_1 rosrun dataset extractOperatingCurves /home/ryan/DATA3 A_1 --output_directory pathToPickledOutputObjects -> to view outputs rosrun slam_analysis plotDetectionStatistics /home/ryan/DATA3 A_1 FAST

3. generate stereo summary statistics

rosrun dataset extractStereo.py