Merge pull request autowarefoundation#762 from CPFL/release/1.4.0

Release/1.4.0 autowarefoundation#762

README.md

@@ -27,7 +27,7 @@ Autoware is protected by the BSD License. Please use it on at your own discretio
 
 Free manuals can be found at [https://github.com/CPFL/Autoware-Manuals](https://github.com/CPFL/Autoware-Manuals). You are encouraged to join maintenance of those manuals. Thanks for your cooperation!
 
-See also Autoware/docs for the tips of Autoware development, including the coding style and branching model.
+See also [branching_model](https://github.com/CPFL/Autoware/blob/master/docs/en/branching_model.md) for the tips of Autoware development, including the coding style and branching model.
 
 ## License
 

docs/en/branching_model.md

@@ -1,22 +1,32 @@
-# Branching model for Autoware development
+# Development Model in Autoware
+## Development Workflow
+To assure traceablity in our code we follow this development process:
+* If you are working on a feature/bug/new-release => create an issue first
+* create a branch, work on the feature/bug/new-release, refer in at least one commit to the issue #number
+* open pull request for given feature/bug/new-release and fill out pull request template. 
+
+This will assure that we know where the feature originated from and the implementation will be linked to the feature request (or bug report). Otherwise there is absolutely no traceability.
+Secondly, with the pull request template it will be easier for the reviewer(s) to understand the changes and we can later on convert "Steps to Reproduce" into integration tests.
+
+## Branching model for Autoware development
 
 In order to keep efficient open-source development of Autoware, we ask all developers to comply with the branching model described below.
-On this model, we maily use five branches - master, develop, release, feature, and hotfix.
+On this model, we maily use five branches - master, develop, release, feature, and bugfix.
 
-## Development
+### Development
 In general, developers should work on the "develop" branch.
 When you develop new functions, please check out a new branch, "feature/[branch_name]", from the "develop" branch, and modify the code there.
 After the modificaiton, please send a pull request to the "develop" branch.
 
-## Release
+### Release
 This situation is true of only the persons in charge of releasing Autoware.
 Once we complete some major development, we make a new release.
 For the release work, please checkout a new branch, "release/[1.xx.yy], from the "develop" branch, and modify the code there.
 After the modification, please send a pull request: "xx" in version of release/[1.xx.yy] is increased when checked out from the "develop" branch, and yy is also increased when bug fixes are done.
 Finally, we merge this branch to the master branch, attaching a version tag.
 
-## Bugfix
-If we encounter bugs in the "master" branch after the release, we check out the "hotfix" branch from the "master" branch and fix the bugs.
+### Bugfix
+If we encounter bugs in the "master" branch after the release, we check out the "bugfix" branch from the "master" branch and fix the bugs.
 This branch is merged to each corresponding branch - master, release, and develop.
 After the merge, the version of the master and the release branches is increased.
 

docs/issue-template.md

@@ -0,0 +1,23 @@
+Fill-out only one section depending on whether you are reporting a bug or a new feature.
+
+# Bug
+## Expected Behavior
+
+
+## Actual Behavior
+
+
+## Steps to Reproduce the Problem
+
+  1.
+  1.
+  1.
+
+## Specifications
+
+  - Ubuntu version:
+  - ROS Version:
+  - Autoware branch:
+
+# New Feature
+Describe where does new feature request come from. Add high-level picture of how the new feature should work.

docs/pull-request-template.md

@@ -0,0 +1,27 @@
+## Status
+**PRODUCTION / DEVELOPMENT**
+
+## Description
+A few sentences describing the overall goals of the pull request's commits.
+
+## Related PRs
+List related PRs against other branches:
+
+branch | PR
+------ | ------
+other_pr_production | [link]()
+other_pr_master | [link]()
+
+
+## Todos
+- [ ] Tests
+- [ ] Documentation
+
+
+## Steps to Test or Reproduce
+Outline the steps to test or reproduce the PR here.
+
+```
+roslaunch pkg_A executable_A
+```
+The car should move.

ros/src/computing/perception/detection/lib/fusion/CHANGELOG.rst

@@ -0,0 +1,83 @@
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+Changelog for package fusion
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+1.4.0 (2017-08-04)
+------------------
+
+1.3.1 (2017-07-16)
+------------------
+
+1.3.0 (2017-07-14)
+------------------
+* convert to autoware_msgs
+* Contributors: YamatoAndo
+
+1.2.0 (2017-06-07)
+------------------
+* fix circular-dependency
+* Contributors: Shohei Fujii
+
+1.1.2 (2017-02-27 23:10)
+------------------------
+
+1.1.1 (2017-02-27 22:25)
+------------------------
+
+1.1.0 (2017-02-24)
+------------------
+
+1.0.1 (2017-01-14)
+------------------
+
+1.0.0 (2016-12-22)
+------------------
+* Fixed a bud. It caused by reference to out of image size
+* Revert "Fixed a bud. It caused by reference to out of image size"
+  This reverts commit 9d9568b650ab372883458bf42a93f13bab0be8e3.
+* Fixed a bud. It caused by reference to out of image size
+* Fix dependency
+* Hotfix segmentation fault in range_fusion
+* Add missing dependencies
+* Add ROS include path to 'include_directories'
+* Add new method for distance calculation of fusion
+  Add new class to handle several distance calculation methods.
+  We can now switch the calculation methos below by this class:
+  * Shortest
+  * Median
+  * Mode (most common value)
+  Confine scope of search_distance.h to lib/fusion
+  Reconstruct search_distance.cpp
+  * use function call instead of class method
+  * fix typo
+  * start function name with lower case
+  * return 0 immediately if argument vector size is empty
+  * use const reference as many as possible to avoid data copy
+  Do not pass reference to std::sort
+* Correct dependency name
+* Delete warning cause
+  * Use 'ranged-for' to reset std::vector
+  * Set nested template parameter to C++11 style
+* Change calculation method of ranging
+  [Before] minimun distance in a detection rectangle
+  [After]  median distance in a detection rectangle
+* Delete image size fixing
+* Use c++11 option instead of c++0x
+  We can use newer compilers which support 'c++11' option
+* Integration of RCNN object detection on Autoware
+  **Added a new library librcnn, which executes the object recognition using the Caffe framework, specifically the fast-rcnn branch.
+  git clone --recursive https://github.com/rbgirshick/fast-rcnn.git
+  -Requires CUDA for GPU support.
+  To take advantage of cuDNN, at least CUDA 7.0 and a GPU with 3.5 compute capability is required.
+  -Compile Caffe, located in caffe-fast-rcnn.
+  Complete the requisites:http://caffe.berkeleyvision.org/install_apt.html
+  -Download the pretrained models:
+  http://www.cs.berkeley.edu/~rbg/fast-rcnn-data/voc12_submission.tgz
+  -Modify the CMakeFiles and point them to your caffe and models directories.
+  **Modified KF to use the new NMS algorithm
+  **Modified Range fusion, it will not execute unnecesary fusions.
+  **Added Configuration Messages to Runtime manager and RCNN node launch files
+* Updated point2image to set minh in the message
+* fix-dependency
+* Initial commit for public release
+* Contributors: AMC, Hiroki Ohta, Manato Hirabayashi, Shinpei Kato, Syohei YOSHIDA, Yukihiro Saito, h_ohta

ros/src/computing/perception/detection/lib/fusion/package.xml

@@ -1,7 +1,7 @@
 <?xml version="1.0"?>
 <package>
   <name>fusion</name>
-  <version>0.0.0</version>
+  <version>1.4.0</version>
   <description>The fusion package</description>
   <maintainer email="[email protected]">Masao KONDOH</maintainer>
   <license>BSD</license>

ros/src/computing/perception/detection/lib/image/dpm_ocv/CHANGELOG.rst

@@ -0,0 +1,63 @@
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+Changelog for package libdpm_ocv
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+1.4.0 (2017-08-04)
+------------------
+
+1.3.1 (2017-07-16)
+------------------
+
+1.3.0 (2017-07-14)
+------------------
+* convert to autoware_msgs
+* Contributors: YamatoAndo
+
+1.2.0 (2017-06-07)
+------------------
+* clean directories
+* ROS Kinectic Upgrade tested on Ubuntu 16.04 and OpenCV 3.2.0
+  Modules not included:
+  -orb_localizer
+  -dpm_ocv node and lib
+  Everything else working
+  Added some libraries for Gazebo on src/extras
+* Update for kinetic
+* Contributors: Yukihiro Saito, Yusuke FUJII, amc-nu
+
+1.1.2 (2017-02-27 23:10)
+------------------------
+
+1.1.1 (2017-02-27 22:25)
+------------------------
+
+1.1.0 (2017-02-24)
+------------------
+
+1.0.1 (2017-01-14)
+------------------
+
+1.0.0 (2016-12-22)
+------------------
+* Fix for rosjava installed platform
+  Some packages don't declare package dependencies correctly.
+  This makes message jar files built failure.
+* Always check return value from CUDA API
+* Correct wrong type name
+* Add CUDA_CHECK after memory accesses
+* Remove extra level of indirection from cuMemFreeHost
+  cuda-memcheck complained about a failure to free, and no doubt the memory leak would soon lead to more problems.
+* Need one more convStream element
+  The loop starting at 493 is off-by-one without it, as are lines 568 and 569.
+  Furthermore, if it so happens that the model has lots of parts but few levels (ie a small λ) such that parts ("n" here) is more than two greater than the HOG pyramid levels ("numLevels"), line 580 will overflow. This is TODO, but a higher priority IMO is to redesign the use of streams and texture memory, as texMap prevents parallelisation of part map processing.
+* Fix off-by-one error
+  cuda-memcheck reported an out-of-bounds read here, which would be followed by an out-of-bounds write. I think changing from <= to < is sufficient.
+* Improve cmake CUDA configuration
+  Use cmake CUDA feature as possible for removing absolute paths.
+* Use c++11 option instead of c++0x
+  We can use newer compilers which support 'c++11' option
+* Use modified OpenCV function for adjusting parameter
+* Update dpm_ocv
+  - support using both GPU and CPU
+  - clean up code
+* Contributors: KenYN, Syohei YOSHIDA

ros/src/computing/perception/detection/lib/image/dpm_ocv/package.xml

@@ -1,7 +1,7 @@
 <?xml version="1.0"?>
 <package>
   <name>libdpm_ocv</name>
-  <version>0.0.0</version>
+  <version>1.4.0</version>
   <description>The opencv_dpm_gpu package</description>
   <maintainer email="[email protected]">Yusuke Fujii</maintainer>
   <author email="[email protected]">hoshika</author>

ros/src/computing/perception/detection/lib/image/dpm_ttic/CHANGELOG.rst

@@ -0,0 +1,44 @@
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+Changelog for package libdpm_ttic
+^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+
+1.4.0 (2017-08-04)
+------------------
+
+1.3.1 (2017-07-16)
+------------------
+
+1.3.0 (2017-07-14)
+------------------
+* convert to autoware_msgs
+* Contributors: YamatoAndo
+
+1.2.0 (2017-06-07)
+------------------
+* fix circular-dependency
+* Contributors: Shohei Fujii
+
+1.1.2 (2017-02-27 23:10)
+------------------------
+
+1.1.1 (2017-02-27 22:25)
+------------------------
+
+1.1.0 (2017-02-24)
+------------------
+
+1.0.1 (2017-01-14)
+------------------
+
+1.0.0 (2016-12-22)
+------------------
+* Remove needless compiling flags
+* Fix CUDA_ERROR_LAUNCH_FAILED error occurred in dpm_ttic
+  * Correct memory allocate & copy size
+  * Correct address calculation on device side
+* Improve cmake CUDA configuration
+  Use cmake CUDA feature as possible for removing absolute paths.
+* Use c++11 option instead of c++0x
+  We can use newer compilers which support 'c++11' option
+* Initial commit for public release
+* Contributors: Manato Hirabayashi, Shinpei Kato, Syohei YOSHIDA

ros/src/computing/perception/detection/lib/image/dpm_ttic/package.xml

@@ -1,7 +1,7 @@
 <?xml version="1.0"?>
 <package>
   <name>libdpm_ttic</name>
-  <version>0.0.0</version>
+  <version>1.4.0</version>
   <description>The dpm_ttic package</description>
   <maintainer email="[email protected]">Syohei YOSHIDA</maintainer>
   <license>BSD</license>