Python based Robot Interactive Development Environment (PyRIDE) for ROS/PR2

Introduction

Python based Robot Interactive Development Environment (PyRIDE) is a middleware that provides a self-contained development environment for rapid interactive programming of robot skills and behaviours using Python scripting language (see Figure 1). PyRIDE functions as a software integration tool that aggregates disparate robot software modules and exposes their functionalities to an embedded Python interpreter engine through a unified programming interface. Robot programmers can access the embedded Python engine and the unified robot function programming interface using a remote interactive shell service. One can code, perform experiments/test, debug programs interactively in the same way as using the standard Python interactive interpreter. Programs can transit seamlessly from development to deployment through a simple bootstrap mechanism. PyRIDE also provides remote user level accesses of the robot functionalities, e.g. real-time robot camera image data, through a client-server mechanism. You can view a live demonstration video here.

Figure 1. PyRIDE System Architecture.

Finally, PyRIDE is written in portable C++ and can be ported to various robot platform with relative ease. Currently, PyRIDE runs on NAO, ROS/PR2 and iOS/ROMO platforms. This repository contains PyRIDE source code for the ROS/PR2 platform.

Compile source code

Prerequisites

PyRIDE on ROS/PR2 uses the standard catkin build system. It requires a full PR2 Hydro installation on a Ubuntu Linux system. In particular, make sure you have the following packages installed on your system:

• ros-hydro-moveit-resources
• ros-hydro-sound-play
• libccrtp-dev

You need to set up a catkin workspace and check out the latest code from this repository into the source directory of a working catkin workspace. Then, do:

catkin_make pyride_pr2
catkin_make install

PyRIDE should be compiled and installed under the catkin install directory along with all necessary dependent libraries.

Generate PyRIDE on PR2 API documentation

You can use doxygen to generate the latest API documentation of PyRIDE on PR2. Simply run:

doxygen pyride.dox

under the pyride_pr2 directory. The API documentation will be generated under doc sub-directory. NOTE: The API documentation is incomplete at this stage. You can find the latest doxygen generated documentation at http://uts-magic-lab.github.io/pyride_pr2

Launch PyRIDE on PR2

Make sure you have started up a full PR2 simulation or on a PR2 robot and sourced appropriate catkin overlay. Do:

roslaunch pyride_pr2 pyride.launch

NOTE: fully functioning PyRIDE depends on several subsystems such as MoveIt!, 2D navigation stack. If you want to use functionalities related to these subsystems, make sure these subsystems are operating before launching PyRIDE.

Access embedded interactive Python console

PyRIDE contains a fully functional Python intepreter. One can access its interactive remote shell via

telnet host 27005

where host is the robot/machine PyRIDE is running on. Figure 2 shows the telnet console displays a list of built-in methods available to PyRIDE on PR2.

Figure 2. PyRIDE Remote Shell Access.

NOTE: You can turn off the remote shell access by setting RemotePythonAccess tag to disable in pyrideconfig.xml configuration file.

Load user-developed Python scripts

PyRIDE on PR2 can automatically load and execute Python scripts on its start up. It searches for a py_main.py under a predefined script directory and executes its main() function. Check scripts/py_main.py under this repository for further details. A working application example that receives and sends messages to a twitter account is also provided.

NOTE: You can specify the directory where you want PyRIDE load Python scripts under script_dir parameter in the pyride.launch file.

Remote client access

One of the key components in PyRIDE is the client server subsystem that offers remote user level access of robot system using PyRIDE. This facility differs from the remote shell access that supports interactive software development. PyRIDE clients allow remote monitoring and control of a robot system developed under PyRIDE software framework. PyRIDE clients should be custom developed according to your own requirements. We provide the client SDK as a separate package. Check https://github.com/uts-magic-lab/pyride_clients for relevant code library and documentation on how to develop custom PyRIDE remote clients. The following subsections describe necessary steps for a remote client to interact with PyRIDE.

User authentication

Similar to other client/server based systems, PyRIDE requires user authentication from remote clients in order to establish communication channels. That is, a user account must be created on the PyRIDE 'server', client must supply the correct password to the server to set up the connection. NOTE: PyRIDE uses a password based authentication system, i.e. each account is associated with a unique password, PyRIDE identifies remote user using the provided password. To created a user account, use PyPR2.addUser method. All user account information is saved in an encrypted form based on SHA256.

Client/Server data communication security

All data communications with client and server are based on a PyRIDE customised messaging protocol on the top of TCP/IP stack. The communication is fully encrypted with a symmetrical blowfish cipher. If the data security is important to you, it is highly recommended that you change the default cipher key (see PyRideCommon.cpp line 30 for details) on both server and client code, recompiling your own binaries.

Remote control commands from client

PyRIDE client can issue command string to PyRIDE server to control the behaviour of the robot system. All commands from a PyRIDE client is routed through PyPR2.onRemoteCommand callback function in the embeded Python engine. You need to write the callback handler function attach to PyPR2.onRemoteCommand to process commands from clients. A remote client command consists of a command type (a integer value) and a free form command argument text string.

As PyRIDE supports simultaneous connections from multiple clients, a client must take exclusive control of the PyRIDE system (see client documentation) before it can issue commands to PyRIDE. Otherwise, all commands from the client will be silently dropped and PyPR2.onRemoteCommand callback will not be invoked.

NOTE: Scripts running on the PyRIDE embeded Python engine can override exclusive control taken by a remote client at anytime by calling PyPR2.blockRemoteExclusiveControl method.

Logging system

PyRIDE uses its own logging system. You can find the log output under .ros/pyride_pr2*.log.

Choose a suitable Inverse Kinematic engine

PyRIDE natively supports MoveIt! as well as S-PR2/Magiks. You can choose either Inverse Kinematic package to drive PR2 arms in the task space (with odometry_combined as the default reference frame). To support adoption and further development of S-PR2/Magiks, PyRIDE selects S-PR2 as its default Inverse Kinematic engine. However, you still need to manually install all required dependencies and clone the entire Magiks repository under the script directory before running PyRIDE. Since S-PR2 runs entirely under PyRIDE. You have direct access to its advanced features and the vast array of library functions within the package through PyRIDE remote console.

Currently, PyRIDE provides only basic level support for MoveIt! and methods relate to MoveIt! have not been fully tested. To enable MoveIt! under PyRIDE, you need to do the following steps:

1. Set use_move_it parameter in the pyride.launch to true.
2. Launch MoveIt! before running PyRIDE.
3. Call py_main.iksResolver.useMoveIt() method in PyRIDE.

Note that regardless which inverse kinematic engine is selected, you can move a PR2 arm use the same PyPR2.moveArmTo method. Check API documentation for further details. Another important issue you need to keep in mind is that the reference frames used in MoveIt! and S-PR2 are slightly different. You cannot switch between MoveIt! and S-PR2 transparently without adjust your target position and orientation calculations. We will improve the wrapper functions so that you can transit between the two seamlessly in the future.

Third-party ROS node integration with PyRIDE

PyRIDE for PR2 already provides some of the most commonly used functionalities on PR2 by integrating the respective ROS nodes into the PyRIDE framework. You may want to integrate additional third party ROS node with PyRIDE. PyRIDE supports two approaches for ROS node integration:

1. "Deep" integration: This approach requires modifications of PyRIDE source code. PR2ProxyManager class is the main class that is responsible for communicating with foreign ROS nodes and providing C++ based wrapper functions to interface with these node. You should create necessary subscriber or publishers to the node in the initWithNodeHandle method and the corresponding clean up code in the fini method. You should write necessary callback functions for your subscriber and command methods for your publisher (or action client) as methods of PR2ProxyManager. PyPR2Module class provides wrapper functions to expose C++ methods (usually) of PR2ProxyManager class to the embedded Python engine. It uses the standard embedded Python programming approach and APIs. See source code further details and examples. NOTE: Future version of PyRIDE may provide a better designed plugin mechanism for deep ROS node integration.

2. "Shallow" integration: This approach represents a simpler way of sending messages from an external ROS node to PyRIDE. PyRIDE constantly listens on a /pyride_pr2/node_status topic. A ROS node can publish NodeStatus (see msg/NodeStatus.msg for message structure) messages to this topic, PyRIDE will automatically forward the received messages to PyPR2.onNodeStatusUpdate Python callback function in the embedded Python environment. See py_main.py for further details on this callback function.

Limited use of rospy under PyRIDE environment

In ideal circumstance, PyRIDE should be used as a full replacement for rospy with custom integration of third party ROS modules. However, it is possible to use rospy in a limited way under PyRIDE to publish messages to other ROS node. Use the following example code snippet to create an addon node and publish messages.

import sys
sys.path.append('/opt/ros/hydro/lib/python2.7/dist-packages') #add rospy path into the system path
import rospy
from std_msgs.msg import String
rospy.init_node('pyride_talker',disable_signals=True)
pub = rospy.Publisher( 'pyride_msg', String, queue_size = 5 )
pub.publish('this is a test message')

NOTE: that you should not use rospy.spin() under PyRIDE.