peekabot—3-D Robotic Visualization

According to the peekabot project's Web site:

"peekabot is a distributed real-time 3-D visualization tool for robotics researchers and developers, written in C++. Its purpose is to simplify the visualization needs faced by a roboticist daily—using visualization as a debugging aid or making fancy slides for a presentation, for example.

Our goal is to provide a flexible tool that will cater to the vast majority of a roboticist's visualization needs, yet is easy to use. Typical scenarios include visualization of simulations, data display from real robots and monitoring of remotely deployed robots.

...to enable remote data visualization, peekabot uses a distributed client-server architecture. All the gory details of networking is handled by the client library, used by your programs."

peekabot's low-level control of actions allows for some very advanced scripting, such as the object pathing shown here.

A number of Blender-created models allow for some snazzy active objects, available freely on the Web.

Installation

Head to the Web site, and grab the latest tarball. In terms of library requirements, the documentation helpfully states the following (note, in the list below: *not required when building only the client API; **needed only if building the unit tests, which are disabled by default):

• A decently recent version of GCC.
• Boost 1.34.0+ (Boost.Thread, Boost.DateTime, Boost.Filesystem*, Boost.ProgramOptions* and Boost.Test**).
• Xerces-C++ 2.7.0+*.
• FLTK 1.1.6+*.
• OpenGL*.
• GLU*.
• libpng*.

I also had to grab these development files: libxerces-c2-dev and libfltk1.1-dev. Once you download the tarball, open a terminal wherever it's saved and enter the following:

$ tar xvzpf peekabot-x.y.z.tar.gz 
$ cd peekabot-x.y.z
$ ./configure --prefix=/usr
$ make
$ sudo make install

Assuming all went well, when the compilation has finished, you can run the program with the command:

$ peekabot

Before we jump in, I have to warn you that we've covered only half the equation. peekabot is made of two key parts: the server and its clients. After the initial building process, you will have the server by itself. The server is the main GUI screen where you'll be testing and interacting with your client programs. The clients generally will be standalone programs that communicate with the peekabot server while following their own coding structure. Although this may be daunting for new users (me included), it does make the system very open, powerful and flexible.

Okay, I'll assume you have the server window open and are ready to take peekabot for a spin. Let's take a look at a working example program and explore the GUI while it's running.

Open another terminal in the peekabot source directory, and look under the examples folder. Here you will see the folders bo-slam, results and skeleton. Enter any of these folders, and run the command make. After make has compiled each example's code, a new program will be available in the same directory. To use bo-slam as an example, here's the terminal commands to enter (we'll assume you have a folder open in the peekabot main folder):

$ cd examples
$ cd bo-slam
$ make
$ ./bo-slam

Run by itself, nothing will happen, and you will get a bunch of error messages in the console. However, when started with the peekabot server running alongside, a robot and a bunch of pylons will appear, with the pylons moving themselves around the world into position and the robot making its way around each pylon on a preprogrammed path. Okay, now that we have something running, let's explore the GUI.

The main window contains the scene you'll be working with, along with a camera whose point of view is adjustable. The left mouse button will pan the camera, and the right button will rotate it. To zoom in and out, use either the middle mouse button or the mouse wheel. If you need the camera controls to be more or less sensitive, the Shift and Ctrl keys will modulate the sensitivity accordingly (useful in extreme close-ups or when looking from very far away, for example).

On the right is the Tree Browser, which contains all the active objects, scenery elements and so on. You can select objects in view by left-clicking, and multiple objects can be combined selectively like any file manager, using Ctrl to toggle select and Shift to add to the selection.

I've covered only the basic operations of this program here, because you'll need to do some actual coding to get into the nitty-gritty of this program. And quite frankly, I'm rubbish at coding! If you're interested in learning more, check out the basic documentation (manual: www.peekabot.org/doc/latest/manual_the_basics.html; models: sourceforge.net/apps/mediawiki/peekabot/index.php?title=Model_repository). There you will find the coding to get you started, along with its structure and so on.

Ultimately this looks to be a powerful project for robotic visualization, albeit a rather intimidating one. Despite the relatively difficult learning streak from its mostly coding-based interface, it's probably this same kind of interaction that will bring it longevity. Not having the restrictions of GUI design to hold back the mechanics surely will be a godsend to those who want to approach their mechanical design at a low level without the restrictions that accommodating to beginners so often imposes.

peekabot—3-D Robotic Visualization (www.peekabot.org)