Embedded Motion Control/Tutorials/Setting up the PICO simulator 2015

Introduction

During the course, we have 10 large groups and only one robot, so test time on the robot is scarce. Fortunately, we have a virtual (software) representation of the robot that can be used to simulate the robot. At the moment it:

• Simulates the movement of the robot
• The laser data of the robot, created by the virtual environment
• Provide a simple visualization

This should already be enough to get you started, and more will be added later (odometry, moving doors, better dynamics, etc).

Updating the simulator

Before you start working with the simulator, make sure you have our latest version by running

emc-update

This will download the latest changes and compile the updated software (framework and simulator). I will notify you when I made changes to the software such that you can run the updater, but feel free to run the command whenever you want.

Starting the simulator

As was already stated before, we will not be using ROS in this course, unless you really want to use it yourself. However, secretly the provided tools are build on top of that; the inter-process communication to be more specific. Don't worry about it too much. The only thing you need to know is that before running the simulator or your software, you need to create a roscore. Simply open a terminal an run:

roscore

and keep it running. This allows processes to 'find' each other and communicate. For example, the software abstraction layer we provide 'talks' to the simulator through this roscore.

Enough about ROS, let's start the simulator! Open a terminal and run

emc-sim

That's it! What you will see, is a visualization of the virtual environment (the white lines are walls) and a top-down view of the robot (the red lines).

Controlling the robot

Now, as you can see, not much is happening. The robot is standing still in a static environment. Let's change that! A first simple way to test the simulator is by controlling the robot using your keyboard. Just run:

pico-teleop

and you will be able to control the robot using the numpad keys. You can rotate the robot using '/' and '*'.

Visualization

As you already noticed, the simulator pops up a visualization windows showing the robot in the virtual world. This shows how the world is (at least in simulation...). It is also very interesting to know how to robot perceives the world through its sensors. Note that this is a big difference! We also created a visualization for that. Just run

emc-viz

You will again see the robot footprint displayed in red. However, the walls are gone and replaced by the actual (simulated) sensor data from the robot (the green dots). The laser data is actually represented in polar coordinates (remember, the LRF is a laser with a rotating mirror, so what it will return are angles and distances) but the visualization 'projects' them back to world coordinates. See how the laser data changes even if the robot is standing still: this is the simulated noise added to the data.

Try driving the robot around using 'pico-teleop' and notice how the laser data changes, and how it differs from the actual virtual world.

Stopping the simulator and visualization

You can stop the simulator and visualization by pressing ctrl-C in the terminal.

Loading a custom heightmap

By default, the simulator loads a pretty ugly maze. Fortunately, you can change the simulation environment very easily! You just have to create a heightmap: an image that specifies for each pixel how 'high' the world has to be at that point. Since the laser only scans at one height, we can use two extremes here: flat (ground level) or high enough to be detected by the laser.

To create an image, simply open your favourite image editor and create a black-and-white image. If you don't know how to create an image, have a look at the section below. White corresponds to the floor, black to the walls (i.e., which will be detected by the laser). You have to keep the following things in mind:

• The simulator converts the pixels to world coordinates. Each pixel corresponds to a block of 2.5 by 2.5 cm, or in other words, 40 pixels correspond to 1 meter.
• The robot always starts in the center of the image. So, if you want to robot to start at the edge of your maze / corridor, just create a bigger image and move the black pixels to the upper part of the image.
• You'll get the best result if the lines drawn are at least 2 pixels wide
• Store your image lossless, i.e. using the png format (which is recommended by the way!), instead of the jpg format.
• Make sure your heightmap images always have a border of white pixels, i.e. there should not be any non-white pixels at the borders of your image

Once you have created an image, simply run the simulator and provide the image file as argument:

emc-sim <YOUR_IMAGE_FILE>.png

That's it!

Create a heigthmap image

There are many linux applications that you use to create images. We suggest using Gimp, an open-source alternative to Photoshop. Although it might be a bit overkill to use for our application, it has great support and the thing we want to do (draw black lines on a white background) isn't hard. To install Gimp, run:

sudo apt-get install gimp

And run it using:

gimp

Then to create a simple image:

• Select File -> New (or ctrl-N)
• Specify the size of your image. Remember, 40 pixels = 1 meter, and the robot starts in the center
• Select the Pencil Tool from the left pane.
• Set the pencil size in the lower left pane to something sensible, e.g. 2 pixels

Now, if you click left on the image, a dot is drawn, but we want lines! To draw a line:

• left click, then hold shift, then left click again.
• While holding the shift button, you can click more times to create a sequence of lines
• To create nice, axis-aligned lines, also hold ctrl

There is two types of saving in Gimp. The first one is the using File -> Save. However, this will only generate an xcf-file, something that can only be read by Gimp. Instead, you should use the File -> Export option:

• File -> Export
• Provide a name for your file. If you put '.png' as extension, it will be saved as png
• Use the default png export settings

That's it!