GzWeb is usually installed on an Ubuntu server. Once the server is set up and running, clients can interact with the simulation simply by accessing the server's URL on a web browser.
The main dependencies for GzWeb are the Gazebo development libraries, version 9 or greater, and NodeJS version 6 or greater.
Take a look at these tutorials to choose the Gazebo installation that best fits your case. The simplest approach would be to install Gazebo 9 as follows:
sudo apt install gazebo9 libgazebo9-dev
Run the following to install dependencies:
sudo apt install libjansson-dev libboost-dev imagemagick libtinyxml-dev mercurial cmake build-essential
Next install nodejs
and npm
using node's version manager nvm
:
# install nvm
curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.35.3/install.sh | bash
# source .bashrc so we can use the nvm cmd
source ~/.bashrc
# install node. Supported versions are 8 to 11.
nvm install 8
You may run into conflict with the libssl version needed by Gazebo and nodejs when trying to install using
apt
on Ubuntu. So the recommended way of installation is to usenvm
.
Clone the repository into a directory in your home folder for example:
cd ~; git clone https://github.com/osrf/gzweb
Enter the GzWeb repository and switch to the latest release branch:
cd ~/gzweb
git checkout gzweb_1.4.1
The first time you build, you'll need to gather all the Gazebo models which you want to simulate in the right directory ('http/client/assets') and prepare them for the web.
Before running the deploy script, it's important to source the Gazebo
setup.sh
file:
If you installed gazebo via deb packages:
source /usr/share/gazebo/setup.sh
If you did a source install then:
source <YOUR_GAZEBO_PATH>/share/gazebo/setup.sh
Run the deploy script, this downloads models from the web and may take a couple of minutes, see more options below.
npm run deploy --- -m
Note: the
-m
flag tells the deploy script to grab all the models from the model database and any other models in yourGAZEBO_MODEL_PATH
. For all subsequent builds, the-m
flag will not be needed.
To skip downloading models from the model database and grab only local models in your Gazebo model path, do:
npm run deploy --- -m local
To generate thumbnails for all the models , run the script with the -t
flag, i.e.:
npm run deploy --- -t
Note: This spins up a
gzserver
with a camera for capturing screenshots of models. So make sure there is rendering support and no background gzerver process running (or set a differentGAZEBO_MASTER_URI
in the terminal).
If you'll use GzWeb on mobile devices, you can create coarse versions of all
models, which are lighter to load (50% of original quality). If generated,
these meshes will automatically be used on mobile devices. If you've already
ran npm run deploy --- -m
, run just:
npm run deploy --- -c
Or you can run both flags at the same time to generate coarse versions as you create the database:
npm run deploy --- -m -c
You also have the option to pick specific models and how much percent to coarsen, running:
./coarse_meshes.sh [percent] [path]
Here, [percent]
is the edges ratio with respect to the original mesh
(0 to 100), and [path]
is the path of a model. For example:
./coarse_meshes.sh 20 http/client/assets/bowl/
Running GzWeb involves the following pieces:
gzserver
running the headless Gazebo simulation (runs by default on
http://127.0.0.1:11345)
GzWeb's NodeJS server which communicates with gzserver
using
Gazebo Transport.
It works as a bridge between the Javascript and the C++ code.
An HTTP server which serves static content such as models and website assets (icons, HTML, CSS, client-side Javascript...)
A Websocket server which forwards simulation updates coming from gzserver
to the browser
A browser client which connects to the HTTP and websocket servers
Start them as follows:
On the server machine, start gazebo
or gzserver
first, it's recommended
to run in verbose mode so you see debug messages:
gzserver --verbose
Tip: see the port where the Gazebo master is communicating, such as
[Msg] Connected to gazebo master @ http://127.0.0.1:11345
On another terminal, from your GzWeb directory, run the following command to start both the HTTP and Websocket servers:
npm start
Tip: You can use the
-p
option to choose an arbitrary port, for example:npm start -p 1234
. By default, it serves on port 8080.
Open a browser that has WebGL and websocket support (i.e. most modern browsers) and point it to the IP address and port where the HTTP server is started, for example:
http://localhost:8080
To stop gzserver
or the GzWeb servers, just press Ctrl+C
in their terminals.
Q: When installing node package modules, I see errors along the lines of:
npm ERR! Error: failed to fetch from registry: node-gyp
A: Try setting the npm registry first then install the modules again.
npm config set registry http://registry.npmjs.org/
Q: When installing websocket, I see errors along the lines of:
sh: 1: node: not found
npm ERR! error installing websocket@1.0.8
npm WARN This failure might be due to the use of legacy binary "node"
Or along the lines of:
/usr/bin/env: node: No such file or directory
There are node-gyp build errors, exiting.
A: In Debian systems, the binary file "node" has been renamed to "nodejs" to avoid a name conflict. Try adding a symlink to the correct name:
sudo ln -s /usr/bin/nodejs /usr/bin/node
You may also find that your repository is too old and you should just install recent versions of node and npm directly.
Q: When running npm run deploy ---
, I see errors along the lines of:
gyp ERR! configure error
A: There might be a conflict between the gyp version installed and the gyp version in node-gyp. Try removing gyp:
sudo apt-get remove gyp
Q: When running npm run deploy ---
, I have problems finding GTS, like this:
~/gzweb/tools/gzcoarse.cc:18:17: fatal error: gts.h : no such file or directory, #include <gts.h>
A: It seems that your Gazebo installation didn't install GTS headers. Try installing them manually:
sudo apt-get install libgts-dev