ducksim

ducksim is a robot simulator inspired by turtlesim. Each robot can pickup and drop objects in the environment. Robots are coordinated to collect all of the objects and take them to a goal. The simulation is accomplished using a custom physics simulator. The progress of the simulation can be viewed in Rviz.

This simulator was created as a part of a final project for CS 4287: Cloud Computing at Vanderbilt University in Spring 2021. It includes the necessary Docker and Kubernetes files to demonstrate how a ROS application can be distributed across Cloud and local infrastructure.

Step 1: Install Ducksim locally

Before you can simulate a local robot or run Rviz, you must first install ROS and then build the ducksim package from source. This is the same process as setting up a real robot. The package was tested with ROS Melodic but should also work with ROS Noetic.

Build ducksim from source

# Create a workspace and clone the code
mkdir -p ~/catkin_ws/src
cd ~/catkin_ws
git clone https://github.com/jeholliday/ducksim.git src/ducksim
# Build and install the code
catkin_make install

Step 2: Deploy and Configure Cloud Infrastructure

Install Tailscale on Cloud VMs

You will need to change the advertised routes to match the pod CIDRs of your cluster. Advertised routes allow any other device on the Tailscale network to access the internal pod network. Tailscale needs to be installed on each cloud VM, so that connections to the pods running on that VM from local robots can be directly routed to the machine instead of through the overlay network.

# Install Tailscale
# https://tailscale.com/download/linux

# Start Tailscale
echo 'net.ipv4.ip_forward = 1' | sudo tee -a /etc/sysctl.conf
echo 'net.ipv6.conf.all.forwarding = 1' | sudo tee -a /etc/sysctl.conf
sudo sysctl -p /etc/sysctl.conf
sudo tailscale up --advertise-routes=10.2.0.0/16

# Allow the route in the Tailscale admin dashboard
# https://tailscale.com/kb/1019/subnets/

Start a Docker registry on one cloud VM

# Start a insecure docker registry
docker run -d -p 5000:5000 --restart=always --name registry registry:2

Build Docker Images

Change the IP address to that of your docker registry

sudo docker build --build-arg DUCKSIM_VER=$(date +%Y%m%d-%H%M%S) -t ros .
sudo docker tag ros:latest <docker_registry>:5000/ros
sudo docker push <docker_registry>:5000/ros

Start Kubernetes Pods

In each of the provided jobs, you will need to replace the IP address with where ROS master can be found.

kubectl apply -f infrastructure/kubernetes/couchdb-job.yaml
kubectl apply -f infrastructure/kubernetes/roscore-job.yaml

# Wait for CouchDB and roscore to start

kubectl apply -f infrastructure/kubernetes/couchdb-logger-job.yaml
kubectl apply -f infrastructure/kubernetes/simulation-job.yaml
kubectl apply -f infrastructure/kubernetes/static-transform-job.yaml
kubectl apply -f infrastructure/kubernetes/herder-job.yaml

Step 3: Start a Local Robot

Configure Local Robot

Notice the --accept-routes flag, which is necessary on Linux to accept the routes to the pod network published by the Cloud VMs.

# Install Tailscale
# https://tailscale.com/download/linux

# Start Tailscale
sudo tailscale up --accept-routes

Run a simulated robot

A simulated robot can be run locally and interact with the other applications in the cloud.

source ~/catkin_ws/install/setup.bash
# Set ROS_IP to address of tailscale interface
export ROS_IP=`ip a show dev tailscale0 | grep 'inet ' | awk '{print $2}' | cut -d/ -f1`
export ROS_MASTER_URI=http://<master_ip>:11311
roslaunch ducksim duck_nav.launch robot_name:=<robot_name>

Run Rviz visuzlization

Rviz is a tool for visualizing all kinds of data. The provided configuration file can configure Rviz to visualize the current state of the robot simulation.

source ~/catkin_ws/install/setup.bash
export ROS_IP=`ip a show dev tailscale0 | grep 'inet ' | awk '{print $2}' | cut -d/ -f1`
export ROS_MASTER_URI=http://<master_ip>:11311
rviz -d ~/catkin_ws/install/share/ducksim/rviz/turtlesim_demo.rviz

View Logs

The logs from each node can be found by looking in the CouchDB admin panel at: http://<master_ip>:30006/_utils

Step 4: Teardown Cloud Infrastructure

Stop Kubernetes Pods

kubectl delete -f infrastructure/kubernetes/herder-job.yaml
kubectl delete -f infrastructure/kubernetes/static-transform-job.yaml
kubectl delete -f infrastructure/kubernetes/simulation-job.yaml
kubectl delete -f infrastructure/kubernetes/couchdb-logger-job.yaml
kubectl delete -f infrastructure/kubernetes/roscore-job.yaml
kubectl delete -f infrastructure/kubernetes/couchdb-job.yaml

jeholliday / ducksim Goto Github PK

ducksim's Introduction