AE740_FINAL_PROJECT

Final Project

Installation

Tested on Ubuntu 18.04, using Ros Melodic.

Create a workspace:

~/$ cd
~/$ mkdir -p catkin_ws/src

Now clone the entire repo into the src folder:

~/$ cd catkin_ws/src/
~/catkin_ws/src$ git clone <repo url>

This should contain a copy of all/most of the dependencies we need, so hopefully we dont need to run extra installation processes.

From the workspace folder, we just need to build the folder:

~/catkin_ws$ catkin build -j8

and now hopefully all the dependencies can be built.

In addition, you will need:

sudo apt-get install ros-melodic-octomap-rviz-plugins

and you might need to follow the installation instructions for OMPL, FCL, and Octomap (no octovis needed). Essentially you need to do the following:

cd dependencies/<package>
mkdir build
cd build
cmake ..
make -j8
sudo make install

If you have issues with nlopt not building, go into catkin_ws/build/nlopt and sudo chmod +x make_install_nlopt.sh

Testing rotorS

using the basic

roslaunch rotors_gazebo mav_hovering_example.launch mav_name:=firefly world_name:=basic

If you have issues with VMware, when launching gazebo:

export SVGA_VGPU10=0

might help

Testing Trajectory Generation

In separate terminal windows

roscore

Generate the waypoints:

rosrun trajectory_generation test_trajectory_generation_node

Generate the trajectory:

rosrun trajectory_generation trajectory_generation_node 

Expected output:

[ INFO] [1617205726.876506837]: Starting Trajectory Generation Node
Received poseArray 
Received 17 desired Waypoints - generating trajectory
[ INFO] [1617205727.083034159]: Generated optimized trajectory from 18 waypoints

To see the pose array, we additionally need to run

rosrun tf static_transrm_publisher 0 0 0 0 0 0 1 map /world 10

(or maybe rosrun tf static_transform_publisher 0 0 0 0 0 0 1 map /world 10 ) `` and then

rviz

and then add the topic /desired_trajectory_waypoints

Testing the combined cart + drone

In separate terminals:

Start the drone:

roslaunch rotors_gazebo mav_hovering_example.launch mav_name:=firefly world_name:=basic

Spawn the cart:

roslaunch vehicle_description spawn_cart_holonomic.launch 

in this terminal pressing 1 + enter makes the cart move forward

Launch the trajectory generation node:

rosrun trajectory_generation trajectory_generation_node 

Testing the landing sequence

roslaunch interface landing_test.launch

Testing the octomap generation

roslaunch interface landing_test.launch

rosservice call /world/get_octomap '{bounding_box_origin: {x: 5, y: 0, z: 1.75}, bounding_box_lengths: {x: 20.8, y: 20.8, z: 4}, leaf_size: 0.1, filename: output_filename.bt}'

reference: https://github.com/ethz-asl/rotors_simulator/wiki/Generate-an-octomap-from-your-world

rosservice call /world/get_octomap '{bounding_box_origin: {x: 30, y: 0, z: 5}, bounding_box_lengths: {x: 70, y: 50, z: 12}, leaf_size: 0.1, filename: /home/devansh/vna2v_project_ws/src/AE740_FINAL_PROJECT/path_planning/assets/road_2.bt}'

to visualize make sure octomap rviz plugins is installed, run:

rosservice call /world/get_octomap '{bounding_box_origin: {x: 5, y: 0, z: 1.75}, bounding_box_lengths: {x: 20.8, y: 20.8, z: 4}, leaf_size: 0.1, publish_octomap: true}'

and in rviz you should be able to see a topic /world/octomap dont forget to change the referernce frame from map to world

Live generation of octomap:

roslaunch interface landing_test.launch

rosrun path_planning generate_octomap

this can be visualized in the rviz

first goal position:

(5.0, 7.75, 0.75)

Teleop

roslaunch interface landing_test.launch world_name:=outdoor
rosrun teleop_twist_keyboard teleop_twist_keyboard.py
rosrun vehicle_description move_van_holonomic

Testing aruco markers:

roslaunch interface landing_test.launch

navigate somewhere nice and rviz

roslaunch aruco_ros single.launch

now launch something that requires an output from aruco detection (detection analysis stops otherwise) and if the marker is detected, the pose will be published to /aruco_single/pose

Testing landing with driving truck

roslaunch interface landing_test.launch

in a separate terminal

rosrun ackermann_drive_teleop keyop.py

this allows you to drive using arrow keys, stop with spacebar and reset speed and steering using tab

Once the drone is launched you can tell it what to do:

rostopic pub /changeState

press tab enough times for autocomplete. Set the integer to:

1 - hover at current state
2 - plan RRT and visual land on van
5 - plan RRT and arrive at package drop location

To stop motors:

rostopic pub /firefly/arm std_msgs/Bool "data: false" 

To start motors:

rostopic pub /firefly/arm std_msgs/Bool "data: true"

the motors will start again when either the above true is sent or when any deisred pose is sent

Testing the full sequence

roslaunch interface landing_test.launch

Todo:

1. Fix RRT/min snap collision avoidance
2. visual landing
3. add second drone
4. spawn drone on van rather than somewhere else
5. generate better octomaps
6. replan rrt if doesnt end at goal