![Wil Selby](/img/default-banner.jpg)
- Видео 19
- Просмотров 124 150
Wil Selby
Добавлен 18 июл 2011
This Channel features videos showing my research into control systems and mechatronics specifically focusing on autonomous unmanned aerial vehicles. These videos comprise my research spanning from the US Naval Academy, MIT, and other experiments.
DIY Driverless Car With Computer Vision (ROS, RaspberryPi, Elegoo)
DIY Driverless Car With Computer Vision (ROS, RaspberryPi, Elegoo)
09 August 2017
www.wilselby.com/diy-autonomous-vehicle-project/
Wil Selby
This project focused on developing a lane identification and tracking algorithm running on a RaspberryPi 3 that would enable the autonomous navigation of a modified Elegoo vehicle. The system was developing using ROS and Gazebo was used as the simulation environment. This video tutorial briefly covers the following topics:
- A brief overview of the required hardware and software
- Development and implementation of a drive-by-wire system for manual control of the vehicle in Gazebo and in the real world.
- Development of a lane detection and tracking algorith...
09 August 2017
www.wilselby.com/diy-autonomous-vehicle-project/
Wil Selby
This project focused on developing a lane identification and tracking algorithm running on a RaspberryPi 3 that would enable the autonomous navigation of a modified Elegoo vehicle. The system was developing using ROS and Gazebo was used as the simulation environment. This video tutorial briefly covers the following topics:
- A brief overview of the required hardware and software
- Development and implementation of a drive-by-wire system for manual control of the vehicle in Gazebo and in the real world.
- Development of a lane detection and tracking algorith...
Просмотров: 5 239
Видео
FormulaPi Basic Lane Tracking (ROS & Gazebo)
Просмотров 2,3 тыс.7 лет назад
FormulaPi Basic Lane Tracking (ROS & Gazebo) 15 January 2017 www.wilselby.com/research/autonomous-vehicles/formulapi-autonomous-vehicle-racing/ Wil Selby This project focused on developing a simulation environment in ROS and Gazebo to test and develop software for the FormulaPi competition. FormulaPi is an autonomous vehicle racing competition hosted by PiBorg using Yetiborg vehicles. This deve...
Quadrotor 2D Mapping and Navigation (ROS Navigation Stack & Gazebo)
Просмотров 4,7 тыс.8 лет назад
Quadrotor 2D Mapping and Navigation (ROS & Gazebo) 8 May 2016 www.wilselby.com/research/ros-integration/ Wil Selby This project focused on simulating a Kinect sensor and a quadrotor to develop a 2D map of the environment and subsequently using the map for localization, path planning, and obstacle avoidance. This video tutorial briefly covers the following topics: - Manipulating the Kinect senso...
Quadrotor Control Systems Design (ROS & Gazebo)
Просмотров 14 тыс.8 лет назад
Quadrotor Control Systems Design (ROS & Gazebo) 8 May 2016 www.wilselby.com/research/ros-integration/ Wil Selby This project focused on developing attitude and position control systems for a quadrotor utilizing the Robot Operating System (ROS) and Gazebo simulation environment. This video tutorial briefly covers the following topics: - Implementation of an attitude control system on a simulated...
Quadrotor 3D Mapping and Navigation (ROS MoveIt! & Gazebo)
Просмотров 34 тыс.8 лет назад
Quadrotor 3D Mapping and Navigation (ROS & Gazebo) 8 May 2016 www.wilselby.com/research/ros-integration/ Wil Selby This project focused on simulating a Kinect sensor and a quadrotor to develop a 3D map of the environment and subsequently using the map for localization, path planning, and obstacle avoidance. This video tutorial briefly covers the following topics: - General overview of the ROS M...
Custom Quadrotor Autopilot Development
Просмотров 1,9 тыс.8 лет назад
Custom Quadrotor Autopilot Development 18 November 2015 www.wilselby.com/research/arducopter/ Wil Selby The purpose of this project was to develop a basic autopilot utilizing the ArduCopter software libraries, an APM 2.5 board, and a 3DR Quad-C Frame. This video tutorial briefly covers the following topics: -Quadrotor System Modeling describing the coordinate system used to develop the model an...
DRL at MIT Energy Futures Conference, Monaco
Просмотров 569 лет назад
wilselby.com for more information Wil Selby Department of Mechanical Engineering Massachusetts Institute of Technology MIT Energy Futures Conference, Monaco 23-26 September 2010 This video was displayed at the MIT Energy Initiative Futures Conference in Monaco from 23-26 September 2010 at the Oceanographic Museum of Monaco. In addition, a live demonstration of a manually flown quadrotor with re...
ArduCopter Autonomous Photomosaic
Просмотров 80311 лет назад
ArduCopter Autonomous Photomosaic 22 July 2013 www.wilselby.com Wil Selby In this experiment, a small Unmanned Aircraft System (UAS) represented by the ArduCopter is used to create a photomosaic image of a user defined region of interest. This enables a small UAS with a limited imagery payload to act as a wide aerial surveillance platform. This is useful if the target is time sensitive and larg...
ArduCopter Sensor Fusion Utilizing Augmented Reality
Просмотров 69111 лет назад
ArduCopter Sensor Fusion Utilizing Augmented Reality July 15 , 2013 Wil Selby This experiment was a proof on concept demonstration to show the utility of augmented reality. GPS waypoints were used to simulate the location of a point of interest determined by some other sensor type. The points could be pre-programmed or updated in real time depending on the mission. As the GPS location came into...
ArduCopter Derby Circle Waypoint Mission
Просмотров 3,1 тыс.11 лет назад
ArduCopter Derby Circle Waypoint Mission May 19, 2013 Wil Selby wil@wilselby.com This first experiment demonstrated the ability of the ArduCopter to accurately follow pre-programmed GPS waypoints. The ArduCopter was flown manually initially and then switched into auto-mode. Once in auto-mode, the ArduCopter followed the GPS waypoints indicated on the map with green dots. The dotted circles indi...
Autonomous Aerial Navigation and Tracking of Marine Animals
Просмотров 30112 лет назад
www.wilselby.com for more information Autonomous Aerial Navigation and Tracking of Marine Animals Wil Selby Peter Corke Daniela Rus MIT Distributed Robotics Lab Fall 2011 This video compliments our paper at from Australaian Conference on Robotics and Automation 2011 (ACRA). In this paper, we describe the development of an independent and on-board visual servoing system which allows a computatio...
Autonomous Dynamic Object Tracking Without External Localization
Просмотров 2,5 тыс.12 лет назад
www.wilselby.com for more information Autonomous Dynamic Object Tracking Without External Localization MIT Distributed Robotics Lab Spring 2011 In this video we present an autonomous on-board visual navigation and tracking system for an Ascending Technologies Hummingbird quadrotor vehicle to support the whale tracking application independent of external localization. Due to the limited payload ...
Autonomous Dynamic Object Tracking Using GPS Feedback
Просмотров 1,1 тыс.12 лет назад
www.wilselby.com for more information Autonomous Dynamic Object Tracking Using GPS Feedback MIT Distributed Robotics Lab Summer 2011 In this video we present an autonomous on-board visual navigation and tracking system for an Ascending Technologies Pelican quadrotor vehicle to support the whale tracking application utilizing the on-board GPS receiver. Due to the limited payload of the robot, we...
Aerial Tracking of Ground Target Using Motion Capture State Feedback
Просмотров 1,7 тыс.12 лет назад
www.wilselby.com for more information Autonomous Dynamic Object Tracking With Motion Capture State Feedback MIT Distributed Robotics Lab Spring 2011 In this section we describe an autonomous on-board visual navigation and tracking system for an Ascending Technologies Hummingbird quadrotor vehicle to support the whale tracking application. Due to the limited payload of the robot, we are restrict...
Target Identification and Tracking Using Hue-Saturation Histograms
Просмотров 3,1 тыс.13 лет назад
www.wilselby.com for more information Object Identification and Tracking Algorithm Results Wil Selby Peter Corke Daniela Rus MIT Distributed Robotics Lab Fall 2009 Our object recognition algorithm characterizes the target in the Hue (H) and Saturation (S) plane and then identifies the target in subsequent frames based on this characterization. We require a solution that is robust but also compu...
Observing Whales Using Small Aerial Robots
Просмотров 1,8 тыс.13 лет назад
Observing Whales Using Small Aerial Robots
Autonomous Underwater Tracking With AMOUR
Просмотров 87513 лет назад
Autonomous Underwater Tracking With AMOUR
Sliding Mode Control for Translational Trajectory Following for a Quadrotor Vehicle
Просмотров 18 тыс.13 лет назад
Sliding Mode Control for Translational Trajectory Following for a Quadrotor Vehicle
Ball and Beam Balance Using PID Control
Просмотров 27 тыс.13 лет назад
Ball and Beam Balance Using PID Control
Nice
Excuse me Mr., How can i implement the trajectory to the Simulink model ?
Can I create with the slinding_control.m and the paper the simulink model. I saw 4 ways online to generate the model from the matlab code
I think you working now in Tesla
The link provided is not working? Any alternate links? Please help
www.wilselby.com/research/ros-integration/ works for me currently
@@wilselby The url shows the following error: Warning: mysqli_set_charset(): Error executing query in /app/public.built/wp-includes/wp-db.php on line 756 Warning: Cannot modify header information - headers already sent by (output started at /app/public.built/wp-includes/wp-db.php:756) in /app/public.built/wp-includes/pluggable.php on line 1223
Hi can we utilize moveit with a 8 motor holonomic submarine ? Is there any packages for this?
Following, any updates ?
@@nirajpudasaini4450 ended up using nav2 with a pid to control z axis. Version 2 of my system is from scratch tho: first spline interpolation to create poses then following said poses with velocity commands
@@nikolaiphilipenko9210 thanks for such a prompt response. I am trying to work around with moveit and ros melodic for the past 3 months for my bachelors thesis. I am still not able to succesfully execute the path planning. I would appreciate if you can share any resources that you have made public. I will try to switch to nav2 if it works. Fyi: i am using bluerov2 as a robot, and have successfully assembled and tested its built in functionality.
Can you share the non linear dynamic model of the system?
s3-us-west-2.amazonaws.com/selbystorage/wp-content/uploads/2016/05/Selby_Quad_Robust_Ctrl.pdf
how can this made to work on ROS noetic?
brutal
what file exactly did you modified to implement your controller ? and how did you implement your controller via mission planner or arduini IDE?
You can find all the info here: www.wilselby.com/research/arducopter/autopilot-implementation/
@@wilselby i have checked everything, it doesn't tell whether your burn the APM or upload file. Can you please refer to some specific page?
@@adeelehsan2019 ah sorry, just upload the file github.com/wilselby/CustomFC/blob/master/CustomFC.ino to the arduino
@@wilselby Thanks sir! I appreciate it!
@@adeelehsan2019 hey bro did work for you ? would you please share your circuit diagram
Plaz reply me on instagram sir My Instagram ID _ instagram.com/onkar_choudhari_16?r=nametag
Sir I need a help for this simulation
Can we do 3D navigation for a 4 wheeled robot?. We are actual working on that but unable to find any package for it
I think you could use the navigation stack packages for a ground robot.
hey wil can you help me too... i am unable to understand the git thing
Here's a tutorial I wrote related to this video: www.wilselby.com/research/ros-integration/control-system-implementation/
I've installed your quadrocopter simulator but doesn't work with ROS Melodic...
hey did you find a solution ?
Hi will.....I also work on a similar field ....can you share the code you wrote in Matlab ...or at least the algorithm...so that it can help me . And how did you make the program of yours run in the flight controller of your drone.
Here's the MATLAB code: s3-us-west-2.amazonaws.com/selbystorage/Files/sliding_control.m
used materials ? servo motor model?
HELLO SIR , CAN U HELP ME , I WANT TO DEVELOP THIS TECHNOLOGY IN MY PROJECT
Hi Will, very interesting project! just bought the elegoo kit and digging into ROS right now. It seems the link you provide in the comments is currently dead, would you mind checking?
Hey Wil, how did you give the predefined waypoints during 2d mapping??? Did u move it manually or did mapping using joystick??
Hello. I am trying use lidar YDLIDAR G4 to do 3D reconstruction of a room. Could you please assist me how to contruct the code? Really appreciate your help.
A detailed tutorial is available here: www.wilselby.com/research/ros-integration/3d-mapping-navigation
Wil Selby thank you so much for your help. Can we do this based on Arduino?
Hi Sir, i am working on Robotic arm with Moveit i done all work for path planing but i don't know how to create Octamap, may you help me . i even check octamap website but can't get it , so please show me the right way where i can learn or how i can learn .
Check out my detailed tutorial here: www.wilselby.com/research/ros-integration/3d-mapping-navigation The Occupancy Map Updater is the primary component of MoveIt! that deals with 3D perception. You can use the PointCloud Occupancy Map Updater plugin.
Thank you for your video. I have one question: Is the room map build by the Kinect just an object map(other than collision map)? If so, how can you transfer the object map into collision map (which robot will see it as a collision when navigation) ?
The octomap that is created is a collision map. Toward the end of the video, you can see that MoveIt! creates a trajectory around some of the obstacles identified in the ocotmap.
Thank you so much for your reply. I have created a 3D map by point cloud2. However the PR2 arm dosen't consider it as collision object . Do you know the reason?
That I'm not sure of unfortunately, sorry.
Thank you for your reply. I fix it by reference this link docs.ros.org/indigo/api/pr2_moveit_tutorials/html/planning/src/doc/perception_configuration.html
I do something similar by georeferencing but there are some slipping I guess it is because of the delay between the time odometry and point cloud are obtained. How can your map be that precise?
This is in simulation, so there are no errors in any of the measurements.
I also get ground truth odometry from gazebo but I can not synchronize point cloud and odometry messages
I tried Time Synchronizer of message_filters but I get an error which may be related to boost libraries.
Here's a more detailed tutorial which might help: www.wilselby.com/research/ros-integration/3d-mapping-navigation/
hello wil ,please can you send me code of matlab and code of board controller using c. my email is karakhld@gmail.com thank you.
It's hosted here: www.wilselby.com/research/mit-class-projects/
thank you very much willy
i am so sorry for word willy I did not concentrate when I wrote, again thanks wil
Good luck!
Hi, I am really impressed by the video. I am currently using 3 external stereo cameras for obtaining depth and point cloud map of a room. I am also using a Parrot A.R. Drone and wish to do obstacle detection and avoidance and move the drone around. My question: Can this setup you used, work for me as well ? The main difference here is that the point cloud and 3D map are provided by the external cameras in my case.
I think it should work if you can get RGB-D information and pointcloud information from your sensors.
Very excellent work keep ,hi Wil Selby , I am in trouble with SMC matlab simulation, I tuned the adaptive SMC controllers for altitude and roll pitch and they work very good , but when I tried to tuning YAW controller (SMC) it affect the other states z (altitude), littel bit affect phi , theta , (pitch , roll) ??? I checked my quadrotor model many time , I start thinking that my saturation blocks make the problem !! , I don't Know , do you bhave any advice for me ?.
Try replacing the yaw controller with a simple PI or PID controller and see if that helps.
+Wil Selby thanks for replying , i have remplace m'y yaw smc by pid as you Saïd , but nothing change , do you think i have to remove thé saturation blocks ? , i put the saturation block on contrôler output and in the input of motors w1,2,3,4 , or do you think i have to add (m*g) to U1 (MASS * GRAVITY) + OUTPUT OF CONTROLLER OF ALTITUDE ?
U1 should already have m*g in it. If it doesn't then you should add it. I would remove the saturation blocks for now since this is just a simulation and you are debugging.
I will add (m*g) and remove the saturation blocks , thanks a lot
Cool!
could you public the github address?
github.com/wilselby/diy_driverless_car_ROS
@@wilselby could you tell me how to set the navigation?i can only use the teleop to control the vehicle.And the opencv didnot show the image.
@@janesjako6934 Did you verify that the control system is publishing control commands on the right ROS topic?
@@wilselby Hi,Wil,thanks for your answer.I didnot see the control commands.By the way,i still have a question,if i jsut do the pure simulation without phyical vehicle and material,do i have to use the raspberry camera ?I am not sure if i have to use the camera or not ,if it is needed ,it seems like i have to install a ubuntu mate system,which is very complicate ,bec if i install it,all the environment that i install in my ubuntu 16.04 will be gone.
@@janesjako6934 You do not need the phyiscal camera, the ROS Gazebo environment will simulate it for you!
Hello again! I would like to ask how you went about choosing its tuning parameters.. Did you use a research paper, or some formulas, or maybe trial and error? Thanks!
I tested some values in the simulation to start with but then just trial and error on the actual system.
may I know what values you used? For the x,y,z, roll, pitch, and yaw controllers? I've seen your matlab code however i believe you used a different set of values for the actual quadrotor... Thanks! I'm sorry for any inconvenience!
Ronello, Here is the actual file: s3-us-west-2.amazonaws.com/selbystorage/Files/slide_tan.cpp I believe the values were: double sigma[3] = {1.5,1.5,1}; double meu[3] = {2.5,2.5,1}; double lambda[3] = {2,2,10};
Hello! My I ask how you obtained the values of the moments of inertia? Thanks
I just estimated them. See this page for specifics: www.wilselby.com/research/arducopter/model-verification/ It's in the section labeled "Quadrotor Body Moment of Inertia"
excuse me if you are an academic person is there a way for contact? I am from Egypt and want to complete my post graduate study at this field .thank you for your attention
Mohamed, You can find an email for me at www.wilselby.com
Hi Wanted to know if there is anything for 2-D mapping ? Thanks for help.
Check out my 2D mapping and navigation video here: ruclips.net/video/orYNS6ntjnY/видео.html It has links to a tutorial on my website with sample code.
Thanks for help! Another question: Can I use a bumper sensor to bump with everything in a room and then store the coordinates of all those points. Then I know which all point have an obstacle. So can something like this be used to map a room. Thanks again!
Raghav, I don't have any experience doing this but here is the link to a bumper sensor: wiki.ros.org/evarobot_bumper and you can use this package to convert the bumper data to a pointcloud which can then be an input into a costmap for gmapping: wiki.ros.org/kobuki_bumper2pc -Wil
Thanks a lot! sorry for late reply.
Hi, sorry for bothering again but do you know of any Path Planning Algorithm I can use for traversing a room in best possible way with a known boundary. Thanks a lot for help.
sir i am beginner of sliding mode control it is possible to give simulink model and some tutorial for this control.
Rajesh, The Final Project Report is available and the MATLAB code can be found here. Good luck!
how did you positioning?
I used a Vicon motion capture system to get position and orientation information.
and whether autopilot did you use? Pixhawk? did you generate your code by Matlab simulink ? because I'am doing it
The Ascending Technologies quadrotor ran it's own autopilot software. The autopilot inputs were desired roll and pitch angles, yaw rate, and thrust. I wrote a program using the algorithms developed in MATLAB to compute the inputs for the autopilot.
Nice job!!!
Dear Sir: It is really a good simulation and I am really enjoy it. But I do still have problem while importing the project... I have checked the .dae file code of the track should contain the color properties. But why in Gazebo 7 it can't show any of the texture? (SDF version or the gazebo version)?
I don't have Gazebo 7 so I can't verify the issue. Just to confirm you are using the formulapi_track1.world and formulapi_track_1.dae files on the repository here: github.com/wilselby/formulapi_ROS_simulator/tree/master/formulapi_gazebo/worlds Also, try rotating your camera view. Sometimes the colors are altered based on the camera angle.
Thanks Sir. Really benefit from your tutorial I will have a try on gazebo 2.x(?) Also I am re-implement the simulator in a model of turtlebot3 Your tutorial is awesome and I will pay effort for learning from what you have done
I have try in the gazebo 2.x and the map is implemented successfully Thanks sir
Glad to hear it worked out!
hi, i have some research same with you, but i take to the smooth landing, would you share simulation using matlab at first video?
hi if I want to test algorithm for path and attitude planning than how can I test it ??
Ritesh, If you want to use one of the provided algorithms in MoveIt!, follow this tutorial: www.wilselby.com/research/ros-integration/3d-mapping-navigation/ If you want to use a custom algorithm, look at the MoveIt! API: moveit.ros.org/code-api -Wil
ok so for that I need to run sitl inside the simulator
Great work there. I was trying to use moveit for quadrotor navigation myself. But I did not find good moveit tutorials for dummies, hence I am stuck. Could you lend me some help?
Anurang, Check out my tutorial here: www.wilselby.com/research/ros-integration/3d-mapping-navigation/ It should enable you to recreate the video above and provide detail on the code so you can apply it to your own project. Good luck!
Great work. How did you creat this World file in Gazebo? Is there any tutorial?
Thanks Mohamad! The full ROS repository is available here: github.com/wilselby/ROS_quadrotor_simulator and the outdoor.world file is located here: github.com/wilselby/ROS_quadrotor_simulator/blob/master/quad_gazebo/worlds/outdoor.world You can read more about this research here: wilselby.com/research/ros-integration/ Hope that helps! -Wil
thanks
Hello! What type of micrcontroller/flight controller did you use? And is it possible to create a sliding mode controller/PID controller algorithm and import them in ArduPilot/Pixhawk?
This project used the Ascending Technologies Hummingbird platform and their organic flight controller. It is possible to integrate control algorithms with ArduPilot. Essentially, you want to create a control algorithm that outputs desired roll and pitch angles as well as a yaw rate and thrust command. You can then override the RC commands and have the ArduPilot input come from your control system. I believe DroneKit was meant to help make this process smoother but I'm sure there are examples online as well.
Excellent video. Gives a glimpse of how to 3D map a room with a quadrotor
hi..can you teach how to do the trajectory
Anas, There's more detailed information on my website here: www.wilselby.com/research/mit-class-projects/ to include the Final Project Report and the MATLAB code. Hope that helps!
tq will selby...i will inform u if i need your help
Hi Wil, it is really a nice simulation. Thank you! I have a few questions here.Could I know the order of running your 2d navigation code? Do you have any document regarding the setup besides your web pages? Besides, there is some errors with the robotmodel component in rviz when launching the examples. It says that no transform can be found between others (eg. base_link, camera__link) and the map. May I know where are these links established? Thank you very much!
Hey Keyu, All the code and documentation is on the website here: www.wilselby.com/research/ros-integration/2d-mapping-navigation/ You should only need to run the launch file to get everything running. One launch file is for setting up an environment to create the map (quad_slam_laser_map.launch), the other launch file is for loading that recorded map and navigating using it (quad_2dnav.launch). Those should take care of launching all the required nodes in the proper order to include Gazebo and RVIZ. Regards, Wil
+Wil Selby Thank you very much!
does the above documentation work with ROS-kinetic? or should i use with indigo????
Hi, the link display a blank page
Really nice work! Thank you for sharing!
Hi Wil, I'm currently working on a similar project for my msc thesis. there's any way to get in touch with you via PM?
+Grex0 Feel free to email me accounts [at] wilselby [dot] com.
Excellent control. Hatts off. I have also done work on adaptive backstepping higher order sliding mode control problem for twin rotor. but i used Matlab simulink interface. could you please tell me how did you implement it on board controller using C. did you convert your matlab script file into c? if you have C code then kindly send it to my email adnanjafar1433@gmail.com. your help would be highly appreciated
Adnan, Thanks for the feedback. We created simulations and prototyped our control system in Matlab but used an Arduino microcontroller to read the sensors and sent motor commands. The Arduino code can be read here: s3-us-west-2.amazonaws.com/selbystorage/Files/Arduino_code.pdf and the full project report including Matlab code and hardware specifications can be read here: s3-us-west-2.amazonaws.com/selbystorage/Files/SelbyTrappFInalReport.pdf Hope that helps! -Wil
dude this is genius, what software you using? have you ever told to write augmented reality code for osd hardware such as minimosd? contact me, I have few ideas.
Very cool to see a physical example to all the math I have been doing in my controls class
Hi wilseby can i ask for the matlab code you used just to try that on my simulator and learn few concepts
Amit. Thanks for your interest. I posted my code here: s3-us-west-2.amazonaws.com/selbystorage/Files/sliding_control.m Let me know if you have trouble accessing it. Good Luck!! -Wil
Thanks a lot..wil surely ask..