- 15
- 516 570
Woolfrey
Registrace 27. 05. 2006
Manually Controlling a Manipulator with a Joystick
This was part of a project looking at developing an underwater robot to clean bridge piles with high-pressure water blasting. This was a demonstration of how the manipulator might conduct the cleaning process. It was controlled manually via a joystick.
The manipulator is the igus robolink - a six degree-of-freedom (6DOF), cable-driven robot arm.
The manipulator is the igus robolink - a six degree-of-freedom (6DOF), cable-driven robot arm.
zhlédnutí: 511
Video
Redundancy Resolution with the Sawyer Robot
zhlédnutí 947Před 3 lety
The Sawyer manipulator from Rethink Robotics has 7 joints or degrees of freedom (DOF). Only 6 DOF is required to control the position and orientation of the end-effector in 3D space. This leaves 1 DOF of redundancy. Using this fact, I show how the manipulator can autonomously reconfigure itself to be structurally stiff or compliant in a vertical direction. I did not tell the manipulator how to ...
Visual Servoing
zhlédnutí 7KPřed 3 lety
A demonstration of a control method known as visual servoing. A camera is used in the feedback control loop of the manipulator. There are two types: - Eye-in-hand, where the camera is mounted directly on the manipulator - Eye-to-hand, where the camera observes the system externally. I use both the Sawyer from Rethink Robotics, and the UR3 from Universal Robot. A Blackfly S camera (BFS-U3-16S2C-...
Feedback Control vs Predictive Control
zhlédnutí 760Před 3 lety
Demonstrating the performance of feedback control versus predictive control for tracking a moving target using a UR3 manipulator. A Blackfly S camera (BFS-U3-16S2C-CS) is used to observe the augmented reality (AR) markers. The camera runs at approximately 112 frames per second. The ar_track_alvar package in Robot Operating System (ROS) is used to detect the markers and determine pose of the boa...
5.2 IK via Optimization
zhlédnutí 9KPřed 6 lety
In this lecture, I show how to solve Inverse Kinematics via gradient descent optimization.
5.1 Inverse Kinematics
zhlédnutí 68KPřed 6 lety
In this lecture, I introduce the concept of Inverse Kinematics (IK). That is, for a given end-effector pose of the robot we want to find a set of joint solutions that satisfies the kinematics.
1.0 Overview of Lectures on Robotics
zhlédnutí 12KPřed 6 lety
In this introductory lecture, I present a "road map" that ties together the principle topics behind robot kinematics. I also give some of the conventions I use for linear algebra.
2.1 Relative Pose and Translation
zhlédnutí 12KPřed 6 lety
In this video, I introduce the concept of pose which expresses position and orientation. Furthermore, I cover translation in a common reference frame.
2.2 Rotation Matrices
zhlédnutí 40KPřed 6 lety
In this lecture, I show how to derive a matrix that rotates vectors between 2 different reference frames. This rotation matrix is in the Special Orthogonal group, and I derive some of the mathematical properties associated with it. I also give an example of how to inverse and multiply rotation matrices to find relative rotation between reference frames in an environment.
3.1 Transformation Matrices
zhlédnutí 15KPřed 6 lety
In this lecture, I introduce the concept of a transformation matrix which combines a rotation matrix in the Special Orthogonal group SO(3), and a translation vector in 3D Euclidean space. I develop the inverse of a transformation matrix, show how to perform point transformations between frames, and chain transformation matrices to find the relative transform between reference frames in a given ...
2.3 Rotations in 3D
zhlédnutí 250KPřed 6 lety
In this lecture, I extend the 2D rotation matrix of SO(2) from Lecture 2.2 to SO(3). Rotation matrices can be constructed from elementary rotations about the X, Y, and Z axes. I also cover the problem of Gimbal Lock, and how to express rotation error.
4.1 Forward Kinematics
zhlédnutí 45KPřed 6 lety
In this lecture, I introduce the concept of forward kinematics: finding the end-effector pose of a serial link manipulator given the joint positions. I also introduce the idea of using transformation matrices as an alternative approach to the forward kinematics problem.
4.2 DH Parameters
zhlédnutí 37KPřed 6 lety
In this lecture, I introduce the concept of DH parameters for specifying the kinematic structure of a serial link manipulator. I give some worked examples, and also some common joint-to-joint transforms that you might encounter when trying to model robot arms. My cat Kismet also makes a cameo.
2.4 Derivatives of the Rotation Matrix
zhlédnutí 20KPřed 6 lety
In this lecture, the derivatives of the rotation matrix are introduced. First, I cover the time derivative of a rotation matrix in the Special Orthogonal Group SO(n). I also show how to get an exact numerical solution for the partial derivative of the rotation matrix with respect to a joint angle on a robot kinematic chain.
Sehr gut
hii i need somemore concept can i get your mail
much love and respect! thank you very much
What software was used to draw these frames? Any recommendations?
Great work man! Can i get the link to all the slides of this playlist?
In illustration 2:30 to 3:15, the direction of rotation is wrong for the y-axis :)
at 4:20, I don't get why you say it is a sum of two rotations? R(t) is rotation, but is dt*R'(t) also a rotation? First of all, multiplication by a small value dt is suspicious: rotation saves volume, but if you multiply a matrix by a very small number, it has a very low chance to have the determinant of 1. Even more, as we remember, R'=S(w)*R. The R is a rotation, that's fine. But S(w) is a 3x3 skew-symmetric matrix, and since it is of an odd dimension, it's determinant is always zero: det(S(w)) = 0. This means that det(R')=0. So dt*R'(t) cannot be a rotation matrix.
at 3:40, honestly I don't get what is being proved here. R'RT is skew-symmetric => there exists such vector w, that R'RT=S(w). Next we multiply both sides by R on the right and get that R'=S(w)R
at 2:50 why can we reduce R'RT=-RR'T to R'=-R'T? This is true only in case when R=I, which is not true in general case? especially later you write R'=S(w)R -- so it suggests that R can be non-trivial
Is there any open source code for this project? I really need to learn this. thank you so much
Is there any geometric/visual way of seeing why every rotation is a composition of two rotations.
Australia never existed?!
very good.
How do we calculate a second rotation? For example, after a body pitches 45 degrees, it yaws 45 degrees about its body frame? It would be incorrect to multiply its nose vector, say <1,0,0> by R(y=45 deg) * R(z = 45 deg), since the second matrix that you would be multiplying by is in the inertial (unrotated) frame.
Thanks for making this video. I cannot understand how much of your timr and effort it took to male it. Respectfully disagree in the matrix you got from saying "Express axes of frame [1] as functions of axes of frame [2 ]". I don't see or understand how this was derived. Thank you.❤
Can you help me with an exercise please 😭🙏🏻🙏🏻 I've been struggling for 2 days with just having to give the homogeneous transformation matrix Thank you very much in advance 🌻 It’s a graded practical exercise 🙏🏻🙏🏻😭
This video is surprisingly so good didn't expect this much of quality and simplest of the content
I got a question that the normal quation is qk = qk-1 + alpha*Pseudo-Jacobian(PJ)*deltaX, i know that Gradient descent is a variant of Jacobian but for the deltaX calculation, some paper had wrote the formular "deltaX = J1*dx+ k*(I-J1*J)*GradianH" and that value have been updated by "value = value + deltaX" so what is this Gradient affect on? Thank you
Thank you so much
and thaks for the great lecture i learn a lot from you!!
teacher i have a quetion. i want to know why you change 3x4 to 4x4matrix to make homegeneous matrix. is it for the calculation?
How can I add rotation error to the main objective function, position error, as rotation error is a 3X3 matrix?
Thank you so much! how do I parameterise spherical joints? One of the joint in the robot we have is a spherical joint, thank you!
There seems to be a mistake at 4:21. it says tan^-1(x/y) when I believe it should be tan^-1(y/x). The previous slide says y/x.
excellent work! help understanding DH well
Thanks for the awesome content, Dr. Woolfrey. It would be of great help if you could just make me clear on the shape of the matrices, delx and qk. For my 4DOF robotic arm, are these shapes right? 1. delx= (3X2) where first column for (position error of x,y,z) and second column for (RPY angles of e0.) 2. inverse of Jacobian matrix provides shape of 4X3 . so on doing delq=inv(J)*delx, i get delq of shape 4X2. 3. ultimately, qk of shape 4X2, where first column for desired angles and second for desired RPY values. Thank you so much for your time.
[2:51] rotation on Y is incorrect, should be ccw
Amazing
That was really helpful. Thank u sir simple and clear. Thanks u 🙏
Holy moly! This is pretty dang impressive! Good work dude!
Excellent video!
Hi, I believe the pitch rotation (about the y-axis) may be illustrated incorrectly. A positive pitch rotation should rotate a vector downwards. I.e., a rotation of pi/2 will rotate a vector pointing straight forward [1,0,0] to straight down [0,0,-1].
at 3:54, this w is s frame not b frame, is that correct? which one is more often used?
hi can you say what kind of matrices is this czcams.com/video/7XZuy5NhxhA/video.htmlsi=83294k5J0P04-0X2
Extremely helpful... Please make more videos on robotics and please cover the the dynamics portion and all.
At 7:12 how can you sub S(w) = S(a)q ?
I just simulated the 2-link solution (I put the formulas inside octave and GeoGebra). It completely breaks as soon as you play around with l1,l2,x and y even if there's clearly 2 solutions. Is it possible that there is an error in the formulas?
Thank you 😊
Thx
Good video .How I could substitute Phi =2tan(R23,R33)as function of number elements please & extracting 'Jacobian Matercis' from rotation matrix.
good job do you have the GitHub for it ？
nice work ，do you have a plan to open source code about your project
1 hour to exam and this was great 😅
God bless woolfrey 🙏
dislike for too low voice.
what if i have Q3
The definite answer to the question: "Why would I need matrix operations or linear algebra in general?" Well done!
The matrix "is _given_ by..." 😏
How do we solve for x = 0 (fully horizontal) or y = 0 (fully vertical)?
How you made it so smooth with almost no latency ?