I need to transform any arbitrary vector in a fixed cartesian coordinate system (let's call this coordinate system $(1)$) to a different cartesian coordinate system (let's call it $(2)$). For illustration, $(1)$ is attached to this airplane where the roll axis is x, the pitch axis is y, and the yaw axis is z:

Coordinate axes of (1)

$2$ is defined a coincident origin to (1) with x2 pointing North, y2 pointing East, and z2 pointing along the resultant nadir radial line. I need to be able to take any vector in (1) and find its coordinates in (2) given the yaw, pitch, and roll of the aircraft.

From the linear algebra that I know, any point in (1) (let's call it $P_{1}$) should be able to be transformed from (1) to (2) by use of a 3x3 Matrix $A$. Thus, the resultant point $P_{2}$ can be found using


My problem is finding what A is. While A can be decomposed into separate transformations by rotating about the x1 y1 and z1 axes of (1) separately using separate rotation matrices, my work up to this point has lead me to conclude that those separate rotations cannot be easily phrased in terms of the yaw pitch and roll angles that I have. Specifically, it seems that three separate transformations about the x y and z axes of (1), using transformation matrices X, Y, and Z respectively, are extrinsic transformations (all the transformations move the vector being transformed relative to (1) throughout the transformation process). Mathematically, these extrinsic rotations are performed by $$P_{2}=XYZP_{1}$$ I am aware that the order of X,Y, and Z is important since $AB\neq BA$. I also am not sure what order they should be in.

Ideally, what I would like is to mathematically transform the vector through intrinsic rotations. Specifically, I would first like to transform $P_{1}$ in (1) by a rotation about $z_{1}$ (a yaw). Then take the new orientation of the plane and define axes $x_{1}'$, $y_{1}'$, and $z_{1}'$ along the new roll pitch and yaw axes of the plane. Then I'd like to be able to perform a pitch transformation about the $y_{1}$' axis to get new axes $x_{1}'', y_{1}'', and z_{1}''$. Then finally, the roll transformation about the $x_{1}''$ axis to get axes $x_{2}, y_{2}, and z_{2}$.

If such intrinsic rotations are not possible to perform separately, then how do I setup A so that I can input the yaw pitch and roll angles, and get $P_{2}$ out?

  • $\begingroup$ Does this help? math.stackexchange.com/questions/525670/… $\endgroup$ – law-of-fives May 11 '17 at 16:33
  • $\begingroup$ No. That thread contains a problem with different given information. I am trying to find a matrix A as a function of roll, pitch, and yaw angles. That thread has a setup and end goal different from mine. $\endgroup$ – Unique Worldline May 11 '17 at 16:45

This material is in chapter 1 of any aircraft dynamics book. The standard aerospace sequence is yaw-pitch-roll (321). You get your $P_2$ by calculating $A$ using the angles and then doing the matrix multiplication $AP_1$. The information below is all available from a google search:





  • $\begingroup$ I have gone through an aircraft dynamics book's explanation now, and I have the same problem. The rotations that the matrices X, Y, and Z make are all with respect to (1). Put another way, the angles I have allow the following instructions to be given: yaw the aircraft $\theta_{z}$ about the z axis, then pitch the aircraft above the horizon (about the new y' axis) by $\theta_{y}$, then finally roll the aircraft about the new x'' axis by $\theta_{z}$ axis. This is not the set of instructions that occurs in the transform $P_{2}=XYZP_{1}$. That transformation performs all rotations about ... $\endgroup$ – Unique Worldline May 11 '17 at 21:28
  • $\begingroup$ ... the original x,y, and z axes. Not z, y' and x''! How do I do the former? $\endgroup$ – Unique Worldline May 11 '17 at 21:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.