0

The problem is as stated in the image below. There is a set of right-handed cartesian coordinate systems along $O$, $O'$ and $O''$ with $O$ having the 2D plane. The red plane represents the unrotated plane along the $x$ and $y$ axis of the 3D coordinate system at $O$, the green plane represents the 2D plane rotated by $\alpha$ (degrees) about the $x$ axis and the blue plane is rotated by $\beta$ (degrees) about the vertical axis of the rotated green plane.

enter image description here

Line $PO''$ is equal to $AO$ since $\bigtriangleup AO'O \sim \bigtriangleup PO'O''$ and $OO'$=$O'O''$. $A'$ is contained in the blue plane, $A$ in the red plane and $B'$ on the vertical axis of the green and blue plane.

Also, $B'A' \perp OB'$ and $AO \perp OO'$

The knowns are the lengths $A'O$, $A'B'$ and $OO'$.

A closer image:enter image description here

I am trying to find the length of the line $AO$ using the knowns stated above in order to ultimately find the length $PO''$ but, I do not have any other measurable value other than the knows stated above. Is it possible to just use the knowns or the problem requires more inputs?. How can I find $AO$?, what could be a possible solution?

radk
  • 133
  • I have a few remarks: when you justify why $PO''=AO$, you probably meant triangle $AO'O$? Also your figure suggest that the lines $(AO)$ and $(OO')$ are orthogonal, so you cannot really have $(AP)\perp(OO')$? If that were true, it would mean that $P$ is in the red plane. – N.Bach May 13 '17 at 15:15
  • As for your actual question, assume the red plane, $O$ and $O'$ are fixed. If you have the coordinates of $A'$ and $B'$, you can compute those of $A$ (and $\alpha$ and $\beta$). Unfortunately you only have distances to point $O$, so it is impossible to solve for $AO$ with only these measurements. Knowing $A'O$ and $A'B'$, plus the properties of $B$, you know the shape of triangle $A'B'O$. But you don't know its orientation in 3D space. That means you can rotate the triangle however you want and still obtain a valid configuration. These rotations change the value of $AO$, so it's impossible – N.Bach May 13 '17 at 15:25
  • @N.Bach Thanks for the correction. I made an error while naming it. What if I somehow take a close estimate of the angles $\alpha$ and $\beta$? – radk May 13 '17 at 15:39
  • If both $\alpha$ and $\beta$ are available, my intuition says it is possible, but I'd have to check if the equations hold. Want me to write an answer assuming $\alpha$ and $\beta$ are known/measurable? – N.Bach May 13 '17 at 15:57
  • Yes, that would also help. – radk May 13 '17 at 16:01

1 Answers1

1

Following the discussion in the comments, here is a proposal assuming that angles $\alpha$ and $\beta$ are known.

Notations
I'll take the red plane/coordinates as reference, and take $O$ as everyone's origin. For any point $M$, I'll denote its coordinates by $(x_M^r,y_M^r,z_M^r)^r$ when in the red coordinate system, $(x_M^b,y_M^b,z_M^b)^b$ when in the blue coordinate system and $(x_M^g,y_M^g,z_m^g)^g$ if in the green system.

From the isometric view, I'll assume the $X$ axis points to the right, $Y$ upwards, and $Z$ also to the right. Then for $O'$ its coordinates (red) are $(0,0,OO')^r$; $A'$ is $(A'B',B'O,0)^b$. Using Pythagoras we know that $B'O^2=A'O^2-A'B'^2$.

Strategy for getting $AO$
Now notice that $A$ is the intersection between the line $O'A'$ and the red plane of equation $z^r=0$. One way to solve this is to obtain the parametric equation of line $(A'O')$ in the red system, then find out which point of this line satisfies $z^r=0$. We then obtain the coordinates of $A$ in the red system and can then compute the distance $AO=\sqrt{(x^r_A)^2+(y^r_A)^2}$.

So first up we need the coordinates of $A'$ in the red system (by using angles $\alpha$ and $\beta$).

Coordinate change
In general for any point $M$ \begin{align*} x^g_M &= x^r_M \\ y^g_M &= y_M^r\cos\alpha +z_M^r\sin\alpha \\ z_M^g &= -y_M^r\sin\alpha +z_M^r\cos\alpha \end{align*} These equations are compatible with your figures if $\alpha<0$, otherwise change the sign. \begin{align*} y^b_M &= y^g_M \\ z^b_M &= z_M^g\cos\beta+x_M^g\sin\beta\\ x_M^b &= -z_M^g\sin\beta+x_M^g\cos\beta \end{align*} Again these equations work with your figures if $\beta<0$. (Algebraic vs geometric is always fun.) In particular you have \begin{align*} x^r_M &= z^b_M\sin\beta + x^b_M\cos\beta \\ y^r_M &= y^b_M\cos\alpha -\left( z^b_M\cos\beta -x^b_M\sin\beta \right)\sin\alpha \\ z^r_M &= y^b_M\sin\alpha +\left( z^b_M\cos\beta -x^b_M\sin\beta \right)\cos\alpha \end{align*} So in particular (bis) for $A'$ \begin{align*} x^r_{A'} &= A'B'\cos\beta &(1)\\ y^r_{A'} &= B'O\cos\alpha +A'B'\sin\beta\sin\alpha &(2)\\ z^r_{A'} &= B'O\sin\alpha -A'B'\sin\beta\cos\alpha &(3) \end{align*}

The line $(A'O')$
$$ \text{line }(A'O')= \left\{ \Big( x^r(t),\ y^r(t),\ z^r(t) \Big)^r \mid t\in\mathbb R\right\} $$ where \begin{align*} x^r(t) &= tx^r_{A'} +(1-t)x^r_{O'} & (4)\\ y^r(t) &= ty^r_{A'} +(1-t)y^r_{O'} & (5) \\ z^r(t) &= tz^r_{A'} +(1-t)z^r_{O'} \end{align*} In particular we have $z^r(t)=0 \iff t=-\frac{ z^r_{O'} }{z^r_{A'}-z^r_{O'}} \quad (6)$
Now if you inject equation $(6)$ into $(4)$ and $(5)$, you obtain $x^r_A$ and $y^r_A$ which suffice to conclude.

Edit

Corrected a few typos on equation numbers.

N.Bach
  • 1,826