Reference frames — body frame, inertial frame; rotation between them: Question bank

3.5.1 · D5 · Physics › Guidance, Navigation & Control (GNC) › Reference frames — body frame, inertial frame; rotation betw

Notation ke reminders jo tumhe chahiye honge (sab parent note mein bane hain, yahan dobara state kiye hain taaki yeh page akela kaam kar sake):

$I$ = inertial frame (non-rotating aur non-accelerating, taaron se tied); $B$ = body frame (vehicle pe bolted).
$v^{I}$ , $v^{B}$ = same arrow ke components, har frame mein alag likhe gaye.
$R_{B I}$ = DCM jo inertial components ko body components mein badalta hai: $v^{B} = R_{B I} v^{I}$ .
"Orthogonal" ka matlab hai $R^{T} R = I$ , yani iska inverse sirf iska transpose hai.
$ω$ = angular-velocity vector: yeh current spin axis ke along point karta hai, aur iska length spin rate hai (radians per second). Iska time-derivative $\dot{ω}$ = angular acceleration — spin kitni tezi se speed up ho raha hai ya axis kaise tilt ho raha hai.

Neeche wala figure is page ke almost har trap ke peeche ka mental picture hai: ek physical arrow, do sets of numbers.

True or false — justify

Ek rotation matrix physical arrow ko khud badal deta hai

False. Arrow (space mein ek real direction) kabhi nahi badalta; sirf iske components, yaani teen numbers jo use describe karte hain, ek frame se doosre mein badlate hain — bilkul woh do number-columns ki tarah jo upar figure mein hain.

R_{B I}

aur

R_{I B}

same matrix hain

False. Yeh dono transposes hain:

R_{I B} = R_{B I}^{T}

, kyunki rotation ko reverse karne se har direction cosine ka role reverse ho jaata hai (rows, columns ban jaati hain).

DCM ki har entry

- 1

aur

+ 1

ke beech hoti hai

True. Har entry do unit axes ke beech ke angle ka

cos

hai, aur cosine kabhi

[- 1, + 1]

se bahar nahi jaata; isi liye ise direction cosine matrix kehte hain.

det = - 1

wala matrix ek valid attitude ho sakta hai

False.

det = - 1

ek reflection hai (yeh handedness flip karta hai); real rigid rotations mein hamesha

det = + 1

hota hai, toh iska matlab hai ki kahin left-handed frame aa gaya hai.

Kisi vector ko rotation matrix se multiply karne par uski length badal sakti hai

False. Rotations length aur angles preserve karte hain (

∥ R v ∥ = ∥ v ∥

); agar length badlti, toh

R^{T} R \neq = I

hota aur woh rotation nahi hoti.

Body frame ek inertial frame hai jab tak vehicle thrust nahi kar raha

False. Do tarike se fail ho sakta hai: vehicle phir bhi spin kar sakta hai (Coriolis/centrifugal terms appear hote hain), aur gravity field mein free-fall mein non-thrusting body translationally accelerate kar rahi hoti hai — koi bhi ek inertiality tod deta hai; dekho Coriolis and fictitious forces in rotating frames.

Rotation ke liye

R^{- 1}

compute karne ke liye Gaussian elimination chahiye

False. Orthogonality se

R^{- 1} = R^{T}

milta hai, toh bas transpose karo — koi inversion nahi, koi rounding error nahi.

Do rotations ka order matter nahi karta

False. Matrix multiplication commute nahi karta, isliye generally

R_{1} R_{2} \neq = R_{2} R_{1}

; physically, yaw-then-pitch alag jagah pahunchata hai pitch-then-yaw se.

R_{B I}

ki rows body axes hain jo inertial components mein likhi hain

True. Har row ek body unit vector hai jo teen inertial axes pe project kiya gaya hai, aur yahi mnemonic hai "rows are body axes."

Star trackers star directions ko body frame mein report karte hain

True, phir convert kiya jaata hai. Star tracker vehicle pe bolted hota hai, isliye woh body coordinates mein sense karta hai; software phir

R

use karke isse known inertial catalogue se relate karta hai (stored list of star directions in

I

) — dekho Attitude determination — star trackers.

Spot the error

"Kyunki

R^{T} R = I

, hum yeh bhi jaante hain ki

det R = 1

, isliye sare orthogonal matrices rotations hain."

Error. Orthogonality sirf

det R = \pm 1

force karta hai;

- 1

case (ek reflection) bhi orthogonal hota hai. Use rotation kehne ke liye additionally

det = + 1

require karna padega.

"Body → inertial convert karne ke liye main

v^{I} = R_{B I} v^{B}

use karta hoon."

Error. Woh matrix inertial → body jaata hai. Body → inertial ke liye transpose chahiye:

v^{I} = R_{B I}^{T} v^{B}

"Frames

I, O, B

ke liye main likh sakta hoon

R_{B I} = R_{O I} R_{B O}

Error. Touching subscripts cancel hone chahiye:

R_{B I} = R_{B O} R_{O I}

(rightmost pehle act karta hai,

I \to O \to B

"Elementary

z

-rotation hai

R_{z} (θ) = cos θ sin θ 0 - sin θ cos θ 0 001

Error — galat convention. Yeh active (arrow-ko-rotate-karo) matrix hai. Parent note ka passive frame-to-frame DCM iska transpose hai,

R_{z} (θ) = cos θ - sin θ 0 sin θ cos θ 0 001

, jisme

- sin θ

lower (row 2, column 1) slot mein hai. Neeche wala do-column figure dekho: passive ek fixed arrow ko re-label karta hai, active arrow ko move karta hai — yeh dono ek doosre ke transposes hain.

"Ek gyro meri orientation measure karta hai, toh main seedha attitude read kar sakta hoon."

Error. Ek rate gyro angular velocity

ω

(spin axis aur spin rate) body axes mein measure karta hai, orientation nahi; attitude paane ke liye tumhe ise time ke saath integrate karna hoga (ya star tracker ke saath fuse karna hoga) — dekho IMU and gyroscope sensing.

"Kyunki

F = m a

, main ise directly body-frame accelerations use karke apply kar sakta hoon."

Error.

F = m a

cleanly sirf

I

mein hold karta hai; rotating body frame mein tumhe fictitious forces (Coriolis, centrifugal, aur

\dot{ω} \times r

term) add karne padte hain pehle Newton's law valid hone se.

Why questions

Har DCM entry sine ki jagah cosine kyun hoti hai?

Kyunki ek component ek projection (dot product) hai ek unit axis pe, aur do unit vectors ka dot product un ke beech ke angle ka cosine hota hai — projection hi cosine hai.

Rotation matrix ko transpose karne se rotation undo kyun hoti hai?

Kyunki

R^{T}

ke columns,

R

ki rows hain (body axes), isliye

R^{T}

unhi axes pe wapas project karta hai — yeh exactly same direction cosines ko reverse karta hai.

GNC ko frames ke beech hazaron baar per second convert kyun karna padta hai?

Sensors body frame mein bolte hain jabki physics aur pointing goals inertially defined hain, isliye computer continuously har naye measurement ko us frame mein re-express karta hai jahan agla decision lena hai.

Hum entirely body frame mein navigate kyun nahi kar sakte?

Ek spinning body frame mein angular velocity

ω

hoti hai, isliye wahaan Newton's law mein fictitious terms aa jaati hain — Coriolis term

2 ω \times v

, centrifugal term

ω \times (ω \times r)

, aur Euler term

\dot{ω} \times r

(jab bhi spin rate ya axis badlti hai); star catalogue bhi inertial hai, isliye body-only navigation ka matlab hai physics ko phir se derive karo aur har instant har star ko re-map karo.

Teen elementary rotations kisi bhi attitude describe karne ke liye kyun kaafi hain?

3-D mein kisi bhi orientation ke exactly teen rotational degrees of freedom hote hain, isliye teen axis rotations ka product (jaise yaw-pitch-roll) har attitude tak pahunch sakta hai — dekho Euler angles and gimbal lock.

Engineers DCM ya Euler angles ki jagah quaternions kyun prefer karte hain?

Quaternions Euler angles ki singularities (gimbal lock, neeche edge cases mein explain kiya gaya) se bachte hain aur numerically stable rehte hue sirf chaar numbers use karte hain nau ki jagah — dekho Quaternions for attitude representation.

Frames ko "dominoes ki tarah" compose karne mein adjacent subscripts match kyun karne chahiye?

Har matrix do specific frames ke beech le jaata hai, isliye ek ka target doosre ka source hona chahiye; mismatched subscripts ek aisa jump describe karte hain jo koi single conversion nahi karta.

Edge cases

Ek aisa vector rotate karo jo exactly rotation axis pe lie karta ho (jaise

\overset{z}{^}

under

R_{z} (θ)

Uske components nahi badlenge — axis pe ek point rotation se fixed hota hai, jo sabse tezi sanity check hai ki tumne sahi axis choose ki hai.

Identity rotation

R = I

(zero angle) apply karo.

Body aur inertial components identical hain; do frames momentarily aligned hain, aur

R^{T} R = I

trivially hold karta hai.

θ

se rotate karo aur phir same axis ke baare mein

- θ

se.

Tum wapas start pe aa jaate ho:

R (- θ) R (θ) = I

, aur kyunki

R (- θ) = R (θ)^{T}

, yeh sirf orthogonality property hai action mein.

Zero vector

0

ko rotate karo.

Yeh har frame mein

0

hi rehta hai — koi direction na hone wale point ko reorient karne ke liye kuch nahi hai, aur length (

0

) trivially preserved hai.

Do frames jo same origin share karte hain lekin

z

ke baare mein

18 0^{\circ}

rotate hain.

x

aur

y

components sign flip karte hain jabki

z

rehta hai;

det = + 1

phir bhi hai (yeh genuine rotation hai, reflection nahi), chahe

x y

-plane mein yeh mirror jaisa dikhe.

Gimbal lock: yaw-pitch-roll (3-2-1) sequence jisme pitch

= 9 0^{\circ}

ho.

Pehla aur teesra axis line up ho jaate hain, isliye yaw aur roll ab same direction ke baare mein spin karte hain — ek degree of freedom silently gayab ho jaati hai aur Euler angles DCM se uniquely recoverable nahi rehte. Isi liye

3 \times 3

DCM valid rehta hai lekin use describe karne wala angle triple singular ho jaata hai — quaternions ki motivation Quaternions for attitude representation mein.

Ek rotation matrix ko ek vector do, phir

36 0^{\circ}

se rotate karo.

Tumhe original components wapas milte hain (

R (36 0^{\circ}) = I

); attitude periodic hai, isi liye angle bookkeeping (ya quaternions) ko wrap-around handle karna padta hai.

Recall Poore trap set ka ek-line summary

Same arrow, alag numbers; inverse transpose hai; touching subscripts cancel hote hain; $det = + 1$ ; body frame kabhi bhi inertial nahi hota jab woh spin ya free-fall kare; aur pitch $= 9 0^{\circ}$ gimbal lock karta hai.