Groves, Chapter 2

Coordinate Frames, Kinematics, and the Earth

ECI, ECEF, local navigation, body frames. Euler angles, DCMs, quaternions, Earth models, and gravity.

Prerequisites: Linear algebra basics (matrix multiplication, transpose, inverse).
10
Chapters
1
Simulation
10
Quizzes

Chapter 0: Why Frames Matter

You're standing on the Earth. Relative to the ground beneath you, you're stationary. Relative to the Sun, you're hurtling through space at 30 km/s. Relative to a satellite overhead, you're rotating at 460 m/s. Same person, three different answers for "velocity." The difference? The coordinate frame.

Navigation is fundamentally a multi-frame problem. Inertial sensors measure motion relative to inertial space. GPS measures position relative to a satellite constellation. But you want to know where you are relative to the Earth's surface. Getting from one to another requires careful mathematics.

The core idea: A coordinate frame has an origin and three orthogonal axes (x, y, z). It can serve as a reference (something known) or describe an object (something to be determined). Navigation always involves at least two frames: object and reference.

Four frames dominate navigation:

FrameSymbolOriginAxes
Earth-Centered InertialiEarth's centerFixed relative to stars
Earth-Centered Earth-FixedeEarth's centerRotate with Earth
Local NavigationnUser positionNorth, East, Down
BodybUser positionForward, Right, Down
Check: Why does navigation require multiple coordinate frames?

Chapter 1: The ECI Frame

The Earth-Centered Inertial (ECI) frame has its origin at the Earth's center and its axes are fixed relative to distant stars. Newton's laws hold directly in this frame — no fictitious forces.

The z-axis points along the Earth's spin axis (toward the North Pole). The x-axis points toward the vernal equinox (where the Sun's apparent path crosses the equator at the March equinox). The y-axis completes the right-handed system.

Gyroscopes measure angular rate relative to this frame. Accelerometers measure specific force, which is the acceleration due to all forces except gravity — equivalent to the inertial-frame acceleration minus gravitational acceleration.

Why inertial? "Inertial" means the frame is not accelerating or rotating. Newton's second law (F = ma) applies directly. In a rotating frame, you need to add Coriolis and centrifugal terms. The ECI frame is the natural reference for inertial sensors.
Check: What makes the ECI frame "inertial"?

Chapter 2: The ECEF Frame

The Earth-Centered Earth-Fixed (ECEF) frame also has its origin at Earth's center, but its axes rotate with the Earth. The z-axis is the same as ECI (spin axis). The x-axis points through the intersection of the prime meridian (0° longitude) and the equator. The y-axis completes the right-hand system.

ECEF is the natural frame for GPS. Satellite positions are computed in ECEF coordinates, and the standard GPS position solution is in ECEF. A point fixed on Earth's surface has constant ECEF coordinates (ignoring tectonic motion).

The relationship between ECI and ECEF is a simple rotation about the z-axis at the Earth's rotation rate, ωie ≈ 7.292 × 10−5 rad/s:

Cie = Rzie · (t − t0))

Where Cie is the coordinate transformation matrix from ECI to ECEF, and Rz is a rotation about the z-axis.

Check: How are the ECI and ECEF frames related?

Chapter 3: The Local Navigation Frame

You probably think about position as latitude, longitude, and height — not as Cartesian x,y,z from the Earth's center. The local navigation frame (n-frame) formalizes this intuition.

Its origin is at the user's current position. Its axes point North (x), East (y), and Down (z). This is called the NED convention.

The n-frame is the most natural frame for expressing a navigation solution: "I'm moving 5 m/s north and 2 m/s east" is more useful than Cartesian ECEF velocities.

Key subtlety: The n-frame rotates as the user moves across the Earth's surface. If you move east, the "north" direction changes. This rotation (called the transport rate) must be accounted for in the navigation equations. It adds Coriolis-like terms.

Converting between ECEF and the local navigation frame requires the user's latitude L and longitude λ. The transformation matrix involves sines and cosines of these angles.

Check: What are the axes of the local navigation frame?

Chapter 4: The Body Frame

The body frame (b-frame) is attached to the vehicle (or person). Axes: x forward, y right, z down. In a strapdown INS, the accelerometers and gyroscopes are fixed to the body, so they naturally measure in body-frame axes.

The relationship between the body frame and the local navigation frame is the vehicle's attitude: roll, pitch, and heading. This is what the INS must track to transform accelerometer measurements from body axes to navigation axes.

Other frames: Groves also defines a wander-azimuth frame (like the n-frame but with the horizontal axes rotated by a "wander angle") to avoid singularities at the poles, and platform frames for gimbaled INS. For most applications, i, e, n, and b are sufficient.
Check: Why must an INS track the body-to-navigation frame attitude?

Chapter 5: Euler Angles & DCMs

How do we describe the orientation of one frame relative to another? Three main representations, each with pros and cons.

Euler angles: roll (φ), pitch (θ), and heading/yaw (ψ). Intuitive and compact (3 numbers), but suffer from gimbal lock at ±90° pitch, and the rotation order matters.

Coordinate Transformation Matrix (DCM, also called Direction Cosine Matrix): a 3×3 orthogonal matrix Cbn that transforms a vector from body axes to navigation axes. No singularities, but 9 numbers with 6 constraints (orthogonality).

vn = Cbn vb

Key DCM properties:

PropertyExpression
Inverse = transposeCnb = (Cbn)T
Determinantdet(C) = +1
Chain ruleCbn = Cen Cbe
Key insight: Euler angles are for humans (intuitive), DCMs are for computation (no singularities, easy composition). The navigation processor typically works with DCMs or quaternions internally, converting to Euler angles only for display.
Check: What is the main disadvantage of Euler angles?

Chapter 6: Quaternions

Quaternions represent rotation using four numbers: a scalar part and a 3D vector part. They avoid gimbal lock, are more compact than DCMs (4 vs. 9 numbers), and are efficient for computation.

A quaternion q = (q0, q1, q2, q3) satisfies the constraint q02 + q12 + q22 + q32 = 1 (unit quaternion). It represents a rotation by angle μ about axis e as:

q0 = cos(μ/2),   (q1, q2, q3) = sin(μ/2) · e

Quaternion multiplication composes rotations. The DCM can be directly computed from the quaternion (and vice versa). Most modern INS implementations use quaternions for attitude propagation.

The rotation vector is another representation: a 3-element vector whose direction is the rotation axis and magnitude is the rotation angle. It is used in some attitude update algorithms but suffers from a singularity at ±2π.

RepresentationElementsSingularity?Composition
Euler angles3Yes (±90° pitch)Complex
DCM9 (6 constraints)NoMatrix multiply
Quaternion4 (1 constraint)NoQuaternion multiply
Rotation vector3Yes (±2π)Complex
Check: How many independent parameters does a unit quaternion have?

Chapter 7: The Earth Model

Navigation requires a mathematical model of the Earth's surface. The Earth isn't a perfect sphere — it bulges at the equator. The standard model is an oblate spheroid (ellipsoid of revolution).

Two radii define it: the equatorial radius R0 ≈ 6,378,137 m and the polar radius RP ≈ 6,356,752 m. The flattening f = (R0 − RP)/R0 ≈ 1/298.257.

Geodetic latitude L is the angle between the ellipsoid normal and the equatorial plane — this is what GPS gives you. Geocentric latitude is the angle from the center — different from geodetic due to flattening. Longitude λ is the angle east of the prime meridian. Height h is above the ellipsoid.

Two radii of curvature: The meridian radius RN (along a north-south line) and the transverse radius RE (along an east-west line). They vary with latitude: both are smallest at the equator and largest at the poles. These appear throughout the navigation equations.

The geoid is a surface of constant gravitational potential that best fits mean sea level. Height above the geoid (orthometric height) is what we mean by "elevation" in everyday life. It differs from ellipsoid height by up to ±100 m.

Check: What shape is used to model the Earth's surface in navigation?

Chapter 8: Gravity vs. Gravitation

This distinction trips up many people. Gravitation (γ) is the attractive force due to Earth's mass — what Newton's law gives you. Gravity (g) is what you actually feel when standing still: gravitation minus the centrifugal acceleration due to Earth's rotation.

g = γ − ωie2 R · cos(L)

At the surface, total gravity is about 9.8 m/s2. The centrifugal component contributes up to 0.034 m/s2 (at the equator). Small but significant for precision navigation.

Accelerometers measure specific force f, which is the acceleration due to all forces except gravity. For a stationary object on Earth's surface, the specific force is the reaction to gravity (pointing up), not zero. The relationship:

a = f + g

where a is the inertial acceleration. This means the INS must know gravity to get acceleration from the accelerometer output.

Critical distinction: In inertial-frame navigation equations, you need gravitation (γ). In Earth-frame or local-navigation-frame equations, you need gravity (g = γ − centrifugal). Using the wrong one introduces systematic position error.

The WGS 84 Somigliana model gives surface gravity as a function of latitude. Height variation follows an inverse-square law. For high-precision applications, spherical harmonic gravity models or gravity anomaly databases are used.

Check: What is the difference between gravity and gravitation?

Chapter 9: Summary

Coordinate Frame Hierarchy

Click a frame to highlight its properties.

Key takeaways:
• Four main frames: ECI (i), ECEF (e), Local Nav (n), Body (b)
• Three attitude representations: Euler angles (intuitive but singular), DCMs (robust, 9 numbers), quaternions (compact, robust)
• Earth modeled as oblate spheroid with equatorial and polar radii
• Position: geodetic latitude, longitude, ellipsoid height
• Gravity ≠ gravitation; the difference is centrifugal acceleration from Earth's rotation
• Specific force = acceleration − gravitational acceleration (what accelerometers measure)
• Frame transformations require careful handling of rotating references (Coriolis, transport rate)
Check: Which attitude representation is singularity-free and most compact?