1. Why Large-Angle Maneuver Theory Is Required

Small-angle attitude models assume the spacecraft remains close to a reference orientation, allowing linearization of kinematics and dynamics. This approximation becomes unreliable for practical slews involving large rotations, where the geometry of orientation and the realities of actuation dominate the behavior.

• Nonlinear attitude geometry (SO(3)): attitude is not a Euclidean vector; global error mappings are nonlinear.
• Euler-angle singularities: gimbal lock can occur depending on the attitude path and chosen angles.
• Short vs long rotation path: the same target attitude can be reached by two valid rotation directions.
• Actuator limits: maximum torque, maximum body rate, and reaction wheel saturation constrain feasible slews.

Therefore, large-angle reorientation must be treated as a nonlinear, constrained maneuver design problem using globally valid attitude representations and control laws that remain stable under bounded actuation.

2. Breakdown of Small-Angle Linear Models

Under small-angle assumptions, rotational dynamics are commonly approximated as:

\[ \mathbf{J}\dot{\boldsymbol{\omega}} \approx \mathbf{u}, \qquad \dot{\boldsymbol{\theta}} \approx \boldsymbol{\omega}. \]

This supports simple linear PD control in an angle-error coordinate system. For large rotations, two issues appear:

(a) Nonlinear kinematics

Attitude evolves on the nonlinear manifold $\mathrm{SO}(3)$. A linear 3-vector “angle error” is not globally valid: the mapping between orientation and a local error coordinate changes with the operating point.

(b) Rotation-direction ambiguity

The same physical orientation can be represented in multiple equivalent ways. Without a consistent convention, feedback may command a longer-than-necessary rotation path.

These issues motivate quaternion-based states and quaternion-based error definitions for large-angle maneuver control.

3. Quaternion State Representation: the large-angle friendly state

A standard large-angle maneuver state uses the attitude quaternion and body angular velocity:

\[ q = [\mathbf{q},\,q_4], \qquad \boldsymbol{\omega}. \]

Quaternions avoid Euler-angle singularities, but they introduce an important equivalence:

\[ q \;\text{and}\; -q \quad \text{represent the same physical orientation.} \]

4. Rest-to-Rest Slews: what “good” means

A rest-to-rest maneuver starts and ends with zero body rate:

\[ q(0)=q_0,\;\boldsymbol{\omega}(0)=0, \qquad q(t_f)=q_c,\;\boldsymbol{\omega}(t_f)=0. \]

The controller must do two jobs simultaneously:

• Reorient: drive the attitude to the commanded orientation.
• Stop: remove angular rate at the final time so there is no residual spin.

This coupling is central: large-angle maneuver control cannot treat attitude and rate as independent problems.

5. Quaternion Attitude Error: the correct large-angle error signal

Feedback control uses an attitude error quaternion $q_e$ that represents the relative rotation from the current attitude to the commanded attitude. In practice, $q_e$ is constructed through quaternion multiplication (commanded attitude relative to current attitude).

Design principles for a usable large-angle error

• Construct $q_e$ as a relative attitude (via quaternion composition).
• Use the vector part of $q_e$ to define the torque direction.
• Apply a sign convention (e.g., $q_{e4}\ge 0$) to ensure shortest-path rotation.

This produces an error signal that remains meaningful across large rotations and avoids unintended long-way slews.

6. Quaternion-Feedback Reorientation Law

A widely used nonlinear feedback structure combines an attitude error term and a rate damping term:

\[ \mathbf{u} = -\mathbf{K}\,\mathbf{q}_e - \mathbf{C}\,\boldsymbol{\omega}. \]

• $\mathbf{q}_e$: vector part of the error quaternion (rotation direction).
• $\mathbf{K}$: attitude gain matrix (how strongly the controller drives the rotation).
• $\mathbf{C}$: rate damping matrix (removes kinetic energy and suppresses overshoot).

The quaternion term drives the spacecraft toward the target orientation, while the damping term dissipates rotational energy so the maneuver can end at rest. Correct sign handling of the quaternion error is essential to maintain shortest-path behavior.

7. Eigenaxis Rotation: the clean geometric way to describe a slew

Any reorientation between two attitudes can be expressed as a rotation about a single axis — the eigenaxis. In an unconstrained setting, an eigenaxis maneuver can be viewed as a clean “rotate about one fixed axis” solution.

Why eigenaxis geometry is useful

• It describes the shortest rotation between orientations (minimum-angle reorientation).
• It aligns naturally with quaternion error feedback (the error defines an axis and an angle).
• It provides an intuitive physical picture for large slews.

In real spacecraft, eigenaxis slews must be executed under constraints: torque bounds, rate limits, and internal momentum limits (reaction wheels / CMGs).

8. Time-Optimal vs Effort-Optimal Maneuvers: What changes, what stays

8.1 Time-optimal reorientation (minimum time)

With bounded control (e.g., $|u_i|\le u_{\max}$), minimum-time slews typically produce bang-bang control: torque saturates at its limit and switches sign at specific times. In a decoupled or single-axis approximation, the canonical structure is:

• accelerate at maximum available torque,
• switch and decelerate at maximum opposite torque,
• arrive at the target with zero rate.

8.2 Effort-optimal / fuel-optimal reorientation

If the objective penalizes effort (e.g., $\int \|\mathbf{u}\|^2\,dt$) or propellant usage, the resulting control tends to be smoother:

• lower peak torque demand,
• reduced excitation of flexible / slosh modes,
• longer maneuver time.

The choice reflects a tradeoff between speed and actuator usage (and often, between aggressiveness and structural excitation).

9. Slew-Rate Constraints

Spacecraft often impose explicit rate limits due to sensor tracking constraints (e.g., star trackers), structural loads, wheel momentum buildup, or pointing stability requirements. Constraints may be written as:

\[ \|\boldsymbol{\omega}\| \le \omega_{\max} \qquad \text{or} \qquad |\omega_i| \le \omega_{i,\max}. \]

A practical control strategy is to embed saturation or command shaping into the control law itself so the system:

• behaves like quaternion PD when far from limits,
• smoothly reduces authority as limits are approached,
• remains stable and predictable under bounded actuation.

10. Saturation and Cascade-Saturation Control: how to stay stable with actuator limits

Naively clipping a continuous torque command after designing a controller can degrade stability and performance. Instead, saturation is incorporated into the control structure so bounded actuation is handled in a stable, well-defined manner.

A useful interpretation is a three-phase rest-to-rest slew profile:

1. Acceleration phase: torque saturates to build angular rate efficiently.
2. Rate-limited phase: rate stays near an allowable level (coast-like behavior).
3. Deceleration phase: torque reverses (often saturated) to drive rate to zero at the target.

This viewpoint directly connects bounded-torque and bounded-rate constraints to maneuver execution structure.

11. Pulse-Modulated Attitude Control: when actuators are on/off or quantized

Some spacecraft cannot command smooth continuous torque. Thrusters produce discrete impulses, systems may be limited by a minimum impulse bit, and digital implementations can introduce quantization. In these settings, continuous commands must be converted into implementable pulse sequences.

Key features and consequences

• Discrete actuation: control is executed as on/off pulses rather than continuous torque.
• Minimum pulse width / minimum impulse bit: imposes a hard lower bound on achievable command resolution.
• Limit cycles near the setpoint: small oscillations can persist because arbitrarily small commands are impossible.

Hysteresis (Schmitt-trigger-like switching)

A common approach is to introduce hysteresis to avoid high-frequency chattering:

• switch ON when the error exceeds an upper threshold,
• switch OFF only after the error falls below a lower threshold.

Pulse-width and pulse-frequency modulation extend this idea by converting continuous control demands into pulse patterns while accounting for actuator dynamics and minimum pulse constraints.

12. Maneuver Design Workflow

1. Represent the commanded reorientation with quaternions (avoid singularities).
2. Define an attitude error quaternion $q_e$ and enforce a sign convention (short rotation).
3. Apply quaternion feedback reorientation using a PD-like structure in quaternion space.
4. Embed actuator constraints using saturation / cascade-saturation (avoid naïve clipping).
5. Interpret maneuver phases: accelerate $\rightarrow$ rate-limited (coast-like) $\rightarrow$ decelerate.
6. Select an objective: minimum time (bang-bang) vs reduced effort (smooth).
7. Choose implementation: continuous torque (wheels/CMGs) or pulse modulation (thrusters).

Summary: What D.2 established

D.2 framed large-angle slews as nonlinear, constrained maneuver problems and established the key design concepts:

• Large slews require nonlinear modeling on $\mathrm{SO}(3)$ beyond small-angle linearization.
• Quaternions provide a globally valid attitude representation without Euler-angle singularities.
• Rest-to-rest maneuvers regulate both orientation and angular velocity simultaneously.
• Quaternion attitude error with a sign convention ensures shortest-path rotation behavior.
• Quaternion-feedback reorientation uses a PD-like structure: $\mathbf{u}=-\mathbf{K}\mathbf{q}_e-\mathbf{C}\boldsymbol{\omega}$.
• Eigenaxis rotation supplies a clean geometric interpretation of minimum-angle reorientation.
• Minimum-time slews under bounded torque typically yield bang-bang switching behavior.
• Effort/fuel-optimal slews trade time for smoother, lower-peak actuation.
• Slew-rate and torque constraints should be embedded structurally using saturation-aware control.
• Pulse modulation enables discrete thruster-based control but introduces minimum-pulse limits and limit-cycle behavior.