Abstract
A 3D pendulum consists of a rigid body, supported at a fixed pivot, with three rotational degrees of freedom. The pendulum is acted on by a gravitational force. Symmetry assumptions are shown to lead to the planar 1D pendulum and to the spherical 2D pendulum models as special cases. The case where the rigid body is asymmetric and the center of mass is distinct from the pivot location leads to the 3D pendulum. Full and reduced 3D pendulum models are introduced and used to study important features of the nonlinear dynamics: conserved quantities, equilibria, invariant manifolds, local dynamics near equilibria and invariant manifolds, and the presence of chaotic motions. These results demonstrate the rich and complex dynamics of the 3D pendulum.
Introduction
While the planar (1D) and spherical (2D) pendulums are staples of undergraduate physics, they are idealized models involving point masses. In contrast, real-world mechanical systems—such as satellites, underwater vehicles, and robotic limbs—are rigid bodies with asymmetric mass distributions and three degrees of rotational freedom. This research provides a rigorous mathematical treatment of the 3D pendulum, bridging the gap between simple textbook examples and the complex reality of rigid body dynamics under gravity.
By applying geometric mechanics and symmetry reduction, this study demonstrates that the 3D pendulum is far more complex than its lower-dimensional counterparts, exhibiting a rich variety of behaviors including stable manifolds, relative equilibria, and deterministic chaos.
Mathematical Framework: Geometric Mechanics
To describe the pendulum's attitude without the singularities (such as gimbal lock) associated with traditional Euler angles, the authors utilize Geometric Mechanics. The pendulum's configuration is represented using rotation matrices in the Special Orthogonal Group $SO(3)$.
Symmetry Reduction
Because the system is symmetric with respect to rotations about the vertical gravity axis, the authors employ Lagrange-Poincaré and Lagrange-Routh reduction. This process simplifies the 6-dimensional state space into lower-dimensional "reduced" models on $S^2$ manifolds. This reduction is essential for analyzing the underlying structure of the motion without the redundancy of the rotational symmetry.
Structure-Preserving Integration
To study long-term behavior and chaos, the study utilizes a Lie group variational integrator. Unlike standard numerical solvers, this method preserves the geometric properties of the physical system—such as total energy and momentum—ensuring that numerical artifacts do not obscure the true physical dynamics.
Equilibrium and Stability
One of the most striking differences between the 3D pendulum and simpler models is the nature of its equilibria:
- Equilibrium Manifolds: Unlike a simple pendulum with a single "down" point, the 3D pendulum possesses an entire manifold (a circle) of hanging equilibria. Any rotation of the body around the vertical axis while it is hanging down constitutes an equilibrium state. The authors prove that this hanging manifold is Lyapunov stable, while the corresponding "inverted" manifold is unstable.
- Relative Equilibria: The paper identifies "dynamic" equilibria where the pendulum rotates steadily at a fixed angle. These are categorized into five distinct cases based on the body's principal moments of inertia and the location of its center of mass.
The Transition to Chaos
The 3D pendulum is a Hamiltonian system that exhibits a classic transition from predictable to chaotic motion as energy levels increase. The authors visualize this transition using Poincaré maps (cross-sections of the trajectory):
- Low Energy: At low energy levels (e.g., $E = -2.65$), the motion is restricted to simple, predictable loops.
- High Energy: As energy increases (e.g., $E = 0$), the Poincaré map dissolves into a "cloud" of points. This dispersion is a signature of deterministic chaos, where the system's state becomes highly sensitive to initial conditions.
Additionally, the research derives the geometric phase, which quantifies how the pendulum's orientation changes after completing one full cycle of its reduced motion.
Conclusion and Future Work
This work establishes the 3D pendulum as a unifying model, showing that the classic 1D planar and 2D spherical pendulums are mathematically "special cases" that emerge under specific symmetry assumptions.
While this study assumes a frictionless pivot, it lays the theoretical groundwork for the Triaxial Attitude Control Testbed (TACT), a physical laboratory setup. Future research will focus on incorporating active control laws to stabilize the pendulum in non-trivial positions and accounting for the damping and friction present in real-world mechanical bearings. Experimental validation at the TACT facility will further bridge the gap between these theoretical proofs and practical aerospace and robotic applications.
Summary updated