Where This Sits — Prior Art and the Pulsed Offset Gyro
Before you claim a finding, check who got there first.
Context
Yesterday I published a GPU-accelerated simulation showing that pulsing the spin speed of counter-rotating masses on an offset circular race produces a persistent, tuneable, directional force bias on the frame. Constant spin averages to zero. Pulsed spin doesn't. 67,000 samples over 22 minutes, clean control, linear impulse growth.
The effect is real mechanics. It's also not new territory in the broad sense — people have been trying to extract directional force from rotating eccentric masses since 1959. What matters is whether the specific mechanism and the quantified characterisation add anything to the field.
So I went looking.
The Dean Drive (1959)
The grandfather of all mechanical inertial propulsion. Norman Dean patented counter-rotating eccentric masses (US 2,886,976) and claimed reactionless thrust. John W. Campbell promoted it in Astounding Science Fiction. It was debunked: the apparent thrust came from anisotropic friction — stick-slip interaction with the surface beneath the device.
Dean used constant-speed rotation. He did not pulse the angular velocity. The device produces symmetric oscillating forces that average to zero in free space. On a surface with direction-dependent friction, the oscillation "ratchets" in one direction. That's vibration-driven locomotion, not reactionless thrust.
The Dean Drive is important context because it established the pattern: someone builds an eccentric mass device, claims thrust, the community investigates, and the effect turns out to be coupling to the test surface. The physics is real — the conclusion about reactionless propulsion is not.
Thornson Drive and PIE (1986–Present)
Brandson Thornson (US 4,631,971, 1986) is the closest ancestor to the pulsed offset gyro. He used eccentric masses on an epicyclic gearset — planet gears orbiting a sun gear. The gear geometry inherently creates variable angular velocity: masses speed up at certain phases and slow down at others. The device reportedly propelled a canoe.
Bryan St. Clair has continued this work with the Pulsed Inertial Engine (PIE), now on iteration PIE7. St. Clair explicitly uses electronic control to boost speed during specific phases and harvests thrust when masses hit mechanical stops. His device produces two forward impulse events per orbit.
The key difference from my approach: Thornson/PIE achieves variable speed through gear geometry and mechanical impacts. My concept uses a geometrically simpler system — a circular race offset from the spin axis, with the motor directly pulsing the spin speed via omega(t) = omega_base * (1 + pulse * cos(theta)). No gears, no stops, no impacts. The symmetry break comes purely from the speed profile.
Same underlying principle. Different mechanism.
Controlled Unbalance Propulsion — CUP (2024)
This is the most directly relevant academic work. A 2024 paper in MDPI Applied Sciences describes using an airborne gyro rotor spinning at constant speed, with short-duration mass unbalances introduced at specific angular positions during each revolution. The centrifugal force pulses are directed in a chosen direction. The time-averaged force is non-zero because the unbalance is only active during a selected angular range.
CUP achieves rectification by toggling an unbalance on and off while the rotor spins at constant speed. My approach achieves rectification by varying the spin speed while the offset (unbalance) is permanently fixed.
Same physics principle — breaking the time symmetry of centrifugal force. Different control variable. CUP modulates the geometry. I modulate the velocity.
The CUP paper is peer-reviewed and published. It claims theoretical performance up to orbital velocities, though this requires coupling to a medium.
Gerocs & Gillich — Balls on Eccentric Trajectories (2020)
Gerocs, Gillich et al. published "A Multibody Inertial Propulsion Drive with Symmetrically Placed Balls Rotating on Eccentric Trajectories" in MDPI Symmetry. Eight steel balls on an eccentric path, two counter-rotating symmetric drivers. Their simulation showed 0.22mm displacement in 0.2 seconds.
This is geometrically close to my setup. The critical difference: they used constant angular velocity. They did not pulse the spin speed. Their net motion comes from collision dynamics and friction interaction, not from speed modulation. They did not identify or quantify the force bias that emerges from pulsing.
A 2023 follow-up tested the concept on a physical wheeled vehicle.
Provatidis — The Formal Proof (2025)
Christopher Provatidis at the National Technical University of Athens is the most prolific academic researcher in this space. His 2024 review paper covers 536 references across the entire field of inertial propulsion.
His 2025 paper is the definitive word on the vacuum case: "On the Incapability of Inertial Forces as a Means of Repeated Self-Propulsion of an Object in a Vacuum." Formal mathematical proof. Inertial drives cannot produce net displacement of the centre of mass in free space. Period.
But — and this is the important part — Provatidis also demonstrated (2014) that contra-rotating masses on a floating object can produce motion through water resistance. The internal forces couple to the external medium. Blasius drag converts asymmetric oscillation into net displacement.
The conclusion is precise: these devices do produce real forces on the frame. They can produce real locomotion when coupled to a medium. They cannot self-propel in vacuum.
My simulation confirms the first point. The second is the next experiment. The third was never in question.
Vibration-Driven Locomotion — The Legitimate Physics
This is the established, respectable field that the pulsed offset gyro actually belongs to. The core principle: internal periodic excitation combined with asymmetric system properties produces net locomotion. The asymmetry can be in friction (different forward vs. backward resistance), fluid drag, or electromagnetic interaction.
The field is well-published in major journals — Nonlinear Dynamics, Nature Communications, PLOS ONE. Real applications exist: capsule robots for medical endoscopy, microrobots, vibration-driven walkers.
The pulsed offset gyro is a vibration-driven locomotion mechanism. The pulsed speed creates a time-asymmetric force waveform. Coupling to any anisotropic medium converts that into net thrust. The contribution — if there is one — is the specific geometry and the quantified characterisation of how pulse strength maps to force bias.
What Appears to Be New
After searching the published literature, patents, and maker community:
-
The specific combination of a fixed offset circular race with directly pulsed angular velocity as the symmetry-breaking mechanism. Thornson uses gear geometry. CUP toggles the unbalance. This approach keeps the geometry fixed and pulses the speed. A distinct third mechanism for centrifugal force rectification.
-
Quantified force bias as a function of pulse strength for this geometry. The clean A/B comparison — pulse 0.0 averages to zero, pulse 0.5 produces -3.05 N bias, higher pulse produces proportionally more — with 67,000 samples over 22 minutes. I haven't found this characterisation published for the offset-race geometry.
-
GPU-verified numerical evidence with a clean control. The simulation runs on CUDA via Taichi with real-time visualisation. The impulse growth is linear over 22 minutes with no sign of convergence to zero.
What is NOT new: the general principle that time-asymmetric internal oscillation coupled to an external medium produces net locomotion. That's textbook vibration-driven locomotion.
What I Learned
-
The field has a credibility problem. Decades of "reactionless thrust" claims have made the physics community allergic to anything involving eccentric masses. The legitimate vibration-driven locomotion work is well-respected, but anything that smells like a Dean Drive gets dismissed. Positioning matters.
-
The vacuum question is settled. Provatidis (2025) proved it formally. No inertial drive can self-propel in vacuum. This is not a limitation of engineering — it's a theorem. Accepting this upfront is what separates serious work from crank physics.
-
CUP is the closest published work. The 2024 MDPI paper uses a different mechanism (toggling unbalance vs. pulsing speed) but the same underlying physics. Referencing it properly is essential.
-
The coupling question is everything. The force bias exists. The impulse grows linearly. The next step is demonstrating net locomotion through a medium — and measuring efficiency. That's where this either becomes useful engineering or stays as a simulation result.
-
Timestamp your work. This blog, the git history, and the published posts create a verifiable record of when this analysis was done and what it showed. If the specific mechanism and characterisation are genuinely novel, the timestamps matter.
What's Next
Build the coupling experiment. Put the pulsed offset gyro in a simulated fluid and measure whether the asymmetric force produces net displacement. If it does, measure efficiency — how much input energy goes to locomotion vs. waste heat. Compare to a simple vibration motor on an anisotropic surface.
The physics says it should work. Provatidis demonstrated it with contra-rotating masses in water. The question is whether the pulsed offset geometry is more efficient, more tuneable, or more practical than existing vibration-driven locomotion mechanisms.
The simulation pipeline is ready. The GPU is warm. The literature review is done.
Time to put it in water.
Prior art searched. Position established. No claims beyond the evidence. Built with Claude Code. Published at indigo-nx.com.