MakarovSuites
English
EnglishDeutschРусскийУкраїнськаFrançaisEspañolItalianoPolski
Contact & licences

Engineering

Parallel motion platform

Fly a take-off in a full-motion simulator in a dark hall: the cockpit capsule rides on six blue-lit actuators of a Stewart platform before a glowing projection dome. The washout turns the aircraft’s push of 0.29 g into a quick surge that comes back to centre as a0 t² e^(-ωt)/2, while tilt coordination pitches the cabin up toward asin(a/g) at 3° per second until gravity gives the pilot the push - and the horizon on the dome rises with the cabin. The six leg lengths follow from the pose in closed form, and the motion computer recovers the pose from them by Newton’s method. Slow the washout and the legs run out of stroke, their lamps turning amber and red; switch tilt off and the push is lost. Hold a pose with an offset payload and read the forces J^-T gives; move the platform joints toward the base pairs and the forces soar near the singularity. The actuators’ work is followed into the cabin’s height and motion.

Parallel motion platform — Actual application image
Actual application image

What happens

A Stewart–Gough platform: six actuators of 2.3–3.3 m join base joints, in pairs on a circle of 2.2 m (±8°), to platform joints on a circle of 1.5 m (±50° by default), so each leg crosses to the neighbouring pair; at mid-stroke the upper joint plane stands 2.378 m above the lower. Inverse kinematics is closed form, l_i = o + R·p_i − b_i; the motion computer recovers the pose from the six lengths by Newton’s method with the exact derivative of the yaw-pitch-roll rotation (two or three iterations a frame, error below 10^-12 m). The Jacobian rows [n_i, (a_i − c) × n_i] give the leg rates and, through J^-T, the forces that hold a wrench; its condition number (angular columns scaled by the platform radius) is √(λmax/λmin) of JᵀJ by Jacobi rotations: 2.29 at neutral, 52 with the joints at 10°, infinite when the hexagons are similar. The washout: surge, heave and sway pass x(s) = s·a(s)/(s + ω)³ with ω = 2 rad/s, so a step a0 gives a0·t²·e^(−ωt)/2 and the cabin returns to centre; tilt coordination pitches and rolls the cabin toward asin(f/g) (at most 20°) with a 1 s lag and a 3°/s rate limit; the aircraft’s pitch and roll rates pass s/(s + 0.5). Each filter is advanced exactly with its input held over the 1 ms sub-step. The cabin of 7 t (radii of gyration 1.0, 1.15 and 1.05 m, centre of mass 1.1 m above the joints) is a rigid body on six force-controlled actuators: the feed-forward J^-T·(m(a_ref + g), 0) plus a servo of 3 MN/m and 100 kN·s/m on each leg’s length error to the reference legs, clamped 2 cm inside the stroke. Euler velocity, trapezoid position, an implicit-midpoint body rate and the exact rotation of the midpoint rate make the actuators’ work equal the kinetic plus potential energy to round-off.

Try this

Watch the take-off: the surge comes back to centre while the tilt builds the push, until the felt surge equals the aircraft’s; switch tilt coordination off, or slow the washout until the lamps turn red; choose Turbulence; choose Hold a pose, pitch with the payload moved forward and bring the joint angle toward 10° to watch the condition number and the forces climb.

The model and its limits

Massless actuators with ideal force servos and no force limit (the largest force is only reported), point joints without angle limits, no collision check of the legs or of finite joints; the cabin a rigid body of fixed inertia. The washout works in hall axes without the aircraft’s attitude transformation and with fixed gains; the felt force is taken at the centre of mass and the vestibular thresholds enter only as the tilt rate limit. A clamped leg stops 2 cm short of its end and the cabin goes where the six lengths put it, so the cue turns false. Near the architecture singularity the servos still hold the pose, slowly and with little damping. The dome is a fixed screen with the eye at its centre.

Equations used in the module
Equations
l_i = o + R(φ,θ,ψ)·p_i − b_i,  L_i = |l_i|,  FK: q ← q − (∂L/∂q)^-1·(L(q) − L_measured)
dL/dt = J·(v, ω),  J_i = [n_i, (a_i − c) × n_i],  f = J^-T·(m(a_ref + g), 0),  κ = √(λmax/λmin) of JᵀJ
x(s) = s·a(s)/(s + ω)³:  a0 step → x = a0·t²·e^(−ωt)/2,  θ → asin(f/g) at ≤ 3°/s,  θ_r(s) = q(s)/(s + k)
m·dv/dt = Σ f_i·n_i + m·g,  I·dΩ/dt + Ω × IΩ = Σ r_i × f_i·n_i,  W_act = ΔE_k + ΔE_p

Related experiments

Back to catalogue
Conveyor sorting station — Actual application image

Engineering

Conveyor sorting station

Run an airport baggage hall on the night shift. Duffels, hard-shell cases, backpacks, heavy cases and golf bags wait on the induction belt until the gap control releases them; each slips on the main belt at mu g until it runs at belt speed, V/(mu g) later. Under the red laser of the scanner arch a photo-eye times the bag, and the controller fires the pusher of its chute at the computed instant: the paddle drives it sideways against the Coulomb friction of the belt, mu m g, while the belt carries it on, and it slides down a steel chute to its cart at the closed-form speed. Flood the line and the queue grows at the closed-form capacity; close the gap and bags crash into pushers still across the belt; weaken the pusher below mu m g and heavy cases do not move at all; add latency and the long golf bags miss until the eye setting moves by V tau.

Reciprocating feeder — Actual application image

Engineering

Reciprocating feeder

Feed metal blanks from a magazine with a reciprocating carriage. Load, friction and drive force determine whether the queue advances or stalls.

Rotary packaging line — Actual application image

Engineering

Rotary packaging line

Coordinate a six-pocket indexing table and a force-limited sealing head. Timing, payload and motor torque determine throughput, rejects and jams.