PRELIMINARY - WHOLE-BODY CONTROL CANDIDATE ONLY - NOT APPROVED FOR CONNECTION, POWERED TESTING, MOTION, WALKING, OR ENERGIZATION

The same complete robot clears the bounded controller screen.

The current tether model, bilateral 40 mm gait, contacts, current ceilings, inverse feedforward and installed 4:1 hip candidate are unchanged. The rejected 40/2 feedback factors are replaced by the frozen 8/0.8 candidate.

PASS

No rotary axis fails the bounded screen.

WS-L01

0.92 deg

maximum rotary error; saturation 0.00%.

PASS SUCCESSOR IDEAL-FIXTURE NUMERICAL SCREEN

WS-R01

0.92 deg

maximum rotary error; saturation 0.00%.

PASS SUCCESSOR IDEAL-FIXTURE NUMERICAL SCREEN

Why the knee and ankle hardware stays

The earlier forward simulation saturated six knee/ankle axes even though the prescribed inverse demand was already below their existing current-limited linear endpoints. The lower feedback gains remove that numerical contradiction without adding motors, pulleys, or distal mass.

AxisInverse peakCandidate endpointEndpoint marginDisposition
L_ANKLE_PITCH6.38 N·m7.57 N·m1.19×WITHIN LINEAR CURRENT ENDPOINT; CONTINUOUS CAPACITY OPEN
L_ANKLE_ROLL0.21 N·m6.06 N·m29.45×WITHIN LINEAR CURRENT ENDPOINT; CONTINUOUS CAPACITY OPEN
L_HIP_PITCH11.45 N·m17.17 N·m1.50×WITHIN LINEAR CURRENT ENDPOINT; CONTINUOUS CAPACITY OPEN
L_HIP_ROLL12.74 N·m17.17 N·m1.35×WITHIN LINEAR CURRENT ENDPOINT; CONTINUOUS CAPACITY OPEN
L_KNEE_PITCH8.68 N·m10.73 N·m1.24×WITHIN LINEAR CURRENT ENDPOINT; CONTINUOUS CAPACITY OPEN
R_ANKLE_PITCH6.38 N·m7.57 N·m1.19×WITHIN LINEAR CURRENT ENDPOINT; CONTINUOUS CAPACITY OPEN
R_ANKLE_ROLL0.21 N·m6.06 N·m29.44×WITHIN LINEAR CURRENT ENDPOINT; CONTINUOUS CAPACITY OPEN
R_HIP_PITCH11.46 N·m17.17 N·m1.50×WITHIN LINEAR CURRENT ENDPOINT; CONTINUOUS CAPACITY OPEN
R_HIP_ROLL12.75 N·m17.17 N·m1.35×WITHIN LINEAR CURRENT ENDPOINT; CONTINUOUS CAPACITY OPEN
R_KNEE_PITCH8.69 N·m10.73 N·m1.24×WITHIN LINEAR CURRENT ENDPOINT; CONTINUOUS CAPACITY OPEN

What this does not prove

This is still an ideal-pelvis-fixture simulation. It does not validate free balance, contact robustness, latency, state estimation, belt compliance, backlash, continuous motor torque, heat, electrical transients, fall restraint, or physical walking. No command from this package may be sent to hardware.

Evidence downloads

sequence results · all 25 axes · 50 Hz samples · gain definition · leg torque margins · MJCF · open holds