Technology 83 · CageSpar

San Francisco · Sep 22, 2026 · REK staging

Lapenna T800

Recurring job: build & save teleop bout sheets / kit compares for training notes

Viral SF cage night (Frankie Lapenna vs EngineAI T800, REK, teleop) is the spark — CageSpar is the notebook you reopen. Dial corners, park named sheets, share a URL, check ANSUR/NHANES percentiles. Keep the curiosity; skip the doom soundtrack.

Named bout library Shareable URL kits ANSUR II · NHANES · T800 specs

Octagon cage cam

Reach Δ —
Mass ratio —
Strike E —
Teleop lag —
Spar balance 50
Cage cam notes

Drag to orbit · pinch/scroll to zoom. Reduced-motion systems start paused.

Spar

—

Corner desks

Build the bout like a fight card, then save it as a named sheet for the next training block — red corner human, cyan corner humanoid, broadcast bar for latency.

T800 published specs load with the data pack.

Red corner

Human

Cyan corner

Humanoid

Control loop
Reach advantage — Who pokes first
Mass ratio — Bot ÷ human
Strike energy — ½mv² ballpark · educational
Latency tax — Reaction window cost

Exports include kit, estimates, checklist, sources, and share URL — reopen later as training notes. Educational only — not medical, legal, or fight-safety advice.

Bout sheet

Park two kits like tape between rounds. Side-by-side delta, no vibes-only arguing.

Tape A

Empty — dial corners, then save.

Tape B

Empty — dial corners, then save.

Delta

Need both tapes.

Bout facts

Bout sheet library

Name it, save it locally, reload next session. Share the URL so a coach or hobbyist lands on the same kit.

Stored in this browser only (localStorage). Export .txt/.json for backups that travel.

Public data lab

Where your human kit sits vs real published distributions — plus EngineAI T800 product specs as a cyan reference.

Height vs ANSUR II males—
Weight percentile—
Reach vs ANSUR span—
ANSUR II stature percentile ruler · red = you · cyan dash = T800
ANSUR II weight percentile ruler
ANSUR II span ruler (reach proxy)
ANSUR II male stature histogram (working DB) with kit markers
Stature × span cloud (ANSUR percentile pairs) · You vs T800
ANSUR II mean vs CDC NHANES 2021–23 adult mean (amber / cyan) · red = your kit

Dataset citations & caveats

Loading curated public summary tables…

Assumptions board

What the heuristics ARE

  • Relative kit compares: reach Δ, mass ratio, simple ½mv² strike ballpark, latency tax, spar-balance dial.
  • Teleop framing — operator in the loop, soft limits, padding flags.
  • Percentile context from published ANSUR II / NHANES summary tables for the human profile.
  • T800 numbers from EngineAI’s public product page (edition-dependent mass range).

What they are NOT

  • Medical, legal, or fight-safety certification.
  • Impact-lab / force-plate measurements.
  • Ground game, clinch, or cage-wall physics.
  • A claim that the viral bout was unsupervised fight AI.
  • Civilian “average American” sizing when using ANSUR II alone (military sample).

A teleoperated humanoid is a high-end puppet with motors, sensors, and a human brain on a delay. Autonomy would mean the robot choosing strikes without that live driver. The SF bout showed the first kind: impressive presence, human timing, and the familiar problems of mapping input onto a biped that still has to stay upright.

If you are building hobby platforms, celebrate the teleop milestone. Full fight autonomy is a different research mountain — not a weekend firmware flash. Fight fans can read the same footage for latency tells: late slips, careful clinches, soft recoveries.

Teleop vs autonomy

Operator in the loop is still hard robotics. It’s just not unsupervised fight AI — and that’s the honest framing.

Reach before bravado

A few centimeters decide who touches first. They matter more when one side’s footwork is latency-lagged.

Mass ≠ highlight KO

Inertia helps in the clinch and hurts acceleration. Soft limits and padding damp any joule ballpark.

Latency tax

Humans close distance in tenths of a second. 120–200 ms of lag eats parries and counters.

Garage → gym floor playbook

  1. Log round-trip latency on your radio or tether before you add punch speed.
  2. Compare reach at equal stance width before arguing mass.
  3. Treat mass ratio as a shove/clinch cue, not a betting line.
  4. Checklist: padding on, soft limits on, operator warm-up, E-stop within arm’s reach.
  5. Export two tapes from the bout sheet when the group chat starts “who wins.”

Scoreboard legend

Teleop
Human drives the robot live through a control link.
Latency
Round-trip delay from input to motion and feedback.
Reach
Wingspan / punch-extension proxy in centimeters.
Mass ratio
Humanoid mass ÷ human mass — clinch inertia cue.
Strike energy
Educational ½mv² ballpark using effective mass × speed cue.
Soft limits
Force/speed caps for demo-safe exhibition modes.
Balance assist
Stabilization that helps a biped recover from shoves.
Spar balance
0–100 evenness dial from reach, mass, latency — not odds.

Ringside FAQ

Was the T800 fighting on its own?

No. Public framing describes teleoperation — a human in the loop. Still hard robotics; not unsupervised fight AI.

Are the joule numbers lab measurements?

No. Educational ballparks from simple kinetic-energy heuristics. Useful for kit compare — not medical or legal evidence.

Can I train for a real fight with this?

Do not. This is a physics toy and bout explainer. Real sparring needs coaches, rules, and safety gear.

Why does latency dominate the HUD?

Humans close distance in tenths of a second. A 120–200 ms teleop lag eats the window to slip, parry, or plant a counter.

Anti-robot or pro-robot?

Neither. Builder / sports / entertainment tone — curious, competent, lightly humorous. Kit and systems, not apocalypse props.

Method & caveats

Reach advantage = robot reach − human reach. Mass ratio = robot mass ÷ human mass. Strike energy uses effective striking mass (~5–8% of body mass scaled by snap cues) and a speed estimate; soft limits and gloves damp it. Balance score blends normalized reach, mass, and latency. No ground game, no legal analysis, no live scrape.