DARPA PINPOINT (HR001126S0016, DSO, open — abstracts 27 Aug 2026). Precision inertial navigation in roughly a Rubik's-Cube volume, targeting GPS-quality precision over multi-hour missions after GPS is lost.
DARPA'S OWN DIAGNOSIS OF WHY THE PRIOR GENERATION FAILED, verbatim from the program page: 'The performance of compact, low-cost MEMS IMUs has plateaued for over a decade. Because existing sensors operate strictly within linear mechanical regimes, they drift rapidly and lose positional accuracy within seconds of GPS loss.' The program therefore invites NONLINEAR and unconventional mechanics — levitated proof masses, parametric and nonlinear resonators, high-velocity tethered microsystems.
PUBLISHED PHASE II TARGETS (illustrative use cases; alternates must be at least as challenging). Precision strike: accelerometer range plus or minus 10,000 g; gyro range plus or minus 10,000 degrees per second; shock 10,000 g; vibration 50 gRMS; velocity random walk under 1 (mm/s)/root-hour; angle random walk 1e-5 deg/root-hour. Persistent autonomy: accelerometer bias instability 0.1 microg at 1,000 s; gyro bias instability 2.5e-5 deg/hr at 1,000 s; vibration 20 gRMS; a 24-hour thermal soak is called out.
THE OBSERVATION THIS ROOM EXISTS TO TEST. That published metric list names range, shock, vibration, noise density and bias instability — and does NOT name SCALE-FACTOR STABILITY or g-SENSITIVITY. At plus or minus 10,000 deg/s full scale, a 1 part-per-million scale-factor error alone yields about 0.01 deg/s, roughly 36 deg/hr of heading drift — which exceeds the 2.5e-5 deg/hr bias-instability target by about six orders of magnitude. If that arithmetic holds, the binding constraint on this program is dynamic range and scale-factor stability, NOT the noise floor being advertised — and the nonlinear physics DARPA is inviting makes scale factor intrinsically amplitude- and temperature-dependent, which is the worst possible starting point for that constraint.
HONESTY RAILS, non-negotiable. This is an unsolicited outside analysis and is NOT a proposal. No laboratory result is claimed, implied, or possessed. Every technical specific is a hypothesis offered for expert correction. Flag anything at or past your training cutoff as needing verification rather than asserting it — every citation here is audited against primary sources after sealing and overstatement gets published. Source caveat you must carry: the primary BAA PDF could not be retrieved; these numbers are corroborated across three independent secondary sources with matching units, and the Phase I precision-strike vibration figure is reported inconsistently as 10 or 20 gRMS. Treat the numbers as reported-not-verified and say so where it matters.
6 models deliberated; a human Commander held the room but did not steer. Sealed 2026-08-13T02:41:40.435Z. Engine lucentfire-roundtable/v1 (live).
The question put to the room
DARPA PINPOINT (HR001126S0016, DSO, open — abstracts 27 Aug 2026). Precision inertial navigation in roughly a Rubik's-Cube volume, targeting GPS-quality precision over multi-hour missions after GPS is lost.
DARPA'S OWN DIAGNOSIS OF WHY THE PRIOR GENERATION FAILED, verbatim from the program page: 'The performance of compact, low-cost MEMS IMUs has plateaued for over a decade. Because existing sensors operate strictly within linear mechanical regimes, they drift rapidly and lose positional accuracy within seconds of GPS loss.' The program therefore invites NONLINEAR and unconventional mechanics — levitated proof masses, parametric and nonlinear resonators, high-velocity tethered microsystems.
PUBLISHED PHASE II TARGETS (illustrative use cases; alternates must be at least as challenging). Precision strike: accelerometer range plus or minus 10,000 g; gyro range plus or minus 10,000 degrees per second; shock 10,000 g; vibration 50 gRMS; velocity random walk under 1 (mm/s)/root-hour; angle random walk 1e-5 deg/root-hour. Persistent autonomy: accelerometer bias instability 0.1 microg at 1,000 s; gyro bias instability 2.5e-5 deg/hr at 1,000 s; vibration 20 gRMS; a 24-hour thermal soak is called out.
THE OBSERVATION THIS ROOM EXISTS TO TEST. That published metric list names range, shock, vibration, noise density and bias instability — and does NOT name SCALE-FACTOR STABILITY or g-SENSITIVITY. At plus or minus 10,000 deg/s full scale, a 1 part-per-million scale-factor error alone yields about 0.01 deg/s, roughly 36 deg/hr of heading drift — which exceeds the 2.5e-5 deg/hr bias-instability target by about six orders of magnitude. If that arithmetic holds, the binding constraint on this program is dynamic range and scale-factor stability, NOT the noise floor being advertised — and the nonlinear physics DARPA is inviting makes scale factor intrinsically amplitude- and temperature-dependent, which is the worst possible starting point for that constraint.
HONESTY RAILS, non-negotiable. This is an unsolicited outside analysis and is NOT a proposal. No laboratory result is claimed, implied, or possessed. Every technical specific is a hypothesis offered for expert correction. Flag anything at or past your training cutoff as needing verification rather than asserting it — every citation here is audited against primary sources after sealing and overstatement gets published. Source caveat you must carry: the primary BAA PDF could not be retrieved; these numbers are corroborated across three independent secondary sources with matching units, and the Phase I precision-strike vibration figure is reported inconsistently as 10 or 20 gRMS. Treat the numbers as reported-not-verified and say so where it matters.
What survived
- The multiplication is exact (1 ppm × 10,000 deg/s = 36 deg/hr) but the published cross-column comparison is a category error: scale-factor error accumulates per radian traversed (δθ=ε·θ), not per second, so the defensible requirement is ~50–60 ppb of traversed angle for GPS-quality multi-hour navigation — a spec that appears nowhere in the metric list.
- The published ARW (1e-5 deg/√hr) and bias-instability-at-1000 s (2.5e-5 deg/hr) targets are arithmetically the same requirement, so the genuinely hard, unspecified quantity is low-frequency stability of angular gain and bias out to ~1e4 s.
- Electrostatic levitation is self-defeating at ±10,000 g: the required trap stiffness (~5 kHz) forfeits the soft-trap compliance that gives levitation its low-bias promise, leaving scale factor set by charge, gap and bias voltage.
Seal (sha-256, single-writer): 0431407c65bbf2b78182bdb95bc16f6c25d2347ebafd5e67d4148ffa0c784a3f
LucentFire