Five milestones before general use. Measures are contacts, veering, percentage of preferred walking speed, and walking speed, all four reported because no single gold standard exists. IRB under 45 CFR 46 with accessible consent, an orientation and mobility specialist present, a trailing sighted guard, the right to withdraw, and data minimization. Blind co researchers are paid authors. Source: docs/15.
G0 bench: every subsystem meets its datasheet spec; latency budget measured; eye safety and battery certifications in hand.
G1 controlled course, sighted blindfolded then blind participants: obstacle contacts, veering, percentage of preferred walking speed (PPWS) against a mobility aid only baseline.
G2 real indoor and outdoor routes: contacts, PPWS, walking speed; glass and drop off cases explicitly tested.
G3 daily life wear over weeks (as a commercial compass belt trial did over seven weeks with eleven blind participants): comfort, wear time, confidence, incident log.
G4 readiness review: only after G0 to G3 pass are running mode and, separately, any assist frame work considered.
Measures: contacts, veering (lateral deviation from a straight path), PPWS (course speed divided by preferred speed, Clark-Carter 1986; Soong 2000), walking speed. No single gold standard exists, so all four are reported.
IRB: 45 CFR 46; accessible consent (Braille, audio, screen reader e consent), an impartial witness where required, an O&M specialist present, a trailing sighted guard, right to withdraw, data minimization. Blind co researchers are paid authors.
Samples: pilot cells of 6 to 12 for G0 and G1, scaling to a few dozen for G2 and G3.
Source: docs/15-evaluation-and-milestones.md, the engineering record. Every number on this page is read from that file. If a fact is not in the record, it is measured on the bench, or it stays off this site.