AUTONOMOUS SECURITY ROBOT — SITE PILOT WORKSHEET
Yenra | 2026-09-07 | https://yenra.com/autonomous-security-robots/

Purpose: test observable site tasks and the complete human response path.
Define targets BEFORE reviewing results. This is a planning aid, not a
claim of legal compliance or demonstrated crime reduction.

Site / pilot owner / date / approved duration and hours:
Exact robot, software, sensors and configuration:
Required observable event(s) and excluded interpretations:
Route map revision, boundaries and required coverage intervals:
Vendor evidence for surfaces, grades, weather and route suitability:
Charging, connectivity, outages, recovery and alternative patrol:
Human alert reviewer / responder / escalation owner:
Baseline period, conditions and existing process:

METRICS — keep numerators, denominators and timestamps
Coverage = completed scheduled segments / scheduled segments.
Staged-event detection = detected eligible test events / eligible test events.
Actionable-alert precision = actionable alerts / all alerts reviewed.
Availability = patrol-ready minutes / required patrol minutes.
Multiply fractions by 100 for percentages; zero denominators are undefined.
Response times: retain event, alert, acknowledgment and response timestamps.
Separate staged-event results from ordinary alert results.

Requirement | Target | Test conditions | Evidence | Owner | Pass/Fail/Untested
           |        |                 |          |       |

Fictional example: 30 actionable out of 120 reviewed => precision 25%.
34 detected out of 40 staged events => detection 85%, six misses.
120 reviews * 2 minutes = 240 minutes = 4 hours of review.
This is sample arithmetic, not Knightscope or any vendor's measured result.

Break down findings by location, lighting, event type and time.
Record misses, manual interventions, consequential interactions,
accessibility observations, complaints and exclusions individually.
Stage events with authorized participants and an approved boundary.

DATA AND OPERATIONS REVIEW
Data collected / purpose / processing and storage location:
Authorized viewers and export rights / access logs:
Retention, deletion and contract-end handling:
Vendor secondary use or training use:
Applicable privacy/recording/accessibility/other review and owner:
Review labor, service, connectivity, facilities and backup patrol costs:

DECISION
Passed requirements and supporting evidence:
Failed or untested requirements:
Change / owner / repeat test:
Proceed / revise / end pilot, with rationale:
Re-evaluation triggers: route, configuration, model, staffing or policy change.

Reference: NIST AI RMF measurement guidance (voluntary framework)
https://airc.nist.gov/airmf-resources/airmf/5-sec-core/