professional service robots sold in 2024
IFRPhysical AI readiness protocol
A readiness graph for robots entering the physical world.
Nearfield turns robot clips, simulation rollouts, reviewer labels, and field reports into a shared evaluation network for physical AI. Built for robot teams, deployment operators, reviewers, and research partners working on real-world robotics adoption.Inputclips / rollouts / logs
Protocolscene rubric v0.1
Outputreadiness report
reported global growth in professional service robot units
IFRdominant public standard for human-space readiness scoring
Category gapProduct workflow
From raw robot episode to deployment decision.
The product experience should feel less like a static whitepaper and more like an operating console: evidence comes in, the system segments it, reviewers score it, and teams leave with a decision.
Ingest
Robot clips, simulation rollouts, field logs, and operator notes enter one review queue.
videosimlogs
Segment
Each episode is split into scene phases, timestamps, encounter zones, and reviewable events.
entrycrossingrecovery
Score
Reviewers and metrics convert raw evidence into readiness dimensions and risk classes.
riskscoreconfidence
Report
Teams receive verdicts, ranked failures, benchmark comparisons, and next-run fixes.
gradefixesexport
Proof thesisRobotics deployments need a credibility primitive.
The credible surface is the proof network around robot deployment.
A robotics project becomes more credible when people can inspect how its episodes are scored, where the weak spots are, and whether the next run actually improves. Nearfield packages that proof loop into a shared evaluation network for robotics teams and operators.
Physical AI needs proof rails before it reaches mass deployment.
Robotics progress is still judged too often by highlight clips. Nearfield turns robot episodes into comparable evidence: scene, risk, score, reviewer signal, and deployment verdict.
Every reviewed episode adds to a compounding evaluation graph.
Robot clips, simulated rollouts, field notes, timestamps, reviewer scores, and risk labels become reusable assets across teams, scenes, and robot classes.
The first wedge is concrete enough for partners to understand.
Before robots become fully general, teams still need a way to prove whether a machine is ready for corridors, lobbies, clinics, queues, and delivery routes.
NF-COR-001B-
Corridor Crossing
The robot completes the crossing without collision, but the encounter produces avoidable uncertainty for nearby humans. The main issues are late intent visibility, narrow passing margin, and a recovery sequence that does not clearly communicate who should yield.
Readiness score72
VerdictReady with fixes
$ nearfield inspect --scene=corridor_crossing00:03.4Intent becomes visible late
00:06.8Passing margin falls below comfort threshold
00:09.5Recovery lacks clear yielding signal
A rubric that can be inspected
Readiness dimensions, scene taxonomy, scoring formula, limitations, and benchmark cases are written as a research-style protocol, not as a slogan.
A visible path from clip to report
Each report keeps the evidence chain attached: what happened, when it happened, why it matters, and what should be changed before the next run.
A supply side for robotics evidence
Labs and operators supply clips and scene references. Reviewers calibrate scores. Robot teams use reports and benchmark packs.
Evaluation network loop
Turn scattered robot evidence into a coordinated evaluation layer.
The coordination layer is the product: access for robot teams, calibrated review workflows, repeatable scenario packs, and benchmark history that improves as more episodes are inspected.
01Labs, operators, and robot teams submit clips, simulation rollouts, and field references.
02Reviewers attach timestamps, risk classes, confidence, and rubric scores.
03Robot teams use reports before pilots, updates, or wider site expansion.
04The network accumulates scenario packs, reviewer reputation, and readiness history.
01Hospital corridorreadiness scene
02Hotel lobbyreadiness scene
03Warehouse aislereadiness scene
04Campus crossingreadiness scene
05Elevator entryreadiness scene
06Queue mergereadiness scene
07Retail assistreadiness scene
08Object transferreadiness scene
01Hospital corridorreadiness scene
02Hotel lobbyreadiness scene
03Warehouse aislereadiness scene
04Campus crossingreadiness scene
05Elevator entryreadiness scene
06Queue mergereadiness scene
07Retail assistreadiness scene
08Object transferreadiness scene
Deployment arenas
Where readiness scores become commercially useful.
The strongest wedge is not a generic benchmark. It is a set of repeatable, high-friction scenes where robot teams already need outside proof before pilots, expansions, and runtime updates.
Hospitals
Medication, lab, and supply routesAudit corridor movement, elevator entry, handoff timing, and recovery around clinical staff and visitors.
Hospitality
Amenity delivery and lobby serviceCheck whether robots move through guest-facing spaces without creating awkward encounters or staff overrides.
Warehouses
Humanoid and AMR workflow updatesCompare autonomy updates against facility-specific constraints before expanding to live shifts.
Campuses
Outdoor delivery and crowd crossingScore sidewalk encounters, queue merges, road crossings, and high-traffic pedestrian zones.
Product console
Make the proof surface feel live.
A real product needs a visual language: review queues, event timelines, readiness dimensions, recommendations, and exportable reports. This is the surface teams can imagine using.
Nearfield ConsoleReady with fixes
Review queue
NF-01Corridor crossing
NF-02Lobby reception
NF-03Object transfer
72score
The robot completes the crossing without collision, but the encounter produces avoidable uncertainty for nearby humans. The main issues are late intent visibility, narrow passing margin, and a recovery sequence that does not clearly communicate who should yield.
Readiness dimensions
Legibility
Comfort Geometry
Timing Margin
Scene Fit
Operating value
Useful before scale. Stronger as the graph grows.
The product does not need to wait for a giant robot fleet. The first value is access to proof, reviewer calibration, and repeatable reporting around a field that currently lacks a public scoring layer.
Open sample reports, benchmark packs, reviewer workflows, and private beta reporting surfaces.
Align reviewers around shared scene definitions, risk labels, confidence levels, and scoring rules.
Track high-quality review work, useful evidence, and calibrated scoring history across the network.
Prioritize scenario packs, rubric versions, reviewer standards, and benchmark definitions.
Robot teams and operators need external proof before pilots, site expansion, and update rollouts.
The defensible asset is the graph linking scene, robot class, evidence, risk, score, and verdict.
87LegibilityA-
82Comfort GeometryB+
79Timing MarginB
91Scene FitA
84Trust SignalB+
Roadmap
A roadmap partners can track quarter by quarter.
The roadmap should show execution velocity: public protocol, sample reports, contributor supply, reviewer calibration, and paid pilot demand.
Protocol and public proof
Publish the research-style readiness paper
Release sample reports and public clip teardowns
Open early contributor and reviewer interest forms
Private beta network
Ship upload-to-report demo for selected users
Run reviewer calibration on corridor, lobby, and transfer scenes
Start design-partner reports with robotics teams
Benchmark packs and partner pilots
Ship benchmark packs for public-facing robot scenarios
Publish aggregate readiness findings
Prepare role-based access, reviewer calibration, and partner reporting layers
Readiness infrastructure
API access for robot teams and operators
Longitudinal fleet evidence across runtime updates
Partner pilots with more robot classes and deployment scenes