FOR ROBOTAXI & AV DELIVERY OPERATORS

The evidence layer for the safety case you're already publishing.

If your team writes the document the regulator, the insurer, and the public read, FieldSpace generates the per-claim evidence that backs every behavioral acceptance criterion. The same input produces the same output for your auditors, every time.

Built for the operators publishing safety cases. Maps directly into the Section 2.3 acceptance criteria framework (Favaro et al. 2023). Same structure across Waymo, Zoox, Nuro, and Wayve safety material.

Who FieldSpace is built for

Operators who carry end-state safety liability and have to defend the safety case to regulators, the public, and the entities underwriting their fleet.

PRIMARY ICP
  • Robotaxi operators
    Driverless ride-hailing operations with formal safety cases under active publication and revision.
  • Autonomous delivery operators
    Sidewalk, road, and last-mile delivery fleets operating without an onboard human, with NHTSA exemption petitions and published safety reports.
  • Autonomous trucking operators
    Driverless freight operators publishing safety cases for hub-to-hub routes and operating under state-level autonomous trucking regulations.
  • Learned-policy operators
    End-to-end learned stacks where the policy is not deterministic, but the safety envelope around it has to be.
NOT THE PRIMARY ICP
  • ×
    Tier-1 ADAS suppliers
    Liability is offloaded to the OEM by contract, so a third-party evidence layer is not a budget-line. Available on request as a secondary engagement.
  • ×
    L2 driver-monitoring fleets
    If a human is the rated driver, the safety case is structured differently. FieldSpace can still produce useful evidence, but is not the primary product fit.
  • ×
    Pure simulation platforms
    FieldSpace is not a sim vendor. We sit alongside whatever sim the operator already uses, producing evidence the operator's safety team needs.

How FieldSpace fits inside the safety case

Section 2.3 of the Favaro 2023 paper defines five behavioral acceptance criteria dimensions and explicitly invites third-party evaluative methodologies to map into them. FieldSpace populates all five.

AC DIMENSIONFIELDSPACE EVIDENCEREPRODUCIBILITY ARTIFACT
Severity PotentialTTC-bound, predicted delta-V from potential fieldnuPlan TTC 0.9219
Conflict RoleInitiator / Responder identification from active repulsive fieldPer-frame JSON trace
Regulatory ComplianceSpeed-limit + drivable-area compliancenuPlan 1.000 / 1.000
Conflict AvoidanceRoute progress without conflict initiationnuPlan 0.9660
Collision AvoidanceNo-at-fault collision + early-warning lead timenuPlan 0.9766 / 13-15s lead
Functionality StatusCPU-only engine survives degraded computenfs-modulus crate
Event-Level ReasoningPer-frame trace across 60,019 + 4,550 framesPer-segment JSON
Aggregate-Level ReasoningRates across 64 + 50 + 10 scenario / clip setsManifest-pinned

Source: Favaro et al. 2023 "Building a Credible Case for Safety" (arXiv:2306.01917). Numbers anchored to public benchmark artifacts under reproducibility/.

What an operator receives

Three deliverable types. All reproducible to the bit by the operator's safety team, their auditors, and the underwriters pricing their fleet.

01

Per-claim evidence pack

For each behavioral claim in the operator's published safety case, a reproducibility-pinned evidence trace mapped to the relevant AC dimension. Cite-ready.

02

ODD coverage report

Quantitative coverage map across the operator's ODD, with gap analysis tied to the behavioral capability dimensions. Update when the ODD extends.

03

Counterfactual replay log

On an incident or disengagement, the deterministic should-have-happened trace plus the affected claim and the safety case revision delta.

A focused validation engagement, priced in writing

Three phases, fixed scope, fixed fee. Start with a paid evidence pack before any broader commitment.

Weeks 1-4

Safety case alignment

Map FieldSpace evidence to the operator's published safety case structure. Pick the AC dimensions and claims to cover first.

Weeks 5-8

Evidence generation

Run FieldSpace against the operator-supplied ODD spec and scenarios. Produce the per-claim evidence pack with reproducibility-pinned artifacts.

Weeks 9-12

Auditor walkthrough

Joint review with the operator's safety team, audit partner, or insurer. Confirm the evidence replays identically on their side.

Send the safety case. We'll map the evidence.

30-minute validation call first. We bring the AC-framework mapping and the benchmark trail. You bring the safety case structure and the questions your auditors are asking.