End-to-End AV Knowledge Gap Backlog

This backlog consolidates the 2026-05-08 parallel gap audit across the end-to-end autonomous vehicle knowledge architecture. It complements the dedicated SLAM coverage audit and perception coverage audit, which already track method-level gaps in those two libraries.

For ongoing discovery inputs, use the Active Frontier Source Registry. It keeps source monitoring separate from this backlog: candidates discovered there should be verified, deduplicated, and then routed here only when they become durable cross-domain gaps.

Audit Method

Six parallel research agents audited the corpus by architecture domain:

Agent scope	Repo areas audited
Foundations	10-knowledge-base/
Platform	20-av-platform/
Autonomy stack	30-autonomy-stack/ outside the existing perception and SLAM method libraries
Runtime and cloud	40-runtime-systems/, 50-cloud-fleet/
Safety and validation	60-safety-validation/, related 90-synthesis/ risk docs
Operations and industry	70-operations-domains/, 80-industry-intel/

One autonomy-stack agent exceeded context because 30-autonomy-stack/ is the largest section, so the scope was split into two narrower replacement agents: planning/control/V2X and world-models/VLA/E2E/simulation/maps.

Priorities:

Priority	Meaning
P0	Structural gap that blocks the repo from being a generic end-to-end AV knowledge base. Create or expand before calling the section broadly complete.
P1	High-value gap that improves currentness, deployability, or cross-domain transfer.
P2	Useful follow-up, refresh, or extension after the P0/P1 backlog is under control.

Structural Findings

Finding	Evidence	Implication
Generic operations scope was underbuilt	00-start-here/repo-map.md says 70-operations-domains/ should cover airside, indoor warehouse, outdoor campus, road AV, and deployment playbooks, but README.md and INDEX.md were airside-heavy.	P0 first wave added warehouse, logistics yard, port, mining, agriculture, construction, robotaxi, trucking, and sidewalk delivery robot operations files.
Platform tree promised power and thermal but lacked directories	README.md and 00-start-here/repo-map.md describe power and thermal systems, while 20-av-platform/ had compute, sensors, drive-by-wire, and networking.	P0 first wave added power/electrical, diagnostics, ruggedization, and close-range safety sensing. P1 still tracks vehicle-level thermal management.
Foundations needed reusable primers	10-knowledge-base/ was strong on selected deep dives, but lacked coordinate frames, Bayesian filtering, vehicle dynamics, planning taxonomy, calibration fundamentals, and sensor measurement models.	P0 first wave added five foundation primers. The 2026-05-09 loop added LiDAR, camera, IMU, GNSS/RTK, radar, event/thermal, wheel odometry, time synchronization, and calibration observability fundamentals.
Runtime and cloud needed operations discipline	Existing files covered telemetry, OTA, data pipelines, and MLOps, but not fleet SRE, incident command, SUMS governance, map ops, data governance, and runtime security operations.	P0 first wave added operator-facing runtime/cloud playbooks and evidence models.
Safety needed traceable evidence packages	Safety content was deep, but incident reporting, living safety-case traceability, and EU compliance dossiering were scattered.	P0 first wave added incident reporting, safety-case evidence traceability, and EU AI Act/Machinery/CRA dossier files. P1 still tracks ISO 3450x evidence, HARA/STPA, PLd/SIL, and ML assurance governance.

P0 First-Wave Completion (2026-05-09)

The P0 rows below were promoted into first-class research files by seven writing agents: six parallel workers plus one focused delivery-robot follow-up. The table remains as the provenance record for what was promoted. The next active queue is P1.

Perception, SLAM, and Sensor Loop (2026-05-09)

A follow-up loop focused specifically on method-level perception, method-level SLAM, and sensor fundamentals for perception, SLAM, and mapping. The loop is tracked in Continuous Research Loop.

Track	Files promoted
Perception methods	SplatAD, GaussianFormer, GaussianOcc, streaming Gaussian occupancy, Cam4DOcc, StreamingFlow, Sparse4D, TacoDepth, and RaCFormer.
SLAM methods	MOLA, KISS-SLAM, KISS-Matcher, LVI-SAM, FAST-LIVO/FAST-LIVO2, R2LIVE/R3LIVE, Splat-SLAM, S3PO-GS, Gaussian-LIC, GS-LIVM, VIGS-SLAM, dynamic 4D Gaussian SLAM, and RadarSplat-RIO.
Sensor and estimation fundamentals	LiDAR, camera, IMU, GNSS/RTK, radar, event/thermal, time synchronization, multi-sensor calibration observability, wheel odometry, visible-camera hardware, and IMU/GNSS/RTK hardware.

The next active queue is no longer just P1 cross-architecture work. It also includes method-library loops for temporal occupancy, radar-camera/4D-radar perception, robust SLAM backends, alternative localization sensors, and sensor calibration operations.

Cross-Architecture Gap Fill (2026-05-09)

A second broad web sweep and fill-in research wave re-audited the non-perception/SLAM knowledge base. The source-backed results are tracked in Cross-Architecture Knowledge Base Gap Fill. A follow-up writing wave promoted 30 atomic pages across safety, runtime/data, platform, autonomy evaluation, regulation/deployment, and company evidence.

Priority	Queue	Promoted files
P0	Safety standards and evidence	airside-agvs-regulatory-approval-playbook.md, safety-functions-pld-sil-validation.md, ml-assurance-data-governance.md, iso-ts-5083-ads-safety-vv.md, iata-ahm-908-autonomous-gse-standards.md, and safety-critical-scenario-libraries.md.
P0	Runtime data and release governance	model-governance-release-evidence.md, data-catalog-lineage-quality-ops.md, replay-scenario-mining-ops.md, edge-runtime-supervision-config-management.md, and active-labeling-budget-ops.md.
P0	Platform integration	zonal-ee-harness-connectors.md, vehicle-middleware-dds-someip-zenoh.md, safety-certified-runtime-compute.md, vehicle-thermal-management.md, and av-data-recorder-dssad-hardware.md.
P0	Autonomy evaluation	closed-loop-vlm-vla-evaluation.md, risk-forecasting-long-tail.md, real-to-sim-closed-loop-benchmarks.md, and natural-language-cooperative-autonomy.md.
P0	Regulation, deployment, and company evidence	2024-2026-autonomy-deployment-index.md, cross-domain-autonomy-regulatory-map.md, airside-agvs-faa-caas-regulatory-map.md, us-road-ads-approval-reporting-nhtsa.md, eu-ads-type-approval-2022-1426-2026-481.md, plus five company deployment pages.
P1	Remaining foundations	real-time-scheduling-wcet-mixed-criticality.md, rare-event-statistics-safety-validation.md, fault-trees-stpa-hazard-analysis.md, and odd-scenario-ontology-coverage.md remain queued because 10-knowledge-base pages require visual assets and taxonomy assignments in the same change.

First-Principles Foundations Loop (2026-05-09)

Five web/discovery rounds audited probability/statistics, nonlinear optimization, numerical linear algebra, data association, geometry, mapping, sensors, signal processing, and statistical validation. Five writing agents then promoted the highest-priority gaps into 33 atomic knowledge-base files.

Track	Files promoted
Probability and statistics	Gaussian noise/covariance/information, Mahalanobis and chi-square gating, likelihood/MAP/MLE, robust statistics/RANSAC, and mixture models.
Optimization and solvers	Nonlinear least squares, Gauss-Newton/Levenberg-Marquardt/dogleg, trust regions and line search, Jacobians/autodiff/manifold linearization, and Ceres/GTSAM/g2o solver patterns.
Numerical linear algebra	Cholesky/LDLT, QR/SVD, Hessian conditioning, sparse fill-in/orderings, square-root information/covariance recovery, and Schur/marginalization/PCG.
Geometry, mapping, and geodesy	Lie groups, projective geometry/PnP/triangulation, ICP/GICP/NDT registration, correspondence search structures, occupancy Bayes/evidential/dynamic grids, and map projections/datums.
Association, filters, sensors, and validation	Data association, JPDA/MHT/RFS, information filters/smoothers, particle filters, sensor likelihoods, FFT/filtering, radar ambiguity, CFAR, timestamping, and benchmarking statistical validity.

LIORNet, Removal, and ML Foundations Loop (2026-05-09)

The latest loop followed the broader "approach C" structure: treat removal as a system-level capability rather than one method page. It added method-level LiDAR denoising, classical artifact filtering, weather datasets, map-cleaning methods, safety validation, and machine-learning first principles so learned removal methods can be read from architecture down to training mechanics.

Track	Files promoted
Learned LiDAR denoising/removal	LIORNet, LiSnowNet, SLiDE, TripleMixer, 3D-KNN Blind-Spot Desnowing, 3D-OutDet, AdverseNet, and DenoiseCP-Net.
Broad removal mechanics	Classical LiDAR Outlier Removal, LiDAR Weather Artifact Removal, LiDAR Artifact Removal Techniques, and LiDAR Ghost and Multipath Artifacts.
Weather benchmarks	Weather Robustness Datasets, WADS, CADC/CADC+, SemanticSTF, REHEARSE-3D, RainSense, SemanticSpray, RADIATE, and Seeing Through Fog/DENSE.
Safety and map cleaning	LiDAR Artifact Removal Validation, ERASOR, Removert, and LiDAR Map Cleaning and Dynamic Removal.
Machine-learning foundations	Machine Learning Foundations Overview, perceptron/logistic/MLP/backprop/optimization primers, CNN/RNN sequence-model primers, transformer and vision-transformer first principles, self-supervised learning, foundation-model training, JEPA latent predictive learning, and world-model first principles.

Dynamic/Static Removal and ML Objectives Loop (2026-05-09)

The next wave expanded removal beyond weather/noise and into persistent map correctness: dynamic-object trails, parked or moved objects that should not become static infrastructure, map-change benchmarks, scene-flow evidence, and safety validation for false deletion. In parallel, the ML foundation ladder gained objective, tokenization, calibration, and world-model evaluation notes so modern removal and scene-flow methods can be reviewed from first principles.

Track	Files promoted
Dynamic/static map cleaning	MapCleaner, ERASOR++, 4dNDF, FreeDOM, STATIC-LIO Dynamic-Point Removal, and Dynamic Map Cleaning Benchmarks.
Scene-flow and motion/static evidence	MotionSeg3D, MambaMOS, Neural Scene Flow Priors, Scene Flow for Dynamic Object Removal, Scene-Flow Datasets and Benchmarks, and Moving/Static Separation MOS Datasets.
Map-change and safety validation	Moved-Object and Map-Change Datasets, Occupancy-Flow and 4D Occupancy Benchmarks, and Airside Dynamic Map-Cleaning Benchmark.
ML objective/evaluation foundations	Autoencoders/VAEs, Contrastive InfoNCE, Masked Modeling, Energy-Based Models, Tokenization, Positional Encodings, S4/Mamba, Diffusion/Score/Flow Samplers, Multi-Task Losses, Evaluation/Calibration/Leakage, and World-Model Evaluation.

Web Gap Expansion Loop (2026-05-09)

Five web-search scouts re-audited perception, SLAM, world-model/neural-field, validation, and knowledge-base gaps. Six writing agents then promoted the selected gaps into 31 source-backed files.

Track	Files promoted
Perception radar, FMCW, and temporal methods	CVFusion, 4D Radar-Camera Occupancy, POD FMCW LiDAR Predictive Detection, Cross-Domain LiDAR Scene Flow, and TrackOcc.
Dynamic 3DGS/neural fields and world models	DrivingGaussian, HUGS, SplatFlow, DistillNeRF, Self-Supervised Occupancy Flow, and UniScene.
SLAM robust backends and lifelong localization	Robust PGO / GNC / riSAM, Certifiable PGO, Kimera-RPGO / PCM, Distributed Multi-Robot PGO, LT-Mapper / Khronos, RTMap / DUFOMap, GPR Localization, Radar Teach-Repeat, and MOVES.
Knowledge-base foundations	Probabilistic Graphical Models and Message Passing, Information Theory for Perception and ML, Uncertainty Quantification, Calibration, and Conformal Prediction, Constrained Optimization, MPC, and iLQR, and MDP, POMDP, Belief-Space, and RL First Principles.
Dataset and validation gaps	MUSES, Sensor Corruption Robustness Benchmarks, Open-World OOD and Anomaly Segmentation Benchmarks, FOD and Airport Apron Detection Datasets, FOD Perception Validation, and Knowledge-Base Evaluation Protocol.

Perception, SLAM, KB, and Reliability Gap Loop (2026-05-09)

Six web-search scouts compared the current repo against newer perception, SLAM, dataset, first-principles, and validation sources. Six writing agents then promoted 36 additional files with disjoint ownership.

Track	Files promoted
Perception methods	LiDAR-Camera Occupancy Fusion, Dynamic Occupancy and Freespace, Adverse-Weather Radar-LiDAR 3D Detection, RobuRCDet, SAMFusion, Spatiotemporal Memory Occupancy Flow, OVAD/OVODA Open-Vocabulary 3D Attributes, and Open-Vocabulary Panoptic Occupancy.
Perception datasets and benchmarks	STU 3D LiDAR Anomaly Segmentation, Airside FOD Synthetic Multimodal Benchmarks, RCP-Bench Cooperative Corruption Robustness, and V2X Large-Range Sequential Datasets.
SLAM methods and localization	Scan Context Family, LiDAR Bundle-Adjustment Factors, Kimera-Multi, COVINS/COVINS-G, D2SLAM, UWB / Radio Ranging SLAM, OKVIS2-X, MM-LINS, Event-Camera VIO/SLAM, Thermal-Inertial SLAM, 4D Imaging Radar RIO/SLAM, and Radar-to-LiDAR Map Localization.
First-principles KB	Continuous-Time Trajectory Splines and GP Priors, Rolling Shutter, LiDAR Deskew, and Motion Distortion, Volumetric Map Representations, Volume Rendering, Radiance Fields, and Gaussian Splatting, Detection Theory, ROC/PR, and Operating Points, and Tracking Motion Models, Lifecycle, and Metrics.
Reliability and fleet evidence	Perception/SLAM/Map Reliability Evidence Case, Perception/SLAM Statistical Validity Protocol, Uncertainty Calibration Release Gates, Perception/SLAM Corruption Fault-Injection Protocol, SLAM/Map Benchmark Protocol, and Perception/SLAM Fleet Data Contract.

P0 Backlog

Domain	Proposed file	Topic	Why it matters	Source anchors
Foundations	10-knowledge-base/geometry-3d/coordinate-frames-projections-se3.md	Coordinate frames, projections, SE(3), ENU/NED, ROS frame conventions	Every AV stack depends on correct transforms, uncertainty propagation, and sensor frame semantics.	https://www.ros.org/reps/rep-0103.html, https://autoware.one/docs/tf
Foundations	10-knowledge-base/geometry-3d/sensor-calibration-time-synchronization.md	Calibration and temporal alignment fundamentals	Multi-sensor fusion fails silently when extrinsics or timestamps drift.	https://tier4.github.io/autoware-documentation/latest/how-to-guides/integrating-autoware/creating-vehicle-and-sensor-description/calibrating-sensors/, https://pmc.ncbi.nlm.nih.gov/articles/PMC12431046/
Foundations	10-knowledge-base/state-estimation/bayesian-filtering-and-eskf.md	Bayesian filtering, ESKF, UKF, particle filters, consistency	High-rate recursive state estimation is a foundation for localization, tracking, control, and safety monitors.	https://pmc.ncbi.nlm.nih.gov/articles/PMC12526605/, https://autowarefoundation.github.io/autoware.universe_planning/pr-5583/localization/ekf_localizer/
Foundations	10-knowledge-base/controls/vehicle-dynamics-and-control.md	Kinematic/dynamic bicycle models, tire/slip, PID/LQR/MPC, actuator delay	The corpus has Frenet math but not the lower-level vehicle dynamics/control fundamentals that make plans executable.	https://saemobilus.sae.org/papers/a-survey-vehicle-dynamics-models-autonomous-driving-2024-01-2325, https://autowarefoundation.github.io/autoware_universe/main/control/autoware_smart_mpc_trajectory_follower/
Foundations	10-knowledge-base/robotics/planning-taxonomy-and-trajectory-generation.md	Route, behavior, motion, speed, and validation layers	Creates the reusable planning vocabulary for road AVs, indoor robots, yards, and airside vehicles.	https://tier4.github.io/autoware-documentation/latest/design/autoware-architecture/planning/, https://arxiv.org/abs/2402.01443
Platform	20-av-platform/power-electrical/autonomy-power-distribution.md	Power distribution, hold-up, load shedding, safe-stop energy	Sensors, compute, DBW, and safety I/O need deterministic power during faults and charger/battery transitions.	https://www.infineon.com/products/power/smart-power-switches/efuses, https://www.vicorpower.com/resource-library/articles/automotive/future-proof-advanced-evs
Platform	20-av-platform/diagnostics/functional-diagnostics-uds-doip-sovd.md	UDS, DoIP, SOVD, DTCs, remote service workflow	Fleet AVs need diagnostic sessions, fault memory, maintenance access, and traceable health states.	https://www.autosar.org/fileadmin/standards/R24-11/AP/AUTOSAR_AP_SWS_Diagnostics.pdf, https://www.iso.org/standard/87961.html
Platform	20-av-platform/ruggedization/environmental-emc-qualification.md	Environmental, EMC, IP, vibration, mechanical qualification	Indoor washdown, outdoor dust/rain, and airside EMI/de-icing/jet blast need a qualification matrix.	https://www.iso.org/standard/77579.html, https://www.iso.org/standard/77580.html, https://www.iso.org/standard/76116.html
Platform	20-av-platform/networking-connectivity/deterministic-networking-tsn.md	Extend: whole-vehicle timebase, timestamp provenance, holdover	gPTP/PTP is present, but incident reconstruction and fusion need clock-domain policy and timestamp uncertainty.	https://1.ieee802.org/tsn/802-1dg/, https://www.autosar.org/fileadmin/standards/R24-11/AP/AUTOSAR_AP_EXP_PlatformDesign.pdf
Platform	20-av-platform/sensors/close-range-proximity-safety-sensors.md	Safety laser scanners, ultrasonic/proximity, tactile bumpers, safety PLC fields	Low-speed AVs still need certified near-field protection around workers, pallets, aircraft, and docking targets.	https://www.iso.org/standard/83545.html, https://www.sick.com/us/en/sick-launches-first-ever-outdoor-safety-laser-scanner-outdoorscan3/w/press-outdoorscan3/
Autonomy	30-autonomy-stack/end-to-end-driving/evaluation-benchmarks-navsim-bench2drive.md	NAVSIM, Bench2Drive, closed-loop E2E evaluation	Open-loop imitation metrics do not reliably predict closed-loop behavior.	https://proceedings.neurips.cc/paper_files/paper/2024/hash/32768f7faf1995026ef9821c696f3404-Abstract-Datasets_and_Benchmarks_Track.html, https://arxiv.org/abs/2406.03877
Autonomy	30-autonomy-stack/planning/airside-closed-loop-planning-benchmark.md	Airside closed-loop planning benchmark and metrics	Airside planning needs scenario-level progress, comfort, rule, and safety metrics, not only model descriptions.	https://arxiv.org/abs/2406.15349, https://arxiv.org/abs/2406.03877
Autonomy	30-autonomy-stack/planning/trajectory-tracking-control.md	Nominal trajectory tracking and vehicle dynamics control	The planner/controller boundary is where delay, saturation, slip, actuator faults, and comfort show up.	https://autowarefoundation.github.io/autoware_universe/pr-10047/control/autoware_trajectory_follower_node/, https://arxiv.org/abs/2503.10559
Autonomy	30-autonomy-stack/planning/behavior-planning-maneuver-arbitration.md	Tactical behavior planning and maneuver arbitration	Generic AVs need a layer that turns goals, ODD state, rules, V2X, and fallback policy into maneuvers.	https://arxiv.org/abs/2406.01587, https://link.springer.com/article/10.1007/s44267-025-00095-w
Autonomy	30-autonomy-stack/multi-agent-v2x/v2x-cooperative-planning.md	End-to-end V2X cooperative planning	V2X should change prediction, behavior, and trajectory decisions, not only perception and protocol messages.	https://arxiv.org/abs/2405.03971, https://arxiv.org/abs/2408.09251
Autonomy	30-autonomy-stack/vla-vlm/vlm-vla-reliability-benchmarks.md	Driving VLM/VLA reliability, hallucination, and robustness benchmarks	Language reasoning is useful only if prompt failures, sensor corruption, and wrong answers are measured.	https://arxiv.org/abs/2501.04003, https://openaccess.thecvf.com/content/WACV2025/html/Chen_Automated_Evaluation_of_Large_Vision-Language_Models_on_Self-Driving_Corner_Cases_WACV_2025_paper.html
Autonomy	30-autonomy-stack/end-to-end-driving/airside-autonomy-benchmark-spec.md	Airside autonomy benchmark and dataset specification	Road and indoor benchmarks do not cover stands, ramps, aircraft, GSE, FOD, marshalling, and airport rules.	https://huggingface.co/datasets/nvidia/PhysicalAI-Autonomous-Vehicles, https://arxiv.org/abs/2406.03877
Autonomy	30-autonomy-stack/end-to-end-driving/cooperative-v2x-e2e-driving.md	Cooperative V2X and infrastructure-augmented autonomy	Infrastructure sensors and shared context matter for occlusions, indoor/outdoor campuses, and airside fleets.	https://arxiv.org/abs/2408.09251, https://mobility-lab.seas.ucla.edu/v2x-real/
Autonomy	30-autonomy-stack/world-models/radar-native-world-models.md	4D radar-native world models and radar simulation	Radar is the weather/lighting fallback modality; world models should not be only camera/LiDAR-native.	https://arxiv.org/abs/2411.10962, https://arxiv.org/abs/2504.00859
Runtime/cloud	50-cloud-fleet/operations/fleet-sre-incident-response.md	Fleet SRE, incident command, runbooks, postmortems	Fleet safety depends on operational ownership, severity taxonomy, fleet-stop policy, and post-incident learning.	https://opentelemetry.io/docs/what-is-opentelemetry/, https://foxglove.dev/blog/observability-for-robotics-systems, https://waymo.com/blog/2025/06/safe-to-deploy
Runtime/cloud	50-cloud-fleet/map-operations/hd-map-lifecycle-operations.md	Map lifecycle operations: survey, diff, validate, deploy, rollback	Maps are safety-critical runtime artifacts across warehouses, yards, roads, and airports.	https://arxiv.org/abs/2406.01961, https://www.here.com/products/automotive/hd-live-map
Runtime/cloud	50-cloud-fleet/data-governance/fleet-data-privacy-governance.md	Data governance, privacy, retention, access control	AV logs capture people, facilities, routes, operators, and sensitive operational data.	https://www.ftc.gov/policy/advocacy-research/tech-at-ftc/2024/05/cars-consumer-data-unlawful-collection-use, https://docs.aws.amazon.com/iot-fleetwise/latest/developerguide/what-is-iotfleetwise.html
Runtime/cloud	40-runtime-systems/software-operations/on-vehicle-supply-chain-runtime-security.md	Signed artifacts, SBOM, secure boot, CVE triage, secrets, certs	Runtime security is scattered across OTA, ROS, and ML deployment docs; it needs an operations file.	https://csrc.nist.gov/pubs/sp/800/218/final, https://www.nhtsa.gov/research/vehicle-cybersecurity, https://uptane.org
Runtime/cloud	50-cloud-fleet/ota/software-update-management-system-ops.md	SUMS governance for code, models, maps, config, calibration	OTA mechanics exist, but release approval, rollback drills, cohorts, and readiness gates need a separate playbook.	https://www.vehicle-certification-agency.gov.uk/connected-and-automated-vehicles/cyber-security-and-software-updating/, https://waymo.com/blog/2025/06/safe-to-deploy
Safety	60-safety-validation/safety-case/incident-reporting-post-market-monitoring.md	Incident reporting, near-miss, forensics, post-market monitoring	Transparency and post-deployment monitoring are central to regulator trust and safety-case maintenance.	https://www.nhtsa.gov/laws-regulations/standing-general-order-crash-reporting, https://www.faa.gov/airports/airport_safety/certalerts/part_139_certalert_24_02, https://www.easa.europa.eu/en/node/138789
Safety	60-safety-validation/safety-case/safety-case-evidence-traceability.md	Living safety-case evidence and traceability architecture	Claims, assumptions, evidence IDs, logs, change impact, and review workflows need one artifact model.	https://www.shopulstandards.com/ProductDetail.aspx?productid=UL4600, https://arxiv.org/abs/2404.05444, https://www.york.ac.uk/assuring-autonomy/guidance/amlas/
Safety	60-safety-validation/standards-certification/eu-ai-act-machinery-compliance-dossier.md	EU AI Act, Machinery Regulation, CRA, aviation cyber dossier	Date-sensitive compliance is scattered; the high-risk AI timing changed in May 2026 negotiations.	https://eur-lex.europa.eu/eli/reg/2024/1689/oj/eng, https://digital-strategy.ec.europa.eu/en/news/eu-agrees-simplify-ai-rules-boost-innovation-and-ban-nudification-apps-protect-citizens, https://digital-strategy.ec.europa.eu/en/policies/cra-summary
Operations	70-operations-domains/warehouse/operations/amr-autonomous-forklift-operations.md	Indoor warehouse AMR and autonomous forklift operations	Adds GNSS-denied WMS-integrated indoor autonomy, dock staging, charging, and robot safety.	https://www.businesswire.com/news/home/20240305945069/en/Walmart-and-Fox-Robotics-Expand-Partnership-for-Autonomous-Forklifts, https://group.dhl.com/en/media-relations/press-releases/2024/dhl-supply-chain-passes-unprecedented-500-million-picks-milestone-using-locus-robotics-autonomous-mobile-robots.html
Operations	70-operations-domains/logistics-yards/operations/autonomous-yard-truck-operations.md	Autonomous yard truck and outdoor industrial yard operations	Adds YMS/TMS/WMS integration, trailer spotting, RTK/private wireless, and mixed manual/autonomous yard traffic.	https://venturebeat.com/business/isee-commercially-deploys-worlds-first-fully-autonomous-truck-yard/, https://www.outrider.ai/resources/design-checklist-distribution-yards
Operations	70-operations-domains/ports/operations/autonomous-terminal-tractor-port-operations.md	Autonomous port terminal tractor and container yard operations	Adds TOS integration, quay-yard-stack routing, vessel schedules, terminal safety, and workforce constraints.	https://hhla.de/en/media/news/detail-view/automated-and-sustainable-logistics-hhla-and-fernride-launch-pilot-project-in-estonia, https://www.kalmarglobal.com/news--insights/articles/2025/20250219_kalmar_unveils_kalmar_one_automation_system/
Operations	70-operations-domains/mining/operations/autonomous-haulage-operations.md	Autonomous mining and quarry haulage operations	Mining is one of the most mature AV domains and teaches fleet dispatch, private-road ODDs, haul-road design, and exclusion zones.	https://www.komatsu.com/en/newsroom/2024/komatsu-autonomous-haulage-system-achieves-7-billion-tonnes-of-material-moved/, https://www.cat.com/en_US/by-industry/mining/autonomous-solutions.html
Operations	70-operations-domains/agriculture/operations/autonomous-tractor-field-operations.md	Autonomous tractor and field operations	Adds seasonal ODDs, field boundaries, implement safety, remote supervision, crop-row maps, and low-connectivity workflows.	https://www.deere.com/en/news/all-news/john-deere-reveals-autonomous-machines-at-ces-2025/, https://www.iso.org/standard/73915.html
Operations	70-operations-domains/construction/operations/autonomous-earthmoving-site-operations.md	Autonomous construction and earthmoving site operations	Adds temporary layouts, changing work zones, machine control, teleoperation fallback, and site-production KPIs.	https://www.cat.com/en_US/news/machine-press-releases/caterpillar-demonstrates-first-battery-electric-autonomous-haul-truck.html, https://global.kawasaki.com/en/corp/newsroom/news/detail/?f=20240930_3166
Operations	70-operations-domains/road-av/operations/robotaxi-service-operations.md	Robotaxi service operations	Company profiles exist, but the operations layer needs depots, rider support, remote assistance, launch gates, and incident response.	https://waymo.com/blog/, https://www.nhtsa.gov/laws-regulations/av-step
Operations	70-operations-domains/road-av/operations/autonomous-trucking-lane-operations.md	Autonomous trucking lane operations	Adds hub-to-hub lane design, terminal ops, inspections, enforcement, remote support, and launch governance.	https://blog.aurora.tech/aurora-driver/aurora-launches-commercial-driverless-trucking-service, https://www.gov.uk/government/news/self-driving-vehicles-set-to-be-on-roads-by-2026-as-automated-vehicles-act-becomes-law
Operations	70-operations-domains/delivery-robots/operations/sidewalk-delivery-robot-operations.md	Sidewalk delivery robot operations	Adds sidewalk/curb ODDs, pedestrian interaction, municipal permitting, accessibility constraints, and store handoff.	https://www.serverobotics.com/news/serve-robotics-announces-expansion-of-delivery-operations-to-miami-metro-area, https://www.starship.xyz/press/

P1 Backlog

Domain	Proposed file or update	Topic
Foundations	Promoted: 10-knowledge-base/state-estimation/data-association-and-gating.md, 10-knowledge-base/state-estimation/probabilistic-multi-object-association.md	Kalman/Hungarian/JPDA/MHT, track lifecycle, data association fundamentals.
Foundations	Promoted: 10-knowledge-base/mapping/occupancy-bayes-evidential-dynamic-grids.md	Log-odds occupancy, inverse sensor models, inflation, dynamic occupancy, costmap semantics.
Foundations	10-knowledge-base/systems-engineering/robotics-middleware-real-time.md	ROS 2/DDS QoS, executors, deadlines, jitter, lifecycle, deterministic messaging.
Foundations	10-knowledge-base/systems-engineering/odd-scenario-based-assurance.md	ODD, OpenSCENARIO, ISO 34502, ISO/PAS 8800, safety-case fundamentals.
Platform	20-av-platform/networking-connectivity/zonal-ee-harness-connectors.md	Zonal E/E, automotive Ethernet PHYs, sensor SerDes, harnessing, serviceability.
Platform	20-av-platform/sensors/visible-camera-hardware.md	Camera hardware, optics, HDR/LFM, ISP, triggers, synchronization.
Platform	20-av-platform/sensors/gnss-ins-imu-odometry-hardware.md	PNT resilience, GNSS/INS/IMU, wheel odometry, spoofing/denial.
Platform	20-av-platform/sensors/calibration-bay-fixtures.md	Calibration bay targets, fixtures, surveyed references, fleet workflow.
Platform	20-av-platform/thermal/vehicle-thermal-management.md	Vehicle-level thermal budget across compute, sensors, enclosures, battery, heaters.
Platform	20-av-platform/drive-by-wire/can-bus-dbw.md	Extend or split actuator redundancy, E-stop, STO, brake/steer safety I/O.
Autonomy	30-autonomy-stack/planning/motion-prediction.md	Extend with calibrated prediction uncertainty for planner margins and fallback policy.
Autonomy	30-autonomy-stack/planning/world-model-simulation-planning.md	World-model rollouts for planning, rare-event generation, and scenario synthesis.
Autonomy	30-autonomy-stack/multi-agent-v2x/cooperative-perception-benchmarks.md	Latency, bandwidth, pose-error, packet-loss, and mAP metrics for cooperative perception.
Autonomy	30-autonomy-stack/planning/planner-preference-optimization.md	Human feedback, comfort, assertiveness, yielding, and procedural preference optimization.
Autonomy	30-autonomy-stack/multi-agent-v2x/v2x-protocols-airside.md	Extend with NR-V2X Release 18/19 conformance and QoS profile.
Autonomy	30-autonomy-stack/end-to-end-driving/data-engine-long-tail-curation.md	Long-tail mining, VLM-assisted curation, auto-labeling, and training-set repair.
Autonomy	30-autonomy-stack/world-models/planning-oriented-world-models-rft.md	Planning-optimized latent world models, RL, and reinforcement fine-tuning.
Autonomy	30-autonomy-stack/vla-vlm/action-heads-control-interfaces.md	VLA action heads, trajectory tokens, diffusion policies, and safe planner/controller handoff.
Autonomy	30-autonomy-stack/simulation/dynamic-agent-behavior-models-airside.md	Reactive aircraft, GSE, personnel, vehicle, and mixed-traffic behavior models.
Autonomy	30-autonomy-stack/simulation/closed-loop-safety-benchmarks.md	NeuroNCAP-style closed-loop safety benchmark patterns.
Runtime/cloud	50-cloud-fleet/fleet-management/fleet-operations-center-playbooks.md	Fleet ops center authority model, shift handover, emergency stop, site coordination.
Runtime/cloud	50-cloud-fleet/mlops/model-lifecycle-governance.md	Model cards, approval gates, eval datasets, canary criteria, rollback triggers, audit history.
Runtime/cloud	50-cloud-fleet/data-platform/data-catalog-lineage-quality-ops.md	Data catalog, lineage, schemas, quality gates, deletion propagation, replay reproducibility.
Runtime/cloud	40-runtime-systems/software-operations/edge-runtime-supervision-config-management.md	Watchdogs, config schemas, feature flags, offline-first operation, local fallback.
Safety	60-safety-validation/verification-validation/iso-3450x-airside-scenario-evidence.md	ISO 34501-34505 scenario evidence mapping, including ISO 34504/34505.
Safety	60-safety-validation/safety-case/airside-av-hara-stpa-sotif-analysis.md	Item definition, HARA, STPA, FMEA, SOTIF triggering conditions, safety goals.
Safety	60-safety-validation/standards-certification/safety-functions-pld-sil-validation.md	Per-function PLd/SIL evidence for braking, E-stop, personnel detection, geofence.
Safety	60-safety-validation/standards-certification/ml-assurance-data-governance.md	ML assurance lifecycle, data requirements, model change safety case, ISO/IEC 42001 and TR 5469 alignment.
Operations/industry	80-industry-intel/regulations/cross-domain-av-regulatory-map.md	Cross-domain standards map across industrial mobile robots, road AVs, mining, agriculture, airside, and delivery robots.
Operations/industry	80-industry-intel/market-competitive/cross-domain-autonomy-competitive-landscape.md	Competitive landscape by domain maturity, deployments, vendors, and business models.
Operations/industry	80-industry-intel/companies/<company>/tech-stack.md	First-wave company profiles: Fox Robotics, Locus, Outrider, ISEE, Kalmar, Komatsu, Caterpillar, John Deere, Serve Robotics, Starship.
Operations/industry	70-operations-domains/cross-domain/mapping-operations/indoor-outdoor-map-ops-playbook.md	Operational map lifecycle: site survey, map ownership, route approvals, geofence releases, WMS/YMS/TOS/AODB integration.

P2 Backlog And Extension Queue

Domain	Proposed file or update	Topic
Foundations	10-knowledge-base/machine-learning/av-data-evaluation-fundamentals.md	Dataset splits, scenario coverage, benchmark interpretation, open-loop versus closed-loop metrics.
Platform	20-av-platform/networking-connectivity/vehicle-middleware-dds-someip-zenoh.md	DDS, SOME/IP, zero-copy IPC, Zenoh, service discovery, bridge policy.
Platform	20-av-platform/sensors/automated-sensor-cleaning.md	Extend with cleaning verification, fluid logistics, freeze/de-ice, pump/nozzle telemetry.
Platform	20-av-platform/compute/safety-certified-runtime-compute.md	Safety-certified runtime compute and mixed-criticality partitioning.
Autonomy	30-autonomy-stack/localization-mapping/overview/infrastructure-aided-localization.md	UWB, fiducials, RFID, infrastructure-aided localization for terminals, hangars, docks, and repetitive structures.
Autonomy	30-autonomy-stack/end-to-end-driving/learned-autonomy-safety-assurance.md	Evidence arguments for world models, VLA, and E2E driving.
Autonomy	30-autonomy-stack/planning/map-free-online-map-planning.md	Planning when HD maps are stale, unavailable, or wrong.
Autonomy	30-autonomy-stack/planning/safety-critical-planning-cbf.md	Extend with reachability and runtime assurance beyond CBF.
Autonomy	30-autonomy-stack/planning/reactive-sim-agents-planner-validation.md	Reactive simulation agents for planner validation.
Runtime/cloud	40-runtime-systems/software-operations/sensor-calibration-fleet-ops.md	Calibration artifact versioning, drift remediation, maintenance gates.
Runtime/cloud	50-cloud-fleet/cloud-operations/finops-capacity-planning.md	Fleet data/cloud FinOps and capacity planning.
Runtime/cloud	50-cloud-fleet/fleet-management/fleet-interoperability-standards.md	VDA 5050, Open-RMF, MassRobotics interoperability and adapter policy.
Safety	60-safety-validation/cybersecurity/cybersecurity-airside-av.md	Extend with CSMS/SUMS evidence matrix, SBOM ownership, vulnerability reporting, red-team cadence, SOC exercises.
Safety	60-safety-validation/runtime-assurance/runtime-verification-monitoring.md	Extend with monitor qualification, threshold calibration, WCET proof, false-positive/false-negative acceptance, monitor failure handling.
Safety	60-safety-validation/standards-certification/airside-agvs-regulatory-approval-playbook.md	FAA, EASA/ICAO, CAAS/TR68 approval path for airside AGVS deployments.
Operations/industry	Airside refresh set	Update airside industry, regulatory trajectory, reference airside AV stack production deployment, and TractEasy production deployment for 2024-2026 changes.
Operations/industry	70-operations-domains/deployment-playbooks/generic-site-onboarding-checklist.md	Generic AV site onboarding checklist with domain overlays.
Operations/industry	80-industry-intel/deployments/2024-2026-autonomy-deployment-index.md	Neutral deployment ledger by domain, site, vehicle type, autonomy level, safety operator status, regulatory basis, and source date.

Execution Order

1. Treat P0 as completed at the first-file level, then revisit individual files only for deeper expansion or source refreshes.
2. Promote P1 files where they unlock multiple downstream docs, especially model lifecycle governance, ISO 3450x evidence, VLA reliability, and cross-domain regulatory/competitive maps.
3. Keep P2 as extension work unless a new deployment or repo goal makes a topic urgent.
4. When a P1/P2 gap is completed, move it from this backlog into the relevant domain overview or audit and update README.md, INDEX.md, and METHODOLOGY.md counts.