DAA System — Sizing Estimate


Method	Architecture-driven PERT (v3), revised with codebase analysis
Last Updated	2026-03-22
Source PRD	DAA PRD
Stories	122
Expected Duration	~420 days (down from 587 greenfield)
9-Person Team	~14-17 weeks (moderate)
Confidence	High (Phases 1-4) / Medium (Phases 5-6)

Executive Summary

The DAA system is sized at 122 stories across 6 phases, with a revised PERT expected duration of ~420 days of active work time (~84 weeks). This is down from a greenfield estimate of 587 days, reflecting significant existing code and services that can be leveraged.

Phase 1 is the largest beneficiary of this revision – reduced from 244.7 days to ~93 days (-62%) because most of the detect-and-notify capability already exists in Clojure-based Themis services and the Go-based flight-traffic-exchanger. Phase 1 is primarily an AI-assisted port and integration effort, not greenfield development.

These estimates are still conservative. Further reductions are likely as AI-assisted code translation matures and as teams find additional reuse opportunities during implementation. The per-story PERT values for Phases 2-6 have not yet been revised downward for the existing collision-service scaffolding (C2 commands, telemetry subscriptions, mission integration).

Existing Code Leverage (Phase 1)

A codebase exploration revealed that the core Phase 1 functionality already exists across multiple services:

Phase 1 Feature	Existing Service	What Exists	What’s Needed
OCDV/OCAV/WC/NMAC algorithms	`duck/daac.cljc` (utm-clojure-kit)	~300 lines of pure functions, all volume checks	AI-port to Go, update constants to PRD values
ADS-B ingestion (CAT021 ASTERIX)	`utm-themis-daa` (Clojure, UDP)	Full parser, 1 Hz ingestion, geo-filter	Already consumed by flight-traffic-exchanger
Plot dedup + staleness	`utm-themis-daa` + `flight-traffic-exchanger`	Dedup by aircraft ID, 5s staleness	Already in Go service
Geo filtering (site proximity)	`flight-traffic-exchanger`	Provider abstraction, site-based filtering	Config update to 16,093m
Intrusion state classification	`duck/daac.cljc`	All Clear through NMAC states	Port to Go, add Clear to Proceed (new)
Real-time data streaming	`flight-traffic-exchanger`	gRPC-Web streaming, observation cache	Add intrusion alert events to stream
Compliance reporting	`utm-themis-api`	Report endpoints, intrusion summaries	Port to Go or keep as-is initially
Air traffic viz + color coding	Apollo `TrafficObservationsLayer`	Basic aircraft display on map	Extend with DAA status colors
OCDV/OCAV circle rendering	`utm-themis-spaces` (ClojureScript)	Working prototype with interactive demo	Port to React (adapt OperationalAreaLayer)
Alert infrastructure	Apollo notification system	TTS, toast notifications, severity levels	Add DAA-specific alert types + audio tones
C2 command scaffolding	`uncrew-collision-flight-controller`	Hold, RTL, GoToCLZ, mission replan	Extend for Crouch command (Phase 2)

AI-Assisted Translation Strategy

The Clojure DAA algorithm code (duck/daac.cljc) is ideal for AI translation:

Pure functions with no side effects
Well-defined inputs (lat, lon, altitude, airspeed) and outputs (intrusion state)
Comprehensive test vectors available in utm-clojure-kit test suite
Target: port to Go with property-based tests validating output equivalence

Delivery Timeline (9-Person Team)

Assumptions:

9 engineers available full-time
Completion rate: 0.77 (from Jira historical data – accounts for meetings, reviews, overhead)
Effective team capacity: 9 x 5 x 0.77 = 34.7 productive days/week
Phase dependencies constrain parallelism (not all phases can run concurrently)

Per-Phase Delivery Schedule

Phase	Work (days)	Greenfield Est	Savings	Team Size	Calendar Duration
1 – Detect & Notify	93	244.7	-62%	9	2.7 weeks
2 – Detect & Automated Avoid	90	105.3	-15%	7	3.3 weeks
3 – Onboard ADS-B Sensor Integration	36.1	36.1	–	2	4.7 weeks
4 – Onboard + Cloud DAA Orchestration	82.3	82.3	–	6	3.6 weeks
5 – Terrain & Traffic-Aware Crouch	52.4	52.4	–	3	4.5 weeks
6 – UAS Performance Calibration	66.2	66.2	–	9	1.9 weeks
Total	~420	587.0	-28%

Delivery Scenarios

    gantt
    title DAA Delivery Timeline — Revised (9-person team)
    dateFormat YYYY-MM-DD
    axisFormat %b %d

    section Moderate
    Phase 1 Detect & Notify           :p1, 2026-04-01, 21d
    Phase 2 Automated Avoid            :p2, after p1, 24d
    Phase 3 Onboard ADS-B              :p3, after p1, 33d
    Phase 4 Dual-Authority DAA         :p4, after p2, 25d
    Phase 5 Terrain Crouch             :p5, after p2, 32d
    Phase 6 UAS Calibration            :p6, after p4, 14d

Scenario A – Aggressive (max parallelism, no buffer):

Window	Activity	Duration
Wk 1-3	Phase 1 (full team, AI-assisted port)	3 wk
Wk 4-7	Phase 2 (7 ppl) + Phase 3 (2 ppl)	4 wk
Wk 8-12	Phase 4 (6 ppl) + Phase 5 (3 ppl)	4 wk
Wk 13-14	Phase 6 (full team)	2 wk
	Total	~14 weeks / 3.5 months

Scenario B – Moderate (recommended, ~15% buffer):

Window	Activity	Duration
Wk 1-4	Phase 1	4 wk
Wk 5-9	Phase 2 + Phase 3	5 wk
Wk 10-14	Phase 4 + Phase 5	5 wk
Wk 15-17	Phase 6	3 wk
	Total	~17 weeks / 4 months

Scenario C – Conservative (pessimistic + 25% schedule risk):


Total	~22 weeks / 5.5 months

Core DAA Capability (Phases 1-4 only)

If Phases 5 and 6 are deferred, the core DAA system (detect, avoid, onboard detection, dual-authority) is:

	Aggressive	Moderate	Conservative
Work days	301.4	301.4	301.4
Calendar	~11 weeks	~14 weeks	~17 weeks
Months	~2.5	~3.5	~4

Domain Staffing Guidance

Domain	Share of Work	Suggested Allocation (of 9)	Notes
backend-go	58%	5-6	Heaviest domain; AI-assisted Clojure port in Phase 1
firmware-cpp	21%	1-2	Peaks in Phases 3-4 and 6
frontend-react	17%	2	Steady across phases; leverage existing layers
data-pipeline	3%	0-1	Phase 5 only; can be backend-go engineers
infrastructure	1%	0	One-time HA setup; can be backend or DevOps

Further Optimization Opportunities

These estimates are conservative. Areas where further reduction is likely:

Keep Clojure reporting initially – Phase 1 reporting stories (~10d) could be deferred by keeping utm-themis-api report endpoints running alongside the new Go services during the transition.
Shared test vectors – The Clojure test suite can be exported as JSON fixtures for Go tests, eliminating redundant test case authoring.
AI-assisted frontend port – The ClojureScript visualization code in utm-themis-spaces (traffic layers, OCDV/OCAV circles) can be AI-translated to React/TypeScript with the existing TrafficObservationsLayer as the target scaffold.
Phase 2 reduction – The collision service already has telemetry subscriptions, C2 command integration (Hold, RTL, GoToCLZ), and mission state tracking. Adding Crouch as a new command type is incremental.
Proto reuse – flight-traffic-exchanger already defines observation protos. Extending them for intrusion state is incremental, not greenfield.

Per-Phase Breakdown

Phase	Stories	Expected (days)	Range (95% CI)	Weeks	Confidence
1 – Detect & Notify	59	244.7	226.4 – 263.0	48.9	High
2 – Detect & Automated Avoid	20	105.3	92.1 – 118.5	21.1	High
3 – Onboard ADS-B Sensor Integration	7	36.1	27.2 – 45.0	7.2	High
4 – Onboard + Cloud DAA Orchestration	15	82.3	69.7 – 94.9	16.5	High
5 – Terrain & Traffic-Aware Crouch Targeting	10	52.4	42.9 – 61.9	10.5	Medium
6 – UAS Performance Calibration & Latency Tuning	11	66.2	54.4 – 78.0	13.2	Medium
Total	122	587.0	555.8 – 618.2	117.4	High

Per-Phase Effort Distribution

    %%{init: {'theme': 'neutral'}}%%
graph LR
    subgraph "Phase Effort (expected days)"
        P1["Phase 1: Detect & Notify<br/>244.7d | 59 stories"]
        P2["Phase 2: Detect & Automated Avoid<br/>105.3d | 20 stories"]
        P3["Phase 3: Onboard ADS-B Sensor<br/>36.1d | 7 stories"]
        P4["Phase 4: Onboard + Cloud Orchestration<br/>82.3d | 15 stories"]
        P5["Phase 5: Terrain & Traffic Crouch<br/>52.4d | 10 stories"]
        P6["Phase 6: UAS Performance Calibration<br/>66.2d | 11 stories"]
    end
    P1 --> P2 --> P3 --> P4 --> P5 --> P6

    %%{init: {'theme': 'neutral'}}%%
xychart-beta
    title "Expected Days per Phase"
    x-axis ["P1: Detect", "P2: Avoid", "P3: ADS-B", "P4: Orchestration", "P5: Terrain", "P6: Calibration"]
    y-axis "Days" 0 --> 260
    bar [244.7, 105.3, 36.1, 82.3, 52.4, 66.2]

Per-Domain Breakdown

Domain	Stories	Expected (days)	Share
backend-go	80	370.5	63.1%
firmware-cpp	14	105.8	18.0%
frontend-react	25	89.2	15.2%
data-pipeline	2	16.2	2.8%
infrastructure	1	5.3	0.9%
Total	122	587.0	100%

Phase Details

Phase 1: Detect & Notify (59 stories, 244.7 days)

Core cloud-side detection engine, GCS visualization, alerting, and compliance logging.

#	Story	Domain	Repo	Opt	Likely	Pess	Expected
1	Ingest ADS-B plots at 1-2 Hz from traffic source	backend-go	collision-flight-controller	2	5	10	5.33
2	Ingest drone telemetry at 1-2 Hz per active flight	backend-go	collision-flight-controller	2	5	10	5.33
3	Plot dedup – maintain latest position per aircraft ID	backend-go	collision-flight-controller	0.5	2	5	2.25
4	Plot staleness removal – 5s timeout	backend-go	collision-flight-controller	0.5	2	5	2.25
5	Aircraft type identification from ADS-B emitter category	backend-go	collision-flight-controller	0.5	2	5	2.25
7	Dynamic site management – hot-reload	backend-go	collision-flight-controller	2	5	10	5.33
8	Multi-region support – 1000+ sites	backend-go	collision-flight-controller	0.5	2	5	2.25
9	Site proximity filtering – 16,093m radius	backend-go	collision-flight-controller	0.5	2	5	2.25
10	OCDV detection – cylinder geometry	backend-go	collision-flight-controller	5	10	20	10.83
11	OCAV detection – dynamic radius	backend-go	collision-flight-controller	5	10	20	10.83
12	OCAV radius upper bound capping	backend-go	collision-flight-controller	2	5	10	5.33
13	Time-to-avoid computation – 8-component sum	backend-go	collision-flight-controller	2	5	10	5.33
14	DAA geometric constants – configurable	backend-go	collision-flight-controller	0.5	2	5	2.25
15	DAA timing parameters baseline	backend-go	collision-flight-controller	0.5	2	5	2.25
16	DAA timeline selection per surveillance type + UA model	backend-go	collision-flight-controller	0.5	2	5	2.25
17	Unknown altitude handling – assume co-altitude	backend-go	collision-flight-controller	0.5	2	5	2.25
18	Unknown airspeed handling – assume max 77.17 m/s	backend-go	collision-flight-controller	0.5	2	5	2.25
19	6-state intrusion classification	backend-go	collision-flight-controller	2	5	10	5.33
20	Intrusion state transition event emission	backend-go	collision-flight-controller	2	5	10	5.33
21	Multi-drone evaluation – 50 active, 100 capacity	backend-go	collision-flight-controller	2	5	10	5.33
22	Ended flight cleanup – 60 min timeout	backend-go	collision-flight-controller	0.5	2	5	2.25
23	Calculation frequency – per-measurement / 1 Hz	backend-go	collision-flight-controller	2	5	10	5.33
24	Warning alert data constraint – current data only	backend-go	collision-flight-controller	0.5	2	5	2.25
25	Clear to Proceed logic – 4-measurement separation	backend-go	collision-flight-controller	2	5	10	5.33
26	Alert latency <=2s from ADS-B receipt	backend-go	collision-flight-controller	2	5	10	5.33
27	1000 plot updates/sec throughput	backend-go	collision-flight-controller	2	5	10	5.33
28	Cloud processing latency <500ms P99	backend-go	collision-flight-controller	0.5	2	5	2.25
29	Stream connection recovery + operator indication	backend-go	collision-flight-controller	2	5	10	5.33
30	DAA health gate for BVLOS mission approval	backend-go	collision-flight-controller	2	5	10	5.33
31	DAA service unavailability alert + crouch after 10s	backend-go	collision-flight-controller	2	5	10	5.33
32	DAA service recovery – notify, resume, confirm	backend-go	collision-flight-controller	2	5	10	5.33
33	BVLOS scope enforcement	backend-go	collision-flight-controller	0.5	2	5	2.25
34	99.9% uptime SLA – arch/infra	backend-go	collision-flight-controller	0.5	2	5	2.25
35	Intrusion logging – per-second, full fields	backend-go	collision-flight-controller	5	10	20	10.83
36	Differentiated data retention – 30d / permanent	backend-go	collision-flight-controller	2	5	10	5.33
37	Compliance report generation	backend-go	collision-flight-controller	2	5	10	5.33
38	Report export (CSV/XLSX/PDF)	backend-go	collision-flight-controller	0.5	2	5	2.25
39	Report access control – supervisor/admin only	backend-go	collision-flight-controller	0.5	2	5	2.25
40	Aircraft icons – intrusion status color-coding	frontend-react	apollo-frontend	0.5	2	5	2.25
41	Aircraft detail panel – DAA fields	frontend-react	apollo-frontend	0.5	2	5	2.25
42	Stale data indicators	frontend-react	apollo-frontend	0.5	2	5	2.25
43	Mission console traffic scope – OCDV radius	frontend-react	apollo-frontend	0.5	2	5	2.25
44	OCAV circle rendering – dynamic, red	frontend-react	apollo-frontend	2	5	10	5.33
45	OCDV circle rendering – yellow, toggle	frontend-react	apollo-frontend	2	5	10	5.33
46	Alert circle position tracking + lifecycle	frontend-react	apollo-frontend	0.5	2	5	2.25
47	Audio alert system – 4 levels, queuing	frontend-react	apollo-frontend	2	5	10	5.33
48	Audio audibility check pre-takeoff	frontend-react	apollo-frontend	2	5	10	5.33
49	Volume control	frontend-react	apollo-frontend	0.5	2	5	2.25
50	Suppress audio on page load	frontend-react	apollo-frontend	0.5	2	5	2.25
51	Active alert list – bearing, velocity, range	frontend-react	apollo-frontend	2	5	10	5.33
52	Alert severity sorting + CPA secondary sort	frontend-react	apollo-frontend	2	5	10	5.33
53	Alert-to-map navigation	frontend-react	apollo-frontend	0.5	2	5	2.25
54	Alert count badge with color distinction	frontend-react	apollo-frontend	0.5	2	5	2.25
55	Alert auto-resolution on OCDV exit	frontend-react	apollo-frontend	0.5	2	5	2.25
56	Intruder selection highlights alert in list	frontend-react	apollo-frontend	0.5	2	5	2.25
57	End-to-end alert latency <=3s	frontend-react	apollo-frontend	2	5	10	5.33
58	15 fps with 100 aircraft rendered	frontend-react	apollo-frontend	2	5	10	5.33
59	Define DAA proto messages	backend-go	api-private	2	5	10	5.33
60	HA deployment for 99.9% SLA	infrastructure	infrastructure	2	5	10	5.33

Phase 2: Detect & Automated Avoid (20 stories, 105.3 days)

Automated crouch maneuver execution, avoidance state machine, and degraded-mode handling.

#	Story	Domain	Repo	Opt	Likely	Pess	Expected
6	Telemetry staleness escalation during active avoidance	backend-go	collision-flight-controller	0.5	2	5	2.25
61	Notify-then-act path – transmit, wait, crouch	backend-go	collision-flight-controller	5	10	20	10.83
62	Act-and-notify path – immediate crouch	backend-go	collision-flight-controller	5	10	20	10.83
64	Loss of Well Clear response – immediate crouch	backend-go	collision-flight-controller	2	5	10	5.33
65	NMAC response – continue existing descent	backend-go	collision-flight-controller	0.5	2	5	2.25
66	Notification threshold – configurable 15s baseline	backend-go	collision-flight-controller	0.5	2	5	2.25
67	Crouch initiation latency <=500ms	backend-go	collision-flight-controller	0.5	2	5	2.25
68	Crouch confirmation to backend within 1s	backend-go	collision-flight-controller	2	5	10	5.33
69	Non-confirmation escalation after 3s	backend-go	collision-flight-controller	2	5	10	5.33
70	Maximum crouch duration – hold at min altitude	backend-go	collision-flight-controller	0.5	2	5	2.25
71	Crouch maneuver definition (pause + descend + hold)	backend-go	collision-flight-controller	0.5	2	5	2.25
72	Clear to Resume – all clear + operator confirm	backend-go	collision-flight-controller	2	5	10	5.33
73	Multiple intrusion handling – continue, log separately	backend-go	collision-flight-controller	2	5	10	5.33
74	Avoidance during lost-link RTH	backend-go	collision-flight-controller	2	5	10	5.33
75	Cloud traffic provider failure – crouch all BVLOS	backend-go	collision-flight-controller	5	10	20	10.83
76	C2 loss with VOs – continue detection, alert VOs	backend-go	collision-flight-controller	2	5	10	5.33
77	Prioritized alerting – actively controlled > supervised	backend-go	collision-flight-controller	2	5	10	5.33
79	Integrate DAA crouch trigger into safety_crouch_fsm	firmware-cpp	mavlink-shim	3	7.5	15	8.00
80	Buffer encounter logs during C2 loss	firmware-cpp	mavlink-shim	0.8	3	7.5	3.38
81	Clear-to-Resume UI with operator confirmation gate	frontend-react	apollo-frontend	2	5	10	5.33

Phase 3: Onboard ADS-B Sensor Integration (7 stories, 36.1 days)

Airborne ADS-B receiver relay, validation, buffering, and cloud-side source fusion. This phase introduces the dual-source (ground + airborne) traffic picture.

#	Story	Domain	Repo	Opt	Likely	Pess	Expected
82	Onboard ADS-B receiver relay to cloud	firmware-cpp	mavlink-shim	7.5	15	30	16.25
92	ADS-B receiver parameter configuration from cloud	firmware-cpp	mavlink-shim	1	3	7	3.33
93	Onboard ADS-B validation (lat/lon/speed/alt/ICAO)	firmware-cpp	mavlink-shim	1	3	7	3.33
94	ADS-B relay buffering during C2 loss (60s buffer)	firmware-cpp	mavlink-shim	1	3	7	3.33
95	Cloud ADS-B source fusion – dedup by ICAO	backend-go	collision-flight-controller	2	5	10	5.33
96	Traffic source attribution tagging	backend-go	collision-flight-controller	0.5	2	5	2.25
97	Source attribution display in GCS traffic view	frontend-react	apollo-frontend	0.5	2	5	2.25

Phase 4: Onboard + Cloud DAA Orchestration (15 stories, 82.3 days)

Porting detection algorithms to onboard C++, dual-mode cloud/onboard operation, handoff logic, and consistency verification between the two detection paths.

#	Story	Domain	Repo	Opt	Likely	Pess	Expected
63	Autonomous path – crouch without backend when C2 lost	backend-go	collision-flight-controller	2	5	10	5.33
78	Autonomous recovery after C2 loss – 60s clear	backend-go	collision-flight-controller	0.5	2	5	2.25
83	Onboard conflict detection – port OCDV/OCAV to C++	firmware-cpp	mavlink-shim	7.5	15	30	16.25
84	Onboard-to-cloud detection handoff / dual-mode	firmware-cpp	mavlink-shim	3	7.5	15	8.00
85	Onboard sensor failure handling per UA model	firmware-cpp	mavlink-shim	3	7.5	15	8.00
86	GPS degraded/loss behavior for onboard DAA	firmware-cpp	mavlink-shim	0.8	3	7.5	3.38
87	Cloud fallback when onboard detection active	backend-go	collision-flight-controller	2	5	10	5.33
88	Simultaneous cloud + onboard failure handling	backend-go	collision-flight-controller	2	5	10	5.33
89	Onboard ADS-B detection failure handling	backend-go	collision-flight-controller	0.5	2	5	2.25
90	Extend protos for onboard detection status + handoff	backend-go	api-private	0.5	2	5	2.25
91	Onboard vs cloud detection status indicator in UI	frontend-react	apollo-frontend	0.5	2	5	2.25
98	Onboard intrusion state transmission at 1 Hz	firmware-cpp	mavlink-shim	1	3	7	3.33
99	Detection state disagreement logging (>2s divergence)	backend-go	collision-flight-controller	2	5	10	5.33
100	Onboard encounter log buffer (30 min, 10 intruders)	firmware-cpp	mavlink-shim	2	5	10	5.33
101	Shared reference test vector set for consistency	backend-go	collision-flight-controller	3	7	15	7.67

Phase 5: Terrain & Traffic-Aware Crouch Targeting (10 stories, 52.4 days)

Terrain elevation integration, historical traffic pattern analysis, and intelligent crouch altitude selection to maximize separation from manned traffic.

#	Story	Domain	Repo	Opt	Likely	Pess	Expected
102	Terrain elevation data service integration	backend-go	collision-flight-controller	3	7	15	7.67
103	Terrain-based crouch floor computation	backend-go	collision-flight-controller	2	5	10	5.33
104	Terrain data staleness management (90-day refresh)	backend-go	collision-flight-controller	1	3	7	3.33
105	Historical ADS-B altitude aggregation pipeline	data-pipeline	collision-flight-controller	5	10	20	10.83
106	Traffic pattern weekly recomputation (30-day window)	data-pipeline	collision-flight-controller	2	5	10	5.33
107	Optimal crouch altitude computation (90% traffic band)	backend-go	collision-flight-controller	3	7	15	7.67
108	Crouch altitude delivery via pre-flight initialization	backend-go	collision-flight-controller	1	3	7	3.33
109	GCS display of configured crouch altitude + active target	frontend-react	apollo-frontend	1	3	7	3.33
110	Operator crouch altitude override UI + logging	frontend-react	apollo-frontend	1	3	7	3.33
111	Insufficient traffic data fallback + site flagging	backend-go	collision-flight-controller	0.5	2	5	2.25

Phase 6: UAS Performance Calibration & Latency Tuning (11 stories, 66.2 days)

Automated flight test infrastructure for measuring UA-specific performance characteristics, end-to-end latency profiling, and per-aircraft timing constant calibration for OCAV computation.

#	Story	Domain	Repo	Opt	Likely	Pess	Expected
112	Automated flight test protocol + execution framework	firmware-cpp	mavlink-shim	7	15	30	16.17
113	Performance profile storage per UA model + firmware	backend-go	collision-flight-controller	2	5	10	5.33
114	Profile validity management + re-test notifications	backend-go	collision-flight-controller	1	3	7	3.33
115	Envelope condition test matrix (temp/payload/battery)	firmware-cpp	mavlink-shim	3	7	15	7.67
116	End-to-end latency measurement infrastructure	backend-go	collision-flight-controller	5	10	20	10.83
117	Latency profile storage per C2 link type	backend-go	collision-flight-controller	1	3	7	3.33
118	Production latency monitoring + 50% threshold alerting	backend-go	collision-flight-controller	2	5	10	5.33
119	UA-specific timing constants in OCAV computation	backend-go	collision-flight-controller	2	5	10	5.33
120	Fallback to baseline when no valid profile	backend-go	collision-flight-controller	0.5	2	5	2.25
121	Safety margin preservation (Tmargin floor enforcement)	backend-go	collision-flight-controller	0.5	1	3	1.25
122	Performance calibration admin UI	frontend-react	apollo-frontend	2	5	10	5.33

Comparable Epics

Historical epics from the DroneUp Jira portfolio that provide calibration context:

Epic	Actual Days	Stories	Relevance
CORE-2775: Simulator Failure Injection	38.2	26	Similar firmware + backend coordination
CORE-2588: UTM ACM & SCD Enhancements	117.0	17	Airspace management domain, FAA-adjacent
CORE-1996: Altimeter Milestone 1	1,507.3	77	Multi-phase safety-critical system with certification

The DAA system at 122 stories and 587 expected days sits between the UTM and Altimeter programs in scope. The Altimeter comparison is instructive: safety-critical systems with FAA involvement consistently exceed initial estimates due to certification overhead and iterative flight testing.

Methodology

This estimate uses architecture-driven PERT (v3) methodology:

Unit: Days of active work time (not calendar time).
Per story: Three-point estimate (Optimistic / Likely / Pessimistic), minimum 0.5 days.
Expected: (O + 4L + P) / 6 per story.
Standard deviation: (P - O) / 6 per story.
Phase aggregation: Sum of expected values; sqrt(sum(stddev^2)) for phase uncertainty.
95% confidence interval: Expected +/- 2 standard deviations.
Confidence level: High when stddev/expected < 0.3, Medium when < 0.6, Low otherwise.

Repos explored: uncrew-collision-flight-controller (Go), uncrew-apollo-frontend (React/TypeScript), uncrew-api-private (Proto), uncrew-mavlink-shim (C++).

Risk flags: faa_related, certification, flight_testing, hard_repo_mavlink, airspace_deconfliction.

Evolution from prior estimate: This 6-phase structure expands the original 3-epic estimate (91 stories, 426.9 days) by adding 31 new stories across Phases 3-6 to account for onboard ADS-B sensor integration, source fusion, terrain-aware crouch targeting, and UAS performance calibration. All original PERT values are preserved exactly.

Last updated on May 11, 2026

Flight Alerter & Air Traffic Visualization Migration to Uncrew