1. Executive Summary
Quiver flew. The Houston build completed its first flights on July 9 and closed the month with eleven flights across five sessions, including a 20.7 minute endurance record that ended with the battery capacity failsafe commanding an automatic return, an event that doubled as an impromptu live demonstration of the obstacle avoidance stack when the aircraft climbed clear of trees on its way home. The campaign found and fixed a yaw torque imbalance through motor arm realignment, verified in flight, and every log is published on the flight tracking platform. The Remote ID arming blocker carried from June was resolved in the first week, traced to the MK32’s bundled SIYI build of QGroundControl, and the ESC DroneCAN concern closed without the ordered converter once the true fault was confirmed as CAN termination.
The month’s stubborn problem is the second GPS. A replacement M9N failed validation with the same velocity glitching as the original unit, pointing to systemic interference at the mounting position rather than bad hardware, and the aircraft currently flies on the primary F9P alone. A relocated mounting interface is the working fix. Obstacle avoidance moved from configured sensors to a formal ground commissioning campaign at month end, one sensor at a time under a test plan from Zeynep.
On governance, the June report was published, the May and June treasury transfer landed on July 21, and all commitment grants through June are now paid with the ledger reconciled on chain to the cent. Two bounties were converted to in house design work after going unclaimed, and a wave of AI generated bounty claims on GitHub was triaged and closed, surfacing a policy question for the DAO. The accompanying funding proposal is restructured as a phased project roadmap, a format change requested on the August 4 call to tie spending to phase deliverables rather than monthly asks.
2. Project Progress
Team Formation
The core team structure carried over from June.
| Member | Role |
|---|---|
| errrks.eth | Project Lead |
| Julius | Core Contributor |
| Zeynep | Core Contributor |
| alperenag | Core Contributor |
| Dow Fisher KBM | Core Contributor |
| 21stCenturyAlex | Core Contributor |
Progress Summary:
First flights and the campaign. After the ground run with propellers and Zeynep’s clearance, the Houston build flew four flights on July 9 in STABILIZE, ALTHOLD, and LOITER. The full first flight record lives in the Initial Configuration Guide flight log section. The campaign continued through five sessions: flights 5 and 6 on July 23, flight 7 on July 24, flights 8 and 9 on July 28, and flights 10 and 11 on July 30, eleven flights total. Flight 10 set the Houston platform endurance record at 20.7 minutes and ended when the battery capacity failsafe commanded RTL. The event doubled as an unplanned avoidance test. The BendyRuler path planner configured on the aircraft only runs in autonomous modes (AUTO, GUIDED, and RTL), and with the entire campaign flown manually, the failsafe RTL was the first time an autonomous mode had ever engaged. The planner activated for the first time, detected the trees off the left side, and steered the aircraft up and away from them rather than flying a straight line back to home. Erick recovered control by switching out of RTL from a mapped mode switch and landed normally. Two changes followed. The first battery failsafe stage has since been reconfigured to warn instead of commanding RTL, and the Pilot’s Handbook gains both a note that autonomous modes engage the avoidance path planner and a recovery procedure for overriding an active RTL. All logs are uploaded to the flight tracking platform.
Yaw imbalance found and closed. The July 9 logs showed the clockwise motors running roughly 130 PWM higher than the counterclockwise pair, a yaw torque imbalance consistent with motor arm misalignment. Alperen and KBM walked Erick through the arm adjustment procedure, the front arms were measured and realigned to within 0.1 degrees, and flight 7 verified the fix with the imbalance collapsed to under 20 PWM. A new preflight rule came out of the same investigation: check every motor arm screw and latch before each flight, since screws loosened despite thread locker.
GPS2 (open item). The M9N GPS was replaced on July 13 after the original unit was found with a popped component (Matek contact in progress). The replacement failed validation: the same velocity glitching appeared on July 23, and July 28 was the worst performance yet, with kilometer scale position disagreement at arm while the primary F9P stayed clean. The working hypothesis is systemic interference at the GPS2 mounting position, supported by a stationary test with the unit relocated to the lid that performed measurably better. The near term fix is a new mounting interface relocating the unit, with a motors on interference sweep still to run. The aircraft currently flies single GPS on the F9P, a deliberate field deviation, and automatic GPS switching stays pinned until a relocated unit validates in flight.
Remote ID resolved. The arming blocker carried from June was root caused in the first week: the MK32’s bundled QGroundControl is a SIYI fork that lacks the operator location feature Remote ID requires. Installing mainline QGroundControl on the MK32 closed the full Remote ID chain and the aircraft arms with no warnings.
ESC DroneCAN closed without new hardware. The Datalink converter ordered in June arrived but was never needed. The configuration work proved the original no reply symptom was a CAN termination fault, not a protocol limitation. With termination fixed the factory Hobbywing ESCs enumerate normally over DroneCAN, were renumbered one at a time, and flew the entire campaign on the 1 Mbit bus. The converter goes to spares.
Obstacle avoidance commissioning begins (QGB-02). With configuration behind the team, obstacle avoidance moved to the front. The avoidance behavior was deliberately disabled for the early flight campaign. Zeynep is revising her original forum plan into a specific ground commissioning test plan: verify every sensor’s response and integrity on the ground before flying, then introduce sensors one at a time, the 360 degree lidar first, then the forward radar. The first ground session ran July 31, using distance markers, one sensor at a time, walk arounds, and a fixed obstacle held for twenty minutes. As of publication the top mounted lidar has come through clean, twenty minutes on a fixed obstacle with no dropouts and no phantom artifacts, an encouraging early signal on the in flight lidar dropout question from earlier Texas testing. One anomaly is open. With the forward radar and the lidar enabled together, a phantom object consistently registers about 2.5 meters off the nose, and whether that is a radar fault or a sensor fusion artifact is unresolved. An isolated forward radar test, a 12 meter to 1 meter approach followed by a fixed object held for twenty minutes, runs next. A few more ground sessions remain before the first avoidance enabled flight, and a clean pass would put the QGB-02 bounty in reach of closing.
Camera and gimbal. MK32 dial control of the A8 gimbal was brought up on July 27, replacing the touchscreen only control that was impractical in flight. The gimbal to flight controller return path is still open, and the video stream in QGroundControl is under debug. The HM30 air unit and the forward radar were flipped 180 degrees on the CAD model, and radar parameters adjusted accordingly.
Attachments. The granular dispenser work surfaced the limits of the off the shelf JMRRC unit: electronics that overheat at full pack voltage, a spinner ESC that runs about thirty seconds, and an internal PCB that corrupts the PWM signal. Thomas rewired to bypass the internal PCB and is leaning toward an Arrow dispenser designed in house, mostly 3D printable. Grid pattern flights on the Javelina side flew with the drone performing well. A brownout during dispenser spinup, which twice dropped the radio when running from the 12V switched line, moved the dispenser to the HV line and is steering the attachment interface power philosophy toward attachments regulating their own power from pack voltage. The actuated payload latch and multispectral camera bounties were both converted to in house design work after going unclaimed (see budget section). Attachment interface replacement sourcing remains the priority hardware decision, and a group attachment build for the US meetup is being planned around off the shelf microcontrollers.
Regulatory. Thomas is actively working the FAA Declaration of Compliance filing. The FAA’s evidence requirements are now captured: the ASTM F3586-22 MOC system properties table and checklists, an aircraft level test report, an ICAO manufacturer code letter, and the FCC identifier reference for the BlueMark db201 broadcast module. Correspondence with BlueMark clarified that the aircraft manufacturer owns signal compliance, so the aircraft level Remote ID test campaign is Quiver’s to run, and BlueMark provides completed F3411 tables, a template, and a sample submission to build from.
Governance and community. The June progress report was published to the DAO forum. The May and June treasury transfer executed on July 21 and the June commitment grants were deposited to LlamaPay on July 24, bringing all commitment grants current through June with the ledger reconciled against the chain to the cent. A wave of six AI generated GitHub issues claiming documentation bounties was traced to a third party bounty board scraping published progress reports. All six were closed with a standard reply, and the episode raised a policy question worth DAO attention: how to handle third party boards reselling Arrow’s open work and bot authored PRs whose operators offer to sign the CLA on the bot’s behalf.
3. Documentation
Written this month, pending repository publication:
- First Flight Checklist, drafted from the configuration guide and Pilot’s Handbook, reviewed and approved by Zeynep. Publication to GitHub is queued behind a repository sync.
- Pilot’s Handbook post shipment recalibration section, migrating the recalibration procedure from the configuration guide into the Handbook as a standard operating step.
- GitHub failure report issue form, a structured issue template (date, unit, phase, subsystem, severity, evidence, flags) that replaces scattered Discord threads as the formal failure log.
- Attachment Developer Guide scope document, defining the guide’s contents, the verified electrical interface facts, the documentation gaps and where to source them, and milestone pricing for a bounty posting. Thomas will review the guide, authorship goes to bounty.
- Configuration guide flight records, sections capturing the first flights and each subsequent session, parameter restore points, and field deviations.
- FAA RID DoC evidence requirements, captured from the FAA portal and BlueMark correspondence for the filing work.
4. July Goals — Status Reconciliation
The June report listed 27 goals for July. First flight was the gating objective and it cleared with room to spare. Status below.
| # | Goal | Status |
|---|---|---|
| 1 | Complete first flight of the Houston build |  Flew July 9, campaign reached 11 flights |
| 2 | Resolve Remote ID CAN drop and arming block | Root caused (SIYI QGC build) and fixed July 6 |
| 3 | Resolve ESC DroneCAN reliability with the Datalink converter | Closed without it, real fault was CAN termination |
| 4 | Stabilize M9N GPS acquisition |  Replacement unit failed validation, mounting relocation is the path, flying single GPS |
| 5 | Resolve the in flight RPLidar dropout |  Carries into the OA ground commissioning designed to answer it |
| 6 | Confirm onboard logging and SD card | Proven across 26 campaign logs |
| 7 | Enter Remote ID identity once registered |  Registration not yet done |
| 8 | Close config guide gaps, reconcile first time setup payout | Guide substantially complete, payout stays under review, resolves in the August cycle |
| 9 | Migrate recalibration into Pilot’s Handbook | Drafted, publication queued behind repo sync |
| 10 | Publish first flight checklist to GitHub | Drafted and approved by Zeynep, publication queued behind repo sync |
| 11 | Write the Attachment Developer Guide | Scoped in full with gaps and sourcing mapped, work in progress |
| 12 | Audit manufacturing guide build photos | Deferred to the next real assembly |
| 13 | Correct radar and HM30 orientation, verify params | CAD updated and parameters reflect orientation |
| 14 | Resolve attachment interface supply | Replacement candidate and order quantity still undecided |
| 15 | Order Gray’s replacement PCBs, complete 12V repair | No movement |
| 16 | Confirm battery temp logging on Gray’s firmware | No movement |
| 17 | Complete V1 actuated payload latch (QGB-05a) |  Bounty dropped after stalling, latch moves to in house design |
| 18 | Get multispectral camera claimed and started | Bounty dropped unclaimed, camera moves to in house design |
| 19 | Complete dispenser wiring and Spike testing | JMRRC PCB bypassed, in house Arrow dispenser under consideration |
| 20 | Scope floodlight V1 | Not scoped |
| 21 | Progress QGB-02, QGB-03, QGB-04 | QGB-02 commissioning began, QGB-03 deferred until more flights, QGB-04 stalled |
| 22 | Bounty visibility through Discord | Not done |
| 23 | Product refinement owner and focus | Needs a scoping doc first, owner assignment pending |
| 24 | Complete marketing discovery ($375) | roughly scoped, in review |
| 25 | Complete FAA DoC filing | Actively in progress (Thomas), evidence requirements mapped |
| 26 | Define flight test campaign tiers | Held again, mechanics move to the phased project proposal |
| 27 | Execute treasury transfer, confirm airdrop | Transfer landed July 21, airdrop confirmed, ledger reconciled on chain |
Summary: seven complete, eight advanced, two converted to in house work by decision, and the rest carried with owners or decisions identified. The largest July output, a validated flying aircraft with a published log record, sits behind goal 1 but understates the month.
5. Goals for Next Month
Flight envelope expansion and the GPS2 fix are the gating objectives. The list is deliberately shorter than July’s, with items delegated or converted to bounties where possible.
Flight operations:
- Design and install the relocated GPS2 mounting interface, run the motors on interference sweep, and restore dual GPS after a clean validation flight.
- Complete the obstacle avoidance ground commissioning (360 lidar, then forward radar), deliver the lidar dropout verdict, and fly the first avoidance enabled flight (QGB-02).
- Further troubleshoot the gimbal to flight controller return path and the QGroundControl video stream.
Documentation:
- Publish the approved First Flight Checklist, the Handbook recalibration section, and the failure report form to GitHub, resolving the repository sync.
- Deep review the first time setup line against the configuration guide and reconcile the QGB-01 payout.
- Post the Attachment Developer Guide bounty from the completed scope document.
- Integrate QuiverHub onto the Raspberry Pi and bridge Alex’s documentation into the configuration guide.
Hardware and attachments:
- Decide the attachment interface replacement and order quantity.
- Decide the dispenser path, JMRRC bypass versus an Arrow design under QGB-05.
- Start the in house latch and multispectral camera designs that replace the dropped bounties.
- Produce the meetup attachment build plan and order parts.
Regulatory:
- Run the aircraft level Remote ID test campaign and assemble the FAA DoC evidence package (F3586-22 tables, test report, ICAO code, db201 FCC reference) with Thomas.
- Register the aircraft and enter the Remote ID identity.
Governance and product:
- Distribute the July retroactive grants.
- Publish the phased project proposal covering August through October, carrying the flight test campaign mechanics.
- Write the product refinement scope and assign the owner.
- File the marketing discovery deliverable and take the next step on the sales page.
- Determine uses for the California dev-kit.
6. Budget & Resource Allocation
Team Members Compensation:
The core team continued on commitment grants in July under the grants & bounties model. July commitment grants are requested retroactively in the July funding proposal accompanying this report.
July 2026 — Total Budget: $8,760 USDC + 25,000 $ARROW
| Category | Budget |
|---|---|
| Commitment Grants | $8,760 USDC (requested in the July proposal) |
| Flight Test Campaign | 10,000 $ARROW (carried from April) |
| Retroactive Grants | 15,000 $ARROW |
Disbursement status. The combined May and June treasury transfer ($17,520 USDC + 30,000 $ARROW) executed on chain July 21. June commitment grants were deposited to LlamaPay July 24, bringing all commitment grants current through June. The ledger was reconciled against the live multisig on July 30 and matches to the cent: $20,267.65 USDC + 40,850 $ARROW on the Safe.
Commitment Grants — $8,760 USDC
| Contributor | Grant (USDC) |
|---|---|
| Erick | $3,000 |
| Julius | $1,280 |
| Zeynep | $1,280 |
| Alperen | $1,280 |
| KBM | $960 |
| Alex | $960 |
Grants & Bounties
Technical milestone bounties continue to draw from the March USDC reserve ($26,240). $6,699.29 has been paid to date. Three bookkeeping changes this month. First, the V1 actuated payload latch bounty (QGB-05a, $500) was dropped after the claim stalled, and the multispectral camera bounty ($500 from QGB-FLEX) was dropped unclaimed. Both move to in house design work. No funds leave the multisig: the latch $500 returns to its parent bucket QGB-05, where the in house latch work now lives (QGB-05a was a sub-bounty carved from QGB-05), and the camera $500 returns to QGB-FLEX, the bucket it was assigned from. Second, the month end reconciliation walked this ledger against the live Safe balance to the cent and surfaced a $726.94 surplus that predates this report and has not appeared in a prior one. It decomposes into three pieces. $269.94 was already sitting in the Safe before the March allocation arrived and was never booked. $460.00 came from the May commitment grant cycle: the treasury reimbursed the full $8,760 while the Safe had only deposited $8,300, because the LlamaPay streaming contract still held a $460 residual from earlier topups that covered the difference. A $3.00 deduction applies because the March commitment grants actually cost $8,763 on chain rather than the budgeted $8,760. The full decomposition is recorded in the project ledger, and the surplus is folded into QGB-FLEX so that every dollar on the Safe sits in a named bucket. Third, going forward, dropped sub-bounties return to their parent bucket and QGB-FLEX serves as the standing catch-all for reconciliation surpluses and returns with no parent. After the returns and the fold, available bounty funds stand at $20,267.65, which matches the full USDC balance of the Safe now that all commitment grants through June are paid.
| ID | Name | Reward | Status |
|---|---|---|---|
| QGB-01 | Documentation Wrap Up | $10,000 | In Progress — checklist and Handbook updates drafted, first time setup payout under review, Attachment Developer Guide scoped and going to bounty |
| QGB-02 | Obstacle Avoidance System | $4,000 | In Progress — ground commissioning campaign began at month end under Zeynep’s test plan |
| QGB-03 | Endurance Metrics Study | $600 | Open — deferred until more flight data accumulates (flight 10’s 20.7 min endurance record is the first data point) |
| QGB-04 | Wind Limit Study | $3,000 | Open — stalled, needs a restart decision |
| QGB-05 | Attachment Development | $3,000 | In Progress — dispenser rewired past the faulty JMRRC PCB, in house Arrow dispenser under consideration, latch design folds in here |
| QGB-05a | V1 Actuated Payload Latch | $500 | Dropped — claim stalled, latch moves to in house design, $500 returns to its parent bucket QGB-05 |
| QGB-06 | Transport Case | $1,000 | Paid April 20 |
| QGB-FLEX | Open Bounty Bucket | $5,366.94 | In Progress — $4,640 base + $726.94 surplus fold. $1,729.50 paid, $375 assigned (Marketing Discovery), $3,262.44 unassigned |
| QGB-FLEX | Multispectral Camera Payload | $500 | Dropped — unclaimed, camera moves to in house design, $500 returns to QGB-FLEX |
| QGB-FLEX | Marketing & Sales Page Discovery | $375 | In review before filing |
| QGB-FLEX | QuiverHub V1 Completion & Release | $1,080 USDC + 600 $ARROW | Paid — Alex, all M1–M3 milestones closed in April |
QGB-01 Breakdown
| Item | Allocation | Status |
|---|---|---|
| Dev-Kit Information Notes | $775 USDC | Paid |
| Liquidity Pool Seeding | $994.79 USDC | Paid |
| Manufacturing Guide — Alperen | $1,000 USDC | Paid June 26 |
| Manufacturing Guide — Erick | $700 USDC | Paid June 26 |
| Manufacturing Guide — First time Setup (rolled forward) | $300 USDC | Payout under review against the Initial Configuration Guide, resolves in the August cycle |
| Engineering Report — Erick | $500 USDC | Paid June 26 |
| Platform Engineering Report (reserved) | $500 USDC | Open |
| Pilot’s Handbook | $1,000 USDC | Merged — payment pending field driven updates, now accumulating from the flight campaign |
| Developer SDK | Unpriced | Quick start merged — pending field updates and M4 & M5 (Alex) |
| Maintenance Guide | Unpriced | Merged — payment pending field driven updates |
| Attachment Developer Guide | $750 USDC | Scoped — going to bounty posting, Thomas reviewing |
| Unassigned Funds | $3,480.21 USDC | After priced items |
Retroactive Grants — 15,000 $ARROW
July retroactive grants recognize the flight campaign and the work around it. The distribution below draws on the held retro reserves on the multisig, consistent with the June decision to hold rather than distribute.
| Contributor | Amount |
|---|---|
| Thomas | 2,000 |
| Erick | 1,000 |
| Zeynep | 1,000 |
| KBM | 750 |
| Alex | 750 |
| Alperen | 500 |
| Distributed | 6,000 |
| Held in reserve | 9,000 |
The remaining 9,000 of the July allocation stays held on the multisig with the prior months’ reserves.
Flight Test Campaign — 10,000 $ARROW
The 10,000 $ARROW pool approved in April stays held on the multisig, undistributed. Rather than set per flight rates mid campaign, the mechanics fold into the phased project proposal now in drafting, where flight operations are costed alongside the rest of the roadmap. No payments draw from the pool this month.
Boosh April retro — released. The 350 $ARROW owed to Boosh since April was released on July 30 after 90+ days without a wallet address or response. The amount returns to the unallocated retro pool and can be reinstated if Boosh resurfaces.
Total:
July commitment grants ($8,760 USDC) are requested retroactively in the July funding proposal. All prior commitment grants are paid through June and the multisig ledger is reconciled on chain to the cent ($20,267.65 USDC + 40,850 $ARROW). The bounty reserve stands at $20,267.65 after the two dropped bounties returned $1,000 to QGB-FLEX and the $726.94 reconciliation surplus was folded into the same bucket, so the reserve now equals the full USDC balance of the Safe. The July retroactive distribution (6,000 $ARROW across six contributors, 9,000 held) draws from the held retro reserves at month end.