Skip to content

REVIEW.md

Latest commit

 

History

History
436 lines (290 loc) · 45.7 KB

REVIEW.md

File metadata and controls

436 lines (290 loc) · 45.7 KB
Raw

Review note: PalletPilot

Session 2026-09-25: /populate to a strong TRL 2

Overnight batch run; no questions were asked. Every choice that is Amish's to make is listed below as proposed.

What was done

  • docs/01-problem.md (PLP-PRB-001 v0.2): problem with ergonomic push-force data, users and context, operating environment, constraints (including OSHA 1910.178 training and modification approval, and ISO 3691-4), out of scope, prior work with inline sources, open questions; co-design checklist added in the portfolio's standard form.
  • docs/03-requirements.md (PLP-REQ-001 v0.2): 17 measurable requirements (R1 to R17) with targets, verification and a status column, a design load case and design duty, and a list of requirements not met.
  • docs/02-concept.md (PLP-PRC-001 v0.2): how it works, components table numbered to the BOM and exploded view, first-order numbers (forces, traction, stopping, UWB bearing error, energy, mass and cost) with assumptions, key design choices, safety section, open questions.
  • cad/src/concept_media.py: massing model of a 27 x 48 in donor jack (grey) and 13 numbered kit parts, with a 48 x 40 in pallet and cartons as hero context.
  • media/: hero.png, concept-blueprint.png, .pdf and .svg, model.glb and viewer.html, exploded.png with BOM callouts, cutaway.png, flow.png (energy per shift, estimates marked).
  • bom/bom.csv: 17 lines with indicative USD prices, rows 1 to 13 numbered to match the exploded view; bom/bom-notes.md updated.
  • README.md: hero image, links line, updated problem, concept, components and safety.
  • docs/pdf/: branded PDFs of the three controlled documents.
  • project.yaml: unchanged. The pitch and problem are still accurate (see decision 1 on the pitch wording).

Key results (estimates, to be checked at TRL 3)

Quantity Estimate Requirement
Rolling and breakaway force, 1,100 kg total about 130 N and 270 N Motivates the kit
Available traction (1.1 kN preload, friction 0.5) about 550 N R2 met; R3 (2 % ramp) at risk
Stop from 1.2 m/s, walk mode about 1.56 m Normal walkie practice
Stop from 0.6 m/s, follow mode about 0.42 m, against about 40 mm bumper travel R7 not met
Bumper-only safe speed about 0.2 m/s R8 met at that speed
UWB bearing error, two anchors 450 mm apart about ±18° raw R5 (±10°) not met on paper
Energy per shift, 60 moves about 281 Wh of 410 Wh usable R12 met
Charge time about 4.5 h R13 met
Kit mass about 37 kg R15 mass met
Added length at floor level about 330 mm R15 length (300 mm) not met
Kit parts cost about $1,425 R17 ($1,200) not met, about 19 % over
Walk-only variant about $1,075 Within budget

Requirements not met or at risk: R7 (non-contact stop in follow mode), R5 (bearing accuracy), R15 (added length, by about 30 mm), R17 (cost) and R3 (2 % ramp, traction margin). R1, R6, R9, R10, R14 and R16 are unverified.

Market context found: a complete 1,500 kg lithium walkie now costs about $1,640 and the PowerPallet 2000 retrofit about $2,084. A walk-only kit plus a new donor jack costs about the same as a new walkie, so PalletPilot's case rests on follow mode and on reusing existing jacks.

Proposed, awaiting Amish (now decided)

All nine items below: Decided by Amish, 2026-09-25: go with recommendation (PLP-DDR-001, D1 to D9).

  1. Follow-mode stopping and the pitch. Options: (A) add a low-cost lidar or time-of-flight sensor as the main stopping layer, bumper as last resort, follow mode 0.6 m/s, research use in a closed area; (B) safety-rated laser scanner (over $1,000 on its own); (C) bumper only with follow mode at 0.2 m/s; (D) drop follow mode. Recommendation: A for the prototype, with B named as the route to workplace use. This would change "bumper-based stopping" in the pitch to "layered stopping"; project.yaml is unchanged until you decide.
  2. Budget. Options: (a) raise budget_usd from $1,200 to about $1,450 (about $1,550 with option 1A); (b) keep $1,200 and cut to the walk-only variant (about $1,075); (c) keep $1,200 and cut elsewhere (for example a single hub motor plus a steering actuator, cheaper bumper). Recommendation: (a), since follow mode is the reason to build the kit. project.yaml is unchanged at $1,200.
  3. Drive layout: two hub motors on a sprung module on the steering yoke, steering by differential drive in follow mode. Alternatives: one motor plus a steering motor, or load-rated drive wheels replacing the steer wheels. Recommendation: dual hub motors.
  4. Power system: 24 V LiFePO4, 20 Ah. The SwapCell 48 V pack is not proposed because the kit needs only 24 V and stays on the jack all shift. Recommendation: 24 V LiFePO4.
  5. Mounting: pack and electronics on the steering yoke, not the forks. Recommendation: yoke mounting, checked for handle clearance and bearing load at TRL 3.
  6. Speeds: walk 1.2 m/s handle-end first and 0.8 m/s forks first, creep 0.3 m/s, follow 0.6 m/s (0.2 m/s until option 1 is built and tested).
  7. Kit rating: 1,000 kg design load, 1,500 kg maximum at 0.8 m/s or less, below the donor's 2,500 kg.
  8. Legal route for workplace use (OSHA 1910.178(a)(4) needs the jack maker's written approval for modifications affecting capacity and safe operation). Options: partner with a jack maker for one donor model; document the kit for research and non-workplace use only; or both. Recommendation: document for research use now, and approach a jack maker before any workplace trial.
  9. First site partners: a small warehouse or 3PL, a grocery or retail backroom, or a maker space. Recommendation: one small warehouse and one maker space.

Safety concerns

  • A loaded jack in follow mode will strike a person before a bumper can stop it at any speed above about 0.2 m/s. This is the main safety finding and drives decision 1.
  • Foot crush and pinning at the drive end are the classic walkie injuries; the belly-reverse paddle, handle-angle braking and bumper hoop address them but are unverified.
  • Braking falls by about 40 % on a 2 % downgrade; ramps and dock plates need limits.
  • The 512 Wh LiFePO4 pack needs a BMS, a terminal fuse, a lockable disconnect and safe charging practice.
  • Workplace use raises OSHA training and modification-approval duties; the kit must be presented as a research prototype.

Gaps against the brief

  • The OSHA page could not be fetched during the session (the permission request timed out); the 1910.178 citations use the official regulation URL and a secondary summary for the training requirement. The wording of 1910.178(a)(4) should be checked against the source.
  • Hub motor torque, brake holding torque and prices are for the part class, not a chosen supplier.

Recommended next step

Review this note and the media, and decide items 1 and 2. If approved, run /advance-trl3 to size traction, braking and the stop chain by calculation, choose the motors and the follow-mode sensor, build the UWB error budget, and produce the parametric model and drawing sheet.

Session 2026-09-25: TRL 3

Amish's instruction for this session (2026-09-25): "proceed with all of your recommendations across all batches. Make sure we don't proceed to TRL 4 on any of them." For PalletPilot the approved safety recommendation is the low-cost lidar or time-of-flight stop layer (option 1A), the approved budget is about $1,550, and the pitch changes to "layered stopping". PalletPilot now claims TRL 3 (proof of concept on paper). TRL 4 is on hold by Amish's instruction.

What was done

  • docs/decisions/0001-trl2-review-decisions.md (PLP-DDR-001 v0.1): nine TRL 2 items recorded as decided by Amish, 2026-09-25, going with the recommendation (D1 to D9), and six items that stay open (O1 to O6).
  • docs/04-calcs/01-sizing.md (PLP-CAL-001 v0.1) and docs/04-calcs/sizing.py: mass, axle and bearing loads, traction and preload, motor torque and power, steering by differential drive, layered stopping (lidar, bumper, emergency stop, tag loss, parking), protective field sizing, UWB error budget, energy and charging, handle clearance, length, release lever, cost, and a status line for every requirement. The script imports the model's PARAMS and the BOM, and prints every number the note quotes, tagged [B0] to [R0].
  • cad/src/model.py: parametric build123d model (donor jack as reference; subframe with yoke clamp, hub motors, enclosure and lid, pack, driver, contactor, controller, tiller head, beacon, emergency stops, bumper hoop, UWB anchors, release lever, lidar and bracket) with PARAMS and derived(). Exports cad/step/ and cad/stl/: palletpilot-assembly, palletpilot-kit, drive-module, enclosure, bumper-and-sensors and tiller-head.
  • cad/src/sheets.py and cad/drawings/PLP-DWG-001.svg, .pdf, .png: general arrangement at Rev P1, scale 1:20, with length at floor level, added length, steer axis to drive axle, overall height, pivot height, width, bumper width, anchor baseline and drive track drawn from the model, an isometric view and a dimensions box. It carries "CONCEPT, NOT FOR FABRICATION" and "PRELIMINARY, NOT FOR FABRICATION". The concept sheet is PLP-DWG-010, so DWG-001 was free.
  • bom/bom.csv and bom/bom-notes.md: all 18 lines priced with a supplier or supplier type; new line 18 (obstacle lidar); motor spec raised to 30 N·m peak; two lines checked against live listings.
  • cad/src/concept_media.py now builds the media from model.py, with key figures and flow values from PLP-CAL-001. Every image in media/ was regenerated and checked by eye (exploded callouts reordered by BOM number); the temporary media/_views* and cad/drawings/_views folders were deleted.
  • PLP-PRB-001, PLP-PRC-001 and PLP-REQ-001 revised to v0.3; README.md updated (TRL 3, layered stopping, $1,550, links); project.yaml set to trl: 3, trl_target: 3, budget_usd: 1550, new pitch, evidence listed. PDFs are in docs/pdf/.

Requirements (PLP-CAL-001, Table 5)

9 met, 3 not met, 3 at risk, 3 not verifiable at TRL 3 (R15 counted as two lines, mass and length).

ID Status Value against target
R5 Not met Bearing about ±16° (2σ) after filtering against ±10°; ±10° would need a 706 mm anchor baseline, wider than the 685 mm jack. Gap ±0.06 m is fine
R15 mass Not met 41.7 kg against 40 kg; the 30 N·m motors weigh about 7 kg each
R17 Not met $1,580 against $1,550 ($30, 1.9 %); within motor pricing uncertainty
R7 At risk Lidar stop 0.42 m level, 0.76 m worst case, inside a 0.86 m field that stays 0.19 m clear of the operator; sensor not safety-rated
R8 At risk Chain acts in 0.10 s, but 0.2 m/s needs 62 mm of edge travel against 40 mm; 0.15 m/s fits
R9 At risk Single contactor output; PL d not calculated
R1 Not verifiable at TRL 3 Needs a donor survey
R6 Not verifiable at TRL 3 Timing budget 0.20 s against 0.3 s; no firmware
R16 Not verifiable at TRL 3 Set by bought-part specifications
R2, R3, R4, R10, R11, R12, R13, R14, R15 length Met 25.5 of 30 N·m; ramp margin 1.54; brakes 40 against 21.7 N·m; handle clearance 40 mm at 70°; 292 of 410 Wh; 4.5 h charge; +3.9 % push force; 290 mm added length

Other key numbers: traction 750 N at 1.5 kN preload; peak 554 W and 31 A; emergency stop from 1.2 m/s 1.62 m level and 2.66 m on a 2 % downgrade; wall energy 350 Wh per shift.

Design changes made while sizing, within the decided configuration: preload raised from 1.1 to 1.5 kN (R3 was at risk); motor spec raised to 30 N·m peak (the TRL 2 class was too weak); enclosure lowered below the handle pivot (the TRL 2 layout would have hit the handle at about 51°) with the beacon moved to the tiller head and the second emergency stop to the enclosure's rear; drive axle moved 40 mm closer and the track widened to 260 mm (R15 length now met); lidar placed inside the bumper face with its scan plane at 200 mm. Every TRL 2 number in the docs was checked against the script and corrected (PLP-CAL-001, Table 6). TRL 2 overstated the bumper-only safe speed (0.2 m/s; it is 0.15 m/s).

Other findings: the motors cannot swing the yoke at standstill (98 N·m available against 135 N·m of scrub), so follow mode must steer only while rolling, and standstill hand steering gets about 112 N heavier at the grip. The 1,500 kg rating holds on level floors only (a ramp start needs 35.4 N·m per motor).

Decisions recorded (PLP-DDR-001)

Decided by Amish, 2026-09-25, going with the recommendation: D1 layered stopping with a low-cost lidar layer, bumper last, follow 0.6 m/s, research use in a closed area, safety-rated scanner named for workplace use, pitch "layered stopping"; D2 budget $1,550; D3 dual hub motors on a sprung yoke module; D4 24 V LiFePO4 20 Ah, not SwapCell; D5 everything on the yoke; D6 speeds 1.2, 0.8, 0.3 and 0.6 m/s (follow 0.2 m/s until the lidar layer is tested); D7 1,000 kg design load, 1,500 kg at 0.8 m/s or less; D8 research use now and a jack maker's written approval before any workplace trial; D9 partner types, one small warehouse and one maker space. budget_usd is now 1550 and the pitch is reworded in project.yaml and README.md; the problem line is unchanged (no rewording was recommended). The SwapCell interface decisions do not apply, since PalletPilot has its own pack.

Proposed, awaiting Amish (items 4 to 6 now decided)

  1. Named site and co-design partners (O1); partners are picked per area later.
  2. Follow-mode classification under ISO 3691-4 (O2). No recommendation.
  3. Donor models to support first (O3). No recommendation.
  4. Bumper-only stopping (O4, R8): (a) limit bumper-only speed to 0.15 m/s, or (b) an edge with 65 mm or more of travel (adds about 25 mm of length, to about 315 mm, over the 300 mm in R15). Recommendation: (a), since the lidar is the primary layer in follow mode. Decided by Amish, 2026-09-25: go with recommendation (PLP-DDR-002).
  5. Overruns (O5): cost $30 over and mass 1.7 kg over. Options: accept both as within estimate uncertainty and recheck when a motor is quoted; choose a lighter motor with published torque; or raise the budget to $1,600. Recommendation: accept for now and recheck at the motor quote, with no budget change. R5: angle-of-arrival UWB or lidar leg tracking; recommendation: evaluate both at TRL 4, when allowed. Decided by Amish, 2026-09-25: go with recommendation (PLP-DDR-002; the R5 evaluation is on hold with TRL 4).
  6. Stop chain (O6, R9): add a second contactor or a driver with rated safe torque off (about $30). Recommendation: add the second contactor, which would push cost to about $1,610. Decided by Amish, 2026-09-25: go with recommendation (PLP-DDR-002).

Safety concerns

  • The lidar layer is not safety-rated; its range on black targets falls to 6 m and detection of dark clothing at shin height is unproven. Follow mode stays at 0.2 m/s or less in a closed area with no bystanders until tested.
  • A 40 mm safety edge stops the loaded truck only from 0.15 m/s or less.
  • Emergency braking on a 2 % downgrade needs 2.7 m from 1.2 m/s; no 1,500 kg loads on ramps.
  • The 512 Wh LiFePO4 pack needs its BMS, terminal fuse, lockable disconnect and safe charging practice.
  • The drive adds up to 750 N of horizontal load at the donor's steering bearing; worn jacks must not be converted.
  • Workplace use needs the jack maker's written approval under 29 CFR 1910.178(a)(4) and trained operators.

Citations

The 1910.178(a)(4) and (l)(1)(i) wording was checked against the OSHA page in this session and now appears as a quotation in PLP-PRB-001. The motor and lidar prices were checked against the listings cited in bom/bom-notes.md. No unchecked citation flags remain.

Gaps against the brief

  • Hub motor mass, brake torque and 24 V availability are for the part class, not a quoted motor; R10 and the mass figure rest on the specified minimums.
  • No TRL 4 material exists in the repo; build-log/README.md is the scaffold file only.

Recommended next step

TRL 4 is on hold by Amish's instruction. Next, decide items 4 to 6 above, then run a paper follow-up at TRL 3: a donor survey from published jack drawings (yoke geometry, pivot height), a motor quote with brake data, and a PL estimate of the stop chain. For reference only, TRL 4 would need: a bench build of the drive module on one donor jack, measured rolling resistance, friction and stopping distances, lidar detection tests with dark clothing and a test piece, UWB bearing trials, a stop-chain test, a TST report with environment: lab, and build log entries.

Session 2026-09-25: recommendations accepted

Amish wrote on 2026-09-25: "i accept all your recommendations, go with them across all repos." Every open item with a recommendation is now decided in favor of it and recorded in docs/decisions/0002-recommendations-accepted.md (PLP-DDR-002). Items without a recommendation stay open. TRL stays at 3.

Decisions applied and what changed

Item Decision Before After
O4, bumper-only stopping (R8) Option (a): bumper-only speed limit 0.15 m/s; follow mode also 0.15 m/s until the lidar layer is tested Limit 0.2 m/s, 62 mm stop against 40 mm travel, R8 at risk; interim follow 0.2 m/s Limit 0.15 m/s, 38 mm stop, R8 met on paper; interim follow 0.15 m/s
O5, overruns and R5 Accept the mass and cost overruns for now, recheck at a motor quote, no budget change; evaluate angle-of-arrival UWB and lidar leg tracking at TRL 4 (on hold) Budget $1,550 Budget $1,550 (unchanged); R5 evaluation on hold. Cost overrun decided by Amish, 2026-09-26: budget set to $1,610 (PLP-DDR-002 v0.2)
O6, stop chain (R9) Add a second main contactor in series, one per safety relay channel One contactor; BOM line 6 $45; BOM $1,580; kit 41.7 kg Two contactors; line 6 $75; BOM $1,610 (3.9 % over); kit 42.0 kg

Files changed: docs/04-calcs/sizing.py and PLP-CAL-001 v0.2 (stop table, [S3], requirement table, mass, cost); PLP-REQ-001 v0.4 (R4, R8, R9 restated; status); PLP-PRC-001 v0.4 (components, key numbers, design choices, safety); PLP-DDR-001 v0.2 (O4 to O6 marked decided); new PLP-DDR-002 v0.1; bom/bom.csv line 6 and bom/bom-notes.md; cad/src/model.py (two contactors) with STEP and STL re-exported; cad/src/sheets.py and PLP-DWG-001 to Rev P2; cad/src/concept_media.py key figures and flow (pack 292 to 293 Wh) with all media regenerated; README.md; project.yaml (DDR-002 added as evidence; budget, pitch and problem unchanged). Small knock-on changes: push-force rise 3.9 % to 4.0 %, yoke load at 1,500 kg 6.80 to 6.81 kN.

The README gained the sections Concept rationale, Burning platform, Where it could be used and What sparked the idea. All docs PDFs, the drawing and the media were regenerated with the designmolecule.com footer.

Requirement status (PLP-CAL-001 v0.2)

10 met, 3 not met, 2 at risk, 3 not verifiable at TRL 3 (was 9, 3, 3, 3).

ID Status Value against target
R5 Not met Bearing ±16° (2σ) against ±10°; fixes evaluated at TRL 4, on hold
R15 mass Not met 42.0 kg against 40 kg; accepted for now
R17 Not met $1,610 against $1,550 ($60, 3.9 %); accepted for now
R7 At risk Lidar stop 0.42 to 0.76 m inside a 0.86 m field; sensor not safety-rated
R9 At risk Two contactors in series; PL not calculated
R1, R6, R16 Not verifiable at TRL 3 Donor survey, firmware, bought-part specifications
R2, R3, R4, R8, R10, R11, R12, R13, R14, R15 length Met R8 now met: 0.15 m/s stops in 38 mm of 40 mm travel

Still awaiting Amish

  • O1: named site and co-design partners (picked per area later).
  • O2: classification of follow mode under ISO 3691-4. No recommendation.
  • O3: donor jack models to support first. No recommendation.

Cross-repo actions

None. PalletPilot has its own 24 V pack and no interface with another repo.

Safety

  • R8 now rests on a speed limit, so the 0.15 m/s cap in bumper-only operation and in untested follow mode must be enforced by the controller and checked in any later firmware review.
  • The two-contactor chain still needs a PL calculation before any claim toward PL d.
  • The other safety concerns in the TRL 3 session stand.

TRL 4

TRL 4 remains on hold by Amish's instruction. No build, test, purchase, PCB or firmware work was done.

Recommended next step

Decide O2 and O3. Paper work that remains at TRL 3: a motor quote with brake data to recheck the mass and cost overruns, a PL estimate of the two-contactor stop chain, and a donor survey from published jack drawings.

Session 2026-09-26: sources strengthened

Amish asked on 2026-09-26 to fix the weaker sources in the README. Changes (README only; no controlled document changed):

  • What sparked the idea: VentureBeat replaced by Piaggio Fast Forward's own press release of October 15, 2019 (GlobeNewswire), which gives the $3,250 price, the November 18, 2019 sale date, 6 mph and 40 lb. The wording now says "visual sensors", as the release does. The easyPILOT Follow mention now links Jungheinrich's press release. INSPIRATIONS.md line updated to the new source.
  • By country or region: Japan now cites the Statistics Bureau of Japan (29.3 % aged 65 or over, October 1, 2024). India and Brazil were rewritten to state their World Bank Logistics Performance Index 2023 scores (3.4 and 3.2), the only claim the source supports. Kenya and East Africa was replaced by South Africa (LPI 2023 score 3.7), because no Kenya figure could be verified in the LPI 2023 table.
  • United States: OSHA 1910.178 now carries a link to the standard.
  • Kept and rechecked: BLS warehousing rate (4.8, 2024), BLS private industry rate (2.3, 2024), EU-OSHA MSD figures.

Session 2026-09-26: budget approved

On 2026-09-26 Amish wrote: "i approve all the budget items." The cost overrun accepted for now under O5 is decided: budget set to $1,610 to cover the priced BOM, recorded in PLP-DDR-002 v0.2.

  • project.yaml: budget_usd 1550 to 1610. The priced BOM is exactly $1,610 over 18 lines, so the figure covers it with no margin.
  • R17: not met ($60 over $1,550, accepted for now) to met. Requirement status is now 11 met, 2 not met (R5, R15 mass), 2 at risk and 3 not verifiable.
  • docs/04-calcs/sizing.py now checks against $1,610 and was rerun; PLP-CAL-001 v0.3, PLP-REQ-001 v0.5, PLP-PRC-001 v0.5, PLP-PRB-001 v0.4, README.md and bom/bom-notes.md quote the new figure. The concept blueprint quotes the kit cost only, so media/ was not regenerated.
  • The motor quote remains the next paper check for the R15 mass overrun and for R17, which has no cost margin left. O1 to O3 are still awaiting Amish.

Session 2026-09-26: product appearance model and photoreal renders

Amish chose this repo for the first batch of product renders on 2026-09-26. This session added cad/src/product_model.py, an appearance model for photoreal renders, and pointed the README hero image at media/render-hero.png with a link to media/render-exploded.png. The render files are produced separately.

What product_model.py adds

  • product_parts(), TITLE and three RENDER_VIEWS: hero (kit on the loaded jack with the operator), exploded, and a detail view of the drive end without context.
  • Drive module: filleted subframe plate, trailing cheeks with rounded ends, bolt heads and axle nuts, coil preload springs on their seats.
  • Hub motors: grooved tyres, machined hub faces with face bolts, cable exits.
  • Enclosure: filleted teal body with side louvers and a lid gasket line, a charcoal lid frame with a clear polycarbonate window over the pack, BMS, driver, contactors, controller and safety relay (lit status light), lid screws, cable glands, name plate and battery warning label, the rear emergency stop on its yellow plate, and the red disconnect knob.
  • Contact bumper split into a yellow steel hoop and a black safety edge with marking; white UWB anchor radomes on the corners; lidar puck with base, scan window and cap on its bracket; manual release lever with a red grip.
  • Tiller head: body, rubber grips with end caps, thumbwheel throttles, belly-reverse paddle, key switch, horn button, emergency stop, UWB anchor and an amber beacon lit for walk mode; spiral-wrapped cable from a lid gland up the handle.
  • Context: the donor jack, a 48 x 40 in stringer pallet with cartons, and the shared clay mannequin in a push pose holding the grips, with a belt and the operator UWB tag.

Differences from model.py

Each is Proposed, awaiting Amish. No controlled document or model.py dimension was changed.

  1. Handle angle. The renders show the handle at 40 degrees from vertical (mid walk band); model.py shows 20 degrees, the upright edge of the band. At 20 degrees the grips would sit at about 1.24 m, above a natural hand height for a 1.75 m operator. Recommendation: accept 40 degrees for renders only and leave model.py at 20 degrees.
  2. Tiller head width. The grips span 248 mm over their end caps against the 220 mm head_w, so two hands fit beside the head body. Recommendation: accept for renders and revisit head_w when a tiller head is chosen.
  3. Tiller UWB anchor position. The anchor radome sits under the left grip on the head body instead of 35 mm outboard of the head, where the grips now are. Recommendation: accept; the anchor stays on the head as the concept requires.
  4. Parts not modelled in model.py. The operator tag (BOM 15) and the spiral-wrapped tiller cable (BOM 17) are shown for the renders. Recommendation: keep them in the appearance model only.

TRL

This is an appearance model only: no tolerances, fabrication detail, PCB layout or firmware. trl stays 3 and TRL 4 remains on hold by Amish's instruction.

Session 2026-09-27: kit 1.5.0 and image quality

  • Kit 1.5.0 synced: STANDARDS v1.5 (sections 12 to 15: product renders, storefront images and image quality, public release, authorship and signing), .kit/cards.py, .kit/image_qc.py, .kit/release_gate.py, issue templates, and the /render-product and /release commands. CLAUDE.md now matches .kit/CLAUDE.md.
  • Every media/render-*.png recaptioned from its original render with the new layout: the title, concept label and repository sit in a band above the render and the view note in a band below it, each line wrapped to the image width, so no text overlaps other text or the render or runs off the image. media/card.png and media/social-preview.png regenerated with the same rules.
  • python .kit/image_qc.py and python .kit/release_gate.py pass. trl stays 3.

Session 2026-10-01: kit 1.7.0, constructable design and prototype build plan

Kit 1.7.0 installed (.kit/, .claude/commands/, CLAUDE.md). Following .claude/commands/build-plan.md and STANDARDS section 18, the model was checked for constructability with build123d and made buildable under Amish's 2026-09-30 instruction ("If you are realising that the design cannot be built as per concept - fix the design assumptions to match and be physically feasible as you draw the illustrations."). No git commit or push was made in this copy.

Design changes made for construction (PLP-DDR-003, Draft, open for Amish's review)

  • Yoke clamp. The 20 mm slab and loose tongue became a 6 mm top plate resting on the yoke plate with a 62 mm notch round the pump, and a 10 mm lower jaw with a packing bar; four M12 bolts behind the yoke's edge grip the yoke plate. Nothing is drilled in the jack (R1).
  • Sprung drive arms. The welded trailing cheeks and floating spring towers became two 10 mm drive arms on a 20 mm pivot pin in rear hangers behind the wheels, each pushed down by a 100 N/mm spring at its front end (arm ratio 1.77; 423 N per spring gives 750 N per wheel).
  • Hub motor mounting. Each motor's shaft passes through its arm with a 5 mm spacer and a nut.
  • Manual release. The unconnected lever became a 16 mm cam shaft in front hangers with two 22 mm eccentric cams, a crank, a link and a 400 mm over-center lever on a bracket with a stop (a parallelogram). Raising the lever lifts the saddles 22 mm and, through slotted straps, the drive wheels 10 mm clear; peak effort about 58 N (was estimated at 117 N).
  • Bumper. The solid block mounted on the moving cheeks became a 40 x 20 x 2 tube hoop bolted into rivet nuts on brackets welded under the top plate, with the safety edge on the front (50 mm deep, 40 mm travel) and both sides; side legs shortened to start 120 mm behind the steering axis.
  • Lidar and anchors. Lidar on a bent strap bracket on the hoop; the two anchors on corner plates on the hoop, tops at 180 mm, below the 200 mm scan plane.
  • Enclosure. Four riveted angle feet bolted to the top plate; holes for the rear stop and the disconnect; box 89 mm plus a 6 mm lid (95 mm as before).
  • Steering. The kit's drive end reaches the jack's frame head at about 44° of turn on the reference donor (the top plate corners at 34°). Top plate corners cut back 50 mm and two rubber steering stops (new BOM line 19) meet the frame at 40° each way.
  • Donor reference geometry corrected so the donor's own parts no longer overlap (yoke plate above the steer wheels, frame head 100 mm ahead of the steering axis).

python cad/src/model.py --check passes: 36 components, no overlaps, all 46 contacts touch, no collision with the lever raised, and the stops meet the frame before any other kit part.

Results and numbers

  • Kit mass 44.4 kg (was 42.0 kg); R15 mass not met by 4.4 kg.
  • Value-engineering target: USD 1,610. Estimated cost of the constructable design: USD 1,690 (USD 80 over the target); lines 1, 3 and 13 repriced, line 19 added.
  • PLP-CAL-001 v0.4 rerun: worst motor torque 25.6 N·m of 30; ramp margin 1.53; bumper-only stop 39 mm of 40 mm; 28 % of usable energy left; release effort 58 N; steering range [G5]. Requirement status unchanged: 10 met, 2 not met (R5, R15 mass), 2 at risk (R7, R9), 3 not verifiable (R1, R6, R16); R17 is reported against the target.
  • Documents: PLP-BLD-001 v0.1 (docs/05-build-plan.md, new), PLP-DEC-001 v0.1 (docs/06-design-decisions.md, new), PLP-DDR-003 v0.1 (new), PLP-CAL-001 v0.4, PLP-REQ-001 v0.6, PLP-PRC-001 v0.6, bom/bom.csv, bom/bom-notes.md, README.md (links line and "Building the prototype"), project.yaml (design_state: constructable, evidence).
  • Drawings and pictures: PLP-DWG-001 Rev P4; making sketches PLP-DWG-101 to 111 (11); 10 joint close-ups; 15 step pictures; overview; block-level wiring; concept media (PLP-DWG-010 Rev P2) and STEP/STL regenerated from the model. Pictures come from cad/src/build_plan_media.py.

Proposed, awaiting Amish (in PLP-DEC-001)

  1. Steering range of about 40° each way with the kit fitted, against about 90° bare; the precis no longer says the kit steers like the bare jack. Recommendation: accept for the prototype and measure the turning circle at TRL 4.
  2. Kit mass 44.4 kg against 40 kg. Recommendation: accept for the prototype, recheck at the motor quote.
  3. Wheel lift of 10 mm with the release raised. Recommendation: accept for smooth indoor floors.
  4. to 6. Still open from PLP-DDR-001: ISO 3691-4 classification (O2), donor models (O3), named partners (O1).

Stale until regenerated on Amish's Mac

The photoreal renders (media/render-hero.png, media/render-exploded.png, media/render-detail.png; not present in this cloud copy, made on Amish's Mac), media/card.png and media/social-preview.png, and the appearance model cad/src/product_model.py, still show the concept's trailing cheeks, spring towers, solid bumper block and unlinked release lever. The design changed visibly (sprung arms, tube hoop, cam-and-link release, clamp jaw), so all of them are stale.

Safety

The build plan carries eight safety stops (battery handling, first power with the wheels lifted, stop-chain checks before the motors turn, cordoned area, no follow mode until the lidar layer is tested, charging). The welds on the hangers and hoop carry the drive and spring loads and need a competent welder. The steering stops prevent steel-on-steel contact with the jack frame at full turn.

Recommended next step

Amish to review PLP-DDR-003 and decide open decisions 1 to 3 in PLP-DEC-001; then the donor survey (O3), which sets the packing bar, clamp, notch and stop angle. TRL 4 stays on hold.

Session 2026-10-01: steering range decided

Amish, 2026-10-01: "i accept your recommendations for PalletPilot". He was shown the recommendation for open decision 1 of the design decisions register (PLP-DEC-001 v0.1), the steering range, so only that item (PLP-DDR-003, A1) is decided. The kit mass (A2), the wheel lift (A3), the Table 1 changes of PLP-DDR-003 and the other open items stay open. No design change; trl stays 3; no build or test work was done.

Accepted, as recommended

Register item (v0.1) Decision
1 Steering range, option (a): accept about 40° each way with the kit fitted for the prototype; set the rubber stops from the donor survey; correct the wording so the kit no longer claims to steer like the bare jack. Go or no-go gate: the TRL 4 aisle test must show the kit can make a right-angle turn into a standard pallet bay; if it cannot, the drive layout changes before TRL 5 (option c, for example one centre drive wheel under the yoke, which reopens D3)

What changed

  • docs/06-design-decisions.md (PLP-DEC-001 v0.2): item 1 moved to Decisions made, dated 2026-10-01, with Amish's words, the go or no-go gate and the record (PLP-DDR-003, A1); open items renumbered 1 to 5.
  • docs/decisions/0003-design-for-construction.md (PLP-DDR-003 v0.2): status line records that A1 is decided with Amish's words, and that Table 1, A2 and A3 stay open; Table 3 marks A1 accepted, with the gate added to its recommendation; a consequence states the range and the gate.
  • Steering wording: the range is now stated plainly as about 40° each way with the kit fitted, set by rubber stops (about 90° for a bare jack), a turning radius of roughly 1.5 m about the load wheels, confirmed in the TRL 4 aisle test.
    • README.md: walk-mode paragraph.
    • docs/02-concept.md (PLP-PRC-001 v0.7): summary (no longer points at an open decision) and walk mode, which said only that "the handle steers freely".
    • docs/03-requirements.md (PLP-REQ-001 v0.7): no requirement covers manoeuvrability, so none was added or restated; a TRL 4 verification note for the steering range sits with the requirements not met or at risk, with the aisle test as its pass criterion and gate.
    • docs/05-build-plan.md (PLP-BLD-001 v0.2): section 3.2 (steering stops) states the range and radius; Table 3 (first checks) gains an "Aisle turn" check; references to PLP-CAL-001 v0.5 and PLP-REQ-001 v0.7.
    • docs/04-calcs/01-sizing.md (PLP-CAL-001 v0.5): the steering paragraph called the range "open decision 1"; it now records the decision, the radius and its basis (load rollers 1,130 mm from the steering axis on the reference donor). No number changed; the script was not rerun.
  • project.yaml pitch, CITATION.cff and docs/01-problem.md were checked: none claims the kit steers like the bare jack (the pitch is "Retrofit kit that turns a manual pallet jack into a powered, walk-behind unit..."), so they are unchanged. The bare-jack claim survives only as history in PLP-DDR-003 and the register's decision text.
  • PDFs regenerated with python3 .kit/render.py.

The roughly 1.5 m radius is a plain-geometry estimate: with 40° of steer and the load rollers 1,130 mm behind the steering axis, the turning centre lies about 1.35 m to the side of the load-roller midpoint, and the outer load wheel runs at about 1.6 m. Requirement status is unchanged: 10 met, 2 not met (R5, R15 mass), 2 at risk (R7, R9), 3 not verifiable at TRL 3 (R1, R6, R16); R17 USD 80 over its value-engineering target.

Still open (PLP-DEC-001)

  1. Kit mass of 44.4 kg against 40 kg (A2).
  2. Drive wheel lift of 10 mm (A3).
  3. Classification of follow mode under ISO 3691-4.
  4. Donor jack models and steering yokes to support first.
  5. Named site and co-design partners.

Acceptance of the PLP-DDR-003 Table 1 changes is also still open.

Safety

Unchanged. The rubber stops keep steel parts of the kit off the jack's frame at full turn. Operators used to a bare jack will find the wider turn; the aisle test runs at creep speed in the cordoned test area.

Recommended next step

Amish decides A2, A3 and the Table 1 changes of PLP-DDR-003; then the donor survey (O3), which sets the stop angle, packing bar, clamp and notch. At TRL 4 the aisle test is the first go or no-go gate: if the kit cannot make a right-angle turn into a standard pallet bay, change the drive layout (option c, for example one centre drive wheel under the yoke, reopening D3) before TRL 5. TRL 4 stays on hold.

Session 2026-10-02: open decisions decided

Amish approved every recommendation for the open decisions on 2026-10-02: "i approve your recommendations for all 555 open decisions."

Decisions recorded

Five, all moved to "Decisions made" in PLP-DEC-001 (open items 1 to 5): kit mass of 44.4 kg accepted for the prototype with R15's prototype limit restated as 45 kg and about 1.5 kg of savings carried into the TRL 4 drawings (PLP-DDR-003, A2); 10 mm drive wheel lift for smooth indoor floors only (PLP-DDR-003, A3); follow mode treated as a driverless truck function under ISO 3691-4; a donor survey of three widely sold jacks (a Crown PTH 50 series class model, a distributor's standard model such as Uline's, the Harbor Freight Pittsburgh 5,500 lb jack); Dallas Makerspace and a small food bank or charity warehouse in the Dallas and Fort Worth area as the first candidate partners to approach.

Documents changed

  • docs/06-design-decisions.md PLP-DEC-001 v0.3: decisions made; open decisions section now reads "None"; To confirm item 1 and the value engineering note updated.
  • docs/decisions/0003-design-for-construction.md PLP-DDR-003 v0.3: A2 and A3 accepted (status Draft kept); the Table 1 changes stayed open until Amish accepted them later on 2026-10-02 (see the next session).
  • docs/03-requirements.md PLP-REQ-001 v0.8: R15 restated with a 45 kg prototype limit (now met for the prototype); R7 tied to ISO 3691-4; summary counts.
  • docs/04-calcs/01-sizing.md PLP-CAL-001 v0.6: R15a status and totals; no figures changed.
  • docs/01-problem.md PLP-PRB-001 v0.5 and docs/02-concept.md PLP-PRC-001 v0.8: follow mode classification, donor survey, first candidate partners, R15 and the wheel lift.
  • docs/05-build-plan.md PLP-BLD-001 v0.3: the release is for smooth indoor floors only; safety stop S7 treats follow mode under ISO 3691-4.
  • README.md: Safety note states the ISO 3691-4 treatment of follow mode.

Follow-up actions to carry approved decisions into the design

  1. Decision 1 (calculations): set R15a's prototype limit to 45 kg in docs/04-calcs/sizing.py so that its printed status matches PLP-CAL-001.
  2. Decision 1 (model, drawings): carry the 5 mm top plate, tube lever and hollow pivot pin (about 1.5 kg) into the model and the TRL 4 drawings.
  3. Decision 3 (BOM, calculations): choose a safety-rated personnel detection sensor and define the stopping functions to ISO 3691-4; re-specify the lidar line and rerun the R7 stopping case for it.
  4. Decision 4 (model, drawings): after the donor survey, set the packing bar thickness, clamp holes, notch and steering stop angle for the three surveyed jacks.

Points found in the review

  • The design for construction changes P1 to P10 (DDR-003, Table 1) were still open for Amish's review, with no row in the register for accepting them; the recommendation was to accept. Amish accepted them later on 2026-10-02 (see the next session).
  • R15: the register said "2.0 kg over was accepted for now", but the kit is now 4.4 kg over; the accepted figure was out of date (now superseded by the 45 kg prototype limit).
  • Cost is $1,690 against the $1,610 target, $80 over, all of it from parts added for construction.

Session 2026-10-02: design-for-construction changes accepted

Amish, 2026-10-02: "APPROVED: Design-for-construction changes in 10 repos (CityTwin, CoolShade, PalletPilot, Heliolite, PotholeLog, EarthPress, ReadyKit, CellCheck, CargoMule and ThermaCart)". This accepts the design-for-construction changes P1 to P10 in Table 1 of PLP-DDR-003, which were left open for his review when the open decisions were decided earlier the same day. No other item is decided by it. trl stays 3; no build or test work was done, and the model, BOM, calculations and pictures are unchanged.

Documents changed

  • docs/decisions/0003-design-for-construction.md (PLP-DDR-003 v0.4, status Draft): status line now "accepted" with Amish's words.
  • docs/06-design-decisions.md (PLP-DEC-001 v0.4): Decisions made row added, dated 2026-10-02; the 2026-09-30 row no longer calls the record open for review.
  • docs/05-build-plan.md (PLP-BLD-001 v0.4): section 2 says the Table 1 changes are accepted.
  • PDFs regenerated.

Recommended next step

No change: the follow-up actions of the previous session stand. TRL 4 remains on hold by Amish's instruction.

Session 2026-10-02: approved follow-ups carried out

Amish approved every follow-up action of the open-decision sign-off on 2026-10-02 ("APPROVED CHANGES, COMPLETE THESE"). Four follow-ups were listed above; results:

Approved follow-ups carried out

  1. Calculations, R15a limit: done. docs/04-calcs/sizing.py restates R15a to a 45 kg prototype limit (status Met at 42.9 kg).
  2. Model and drawings, 5 mm top plate, tube lever, hollow pivot: done. The top plate is 5 mm, the lever a 20 x 10 x 2 mm tube, and the single pin is two hollow 20 mm pivot tubes (one per arm; this also frees the middle of the frame for the scanner). Kit mass is 42.9 kg (was 44.4 kg). The enclosure is 1 mm deeper (96 mm) because the plate is thinner. STEP and STL regenerated; constructability checks pass (0 overlaps, 46 contacts, nine new clearance checks).
  3. Safety-rated personnel sensor and ISO 3691-4 stopping: done on paper. BOM line 18 is now a SICK nanoScan3 Core I/O class scanner (USD 3,444.30 listed, plus about USD 11 for a 3 mm shelf; USD 3,455 for the line). Stopping function defined: scanner safety outputs to the safety relay, monitored controlled stop (stop category 1) by the driver, brakes close at standstill; the driver specification now includes safe torque off. R7 case rerun: controlled stop 0.36 to 0.70 m inside a 0.80 m field (delay 0.15 s). Not resolved: if the drive's controlled stop fails and only the brakes act, the 1,500 kg 2 % downgrade case needs 1.15 m, beyond the field; R7 stays At risk. The scan plane is now 173 mm (was 200 mm), the safety edge 68 mm high and the anchors 165 mm tall, all to make the scanner fit.
  4. Donor survey dimensions (packing bar, clamp holes, notch, stop angle): not done. They need the dimensioned drawings of the three surveyed jacks, which are not in the repo; the model keeps the reference donor. This is survey work for the donor jacks, listed under "To confirm when parts are bought" in PLP-DEC-001.

Done 3 of 4.

Key results

  • Requirement changes: none. 11 met, 1 not met (R5), 2 at risk (R7, R9), 3 not verifiable at TRL 3. R15 kit mass 42.9 kg (met for the prototype; 2.9 kg over the 40 kg goal). R12 reserve 24 % (was 28 %, scanner draws about 2 W more). R17: Value-engineering target: USD 1,610. Estimated cost of the constructable design: USD 5,065 (USD 3,455 over the target). budget_usd unchanged.
  • Cost saving worth trying: a cheaper safety scanner that fits in front of the enclosure. The cheaper Hokuyo UAM-05LP-T301 (refurbished USD 1,347.53) has a 143 x 110 x 110 mm body that does not fit as laid out.
  • Assumed until the scanner is bought: its scan plane about 40 mm above its base, 70 ms response, envelope about 107 x 80 x 118 mm, four mounting holes (register item 6).

Documents changed (new versions)

PLP-CAL-001 v0.7, PLP-REQ-001 v0.9, PLP-PRC-001 v0.9, PLP-PRB-001 v0.6, PLP-DEC-001 v0.5, PLP-DDR-003 v0.5, PLP-BLD-001 v0.5, README, bom/bom.csv, bom/bom-notes.md, docs/04-calcs/sizing.py, cad/src/model.py (new clearance checks), cad/src/product_model.py.

Pictures regenerated

General arrangement PLP-DWG-001 Rev P5; concept media (hero, blueprint, cutaway, exploded, flow, viewer; concept sheet Rev P3); all build plan pictures (overview, 11 making sketches PLP-DWG-101 to 111, 10 joints, 15 steps, wiring diagram). Text check on all drawings clean. The earlier note that the pictures were unchanged no longer applies.

Render scenes exported

Appearance model cad/src/product_model.py updated (drive module, release, hoop, scanner and steering stops are now the model's own shapes). python3 .kit/export_views.py /home/claude/renders/palletpilot wrote hero, exploded and detail (.npz and .json each) and palletpilot__jobs.json. Photoreal images, card.png and social-preview.png are for Amish's Mac.

Cross-repo actions

None found for this repo.

Safety concerns

The personnel stop for follow mode is on paper only. Brakes-only worst case (1.15 m) exceeds the 0.80 m field; TRL 4 must size the field from the scanner's own stopping-distance rules before any trial with people present.

Recommended next step

Amish decides whether a USD 3,455 scanner line is acceptable against the value-engineering target or whether to pursue a cheaper PL d scanner that fits; TRL 4 remains on hold.

2026-10-02: photoreal renders redone on the constructable design

Rendered with Blender Cycles on Amish's Mac from the updated appearance model; captioned with .kit/photo_caption.py; media/card.png and media/social-preview.png regenerated with .kit/cards.py. Views: hero, exploded, detail. image_qc passes. Appearance deviations are those logged above as proposed, awaiting Amish.

2026-10-03: decisions recorded

Amish decided on 2026-10-03: "PalletPilot - i accept it" (the PL d safety scanner at USD 3,455; kit about USD 5,065 against the USD 1,610 value-engineering target) and "Cost over target - i accept all the cost variations and overruns".

  • docs/06-design-decisions.md (PLP-DEC-001): row added to decisions made; value engineering section says the overrun was accepted by Amish on 2026-10-03.
  • The 2026-10-02 next step (whether the USD 3,455 scanner line is acceptable) is answered; a cheaper PL d scanner stays a saving worth trying. TRL 4 remains on hold.