vention_assembler v1.1.0

Master-plan

This text file complements our tickets tracking, by providing high-level context, as well as additional notes and details.

Our current sprint is S12. We use milestone:future or floating tickets to park future issues or ideas. We also have a longer-term sprint planning project.

Topmost Priorities (High-Level Context)

Short-Term

• Indexer & flexible robots
• Address issues detected by feature 658
• Fix the light-duty extrusions
• A code-splitting pass
• Collisions
• Connection improvements (e.g. 747 etc.)
• Panel re-generation
• Improved tree etc. ("UI 4.0")
• Performance for large assemblies
• Visual reorg of designs

Longer-Term

• Top-of-funnel / growth

  • better/more parts (500 → 5000); ability to make more machines
  • larger designs (toward whole manufacturing floors)
  • in-CAD automation & simulation (toward a truly all-in-one solution)
  • better usability (simplicity, speed, discoverability, completeness, validation etc.)
  • better pricing and rewards (parts, services, promotions, credits etc.)
  • more active community (purpose, reduced privacy inhibitions, active support, meaningful rewards etc.)
  • more traffic (partnerships, better access, gamification etc.)
  • better-targeted, prettier, faster, more "responsive" website
  • better brand identity and awareness
  • better documentation

• Middle-of-funnel

  • more active/focused/informed/comprehensive tracking of users
  • more material to keep users on the platform: news, tips, tutorials, blogs etc.
  • better user engagement and community management
  • bring design requests to fully advertized state (+ missing features such as inviting designers)
  • better company features/privileges
  • improved referrals

• Bottom-of-funnel / efficiency

  • "Warehouse OS" → multiple warehouses → reduced shipping prices etc.
  • better integration with other systems or suppliers (orders, inventory, shipping, payment, suppliers, panels, CRM, RFIDs, accounting, assembly services etc.)
  • better internal systems (more comprehensive predictive assessments, live inventory status, easier accounting, better/faster access to data etc.)
  • more robust and comprehensive solutions for mechanical designs and automation

Next Priorities (Assembler)

Once the topmost priorities are addressed, according to the advertized priorities below, and following the general intent to prioritize sales-driving learning experiments first, the next big candidates would be:

UX and Fluidity (mostly top-of-funnel impact)

• Bulk performance optimizations + 1046, using mt and/or gpu, especially to improve the speed and fluidity of connections
• UI "4.0" (912), improved part selector
• Connections improvements (#29 etc.)
• Collision detection to reduce the number of irrelevant connection suggestions
• Island connections improvements (e.g. 794, 817 etc.)
• Cartesian experience for tooling plates
• Ball-screw LM cassette
• Improved tutorials
• Marketplace/collaboration improvements

UX for Advanced Users (toward bottom-of-funnel)

• Design validation (1167, 1169 etc.)
• Center of gravity
• Allow to label versions
• Allow to diff versions
• PPT-like procedural 3D alignment tools
• Support more complex contours in panels (gussets, plates etc.); improve performance
• Sub-assemblies
• Support larger assemblies (-> whole floor)
• Multiple connections and types on imported parts
• Further improve R->T
• Richer tools for custom connectors
• Custom connectors on Vention parts - ? (maybe no longer relevant, with group/ungroup)
• Misc issues: #281, #277, #131 etc.
• Prevent impossible connections on two sides of the same hole (double-sided plates)
• Import/export improvements
  • Bridges to external systems, e.g. #167
  • Purge
  • Delete/undelete
  • Misc issues: #180 etc.
• Recycle your design

Motion, Mechanisms, Simulation, In-CAD Automation

• Support more RDOF-capable mechanisms
• Toward full n-bar support:
  • locus analysis
  • macro-geom reduction
  • driving-input constraint
  • loop classification
  • solve rigid chains
  • solve loops
  • final position analysis algo
  • support the complete fragment table (all required joint types, all DOF)
• Control-apps editor (#72, #73, #74, #75 etc.)

Parts Ingestion and Internal Processes

• Improve our internal tools for generating connections (UI, R->T vs T->R etc.)
• Libraries of partners' products
• Upgrade to latest HOOPS version
• Finish light-duty extrusions and heavy-duty extrusions
• re-import imported parts from AWS (related to 437 etc.)
• Textured parts + #220: sensors or ball screws with threads, controllers with too many holes etc.

Systems Integration and Systems Health

• Redis integration (basic first pass)
• Better integration of panels (e.g. 2D contour format, automatic dimensions etc.)
• Find and use solutions for server-side rendering of thumbnails (e.g. assembly instructions, design thumbnails, visual time machine etc.)
• Assembler on landing page (?) - vs scalability aspects
• Integration with 3D-printing supplier (e.g. protolab, xometry etc.)
• Integration with back-end services (inventory tracking etc.)
• Scaling of HOOPS servers (models, import, export)
• stale assets + 1177 + 1151
• GA/adwords issues: 1413
• Three.js replacement (?)

Research

• Improved parts suggestions and compatibility features (e.g. based on pre-computed stats from users)
• Usage of machine-learning to assist design creation and improve ranking of connection suggestions
• Improved search based on classification of sub-assemblies
• Motion planning for in-CAD automation
• integration with video analytics
• integration with VR

Related to Inventory and Sales

• VOPS: no sub-assemblies in supplier orders: 68
• Assembly Instructions 2.0 / better annotations

Advertized Priorities

Etienne maintains another view on the same reality, that he's advertizing through recurrent update mails sent to our beta users, investors etc.:

Other Areas (Summary)

vops

• improved access to data (better pages, open queries, batch updates)
• performance improvements (pagination, search etc.)
• better integration with external systems (orders, inventory, shipping, payment, suppliers, panels, CRM, RFIDs, accounting, assembly services etc.)
• more comprehensive supplier order management
• better management of subassembly-parts
• better panel integration
• "live" inventory status
• improved/optimized shipping
• improved predictive heuristics (qty, time, early-vs-late-state regressions, etc.)
• simpler implementation

rails

• continue work to "crack" community
• support better the most used browsers
• improved site speed
• improved queries
• CDN, asset pipeline, faster delivery of assets
• database caching for elements in high demand
• better scalability (e.g. unicorn vs puma etc.)
• brand identity; continue css refactoring
• better "responsive" support (mobile/tablet)
• more thorough tracking of SEO
• more comprehensive security/privacy enforcements
• infrastructure for faster collaboration using push notifications
• CMS improvements
• linkedin signup
• better hubspot integration
• partners/resellers/CSI (#1664 etc.)
• my profile/company wall (latest news, designs, colleagues' work etc.)
• mouseflow++
• better design version support (e.g. buy specific version etc.)
• UPS/Fedex/shipping improvements (or abstraction)
• option to ship assembled
• internationalization for Europe/Asia
• deployments for Europe/Asia
• users forum (esp. automation)
• ag/imported-parts delete/undelete/purge
• cart transfer

MachineMotion

• evolution toward 1.5 and 2.0 physical designs
• 3 levels of API; support for multiple languages; evolving features and mechanisms
• integration with the CAD (in-CAD automation/simulation)
• improvement of the motion control program on the micro-controller
• io expander and simplification of physical mappings
• integration with other automation/robotic systems (protocols, APIs, plug-ins etc.)
• documentation

Additional Issues

Here I intend to enumerate issues or thoughts that came up, that may be important, but that we don't yet understand well enough to categorize right away as tickets of a milestone:

• Etienne's design 326 had a severe reload issue; design 516 too; design 522-rev4 too; Francois's design 554-rev8.9.7 too
• with recent work on RDOF, I noticed errors when attaching parts to sections that had been rotated around a hinge; as a quick workaround I augmented the tolerance threshold in ilib_tm (from 1e-6 to 2e-6), but I need to get to the bottom of this
• I noticed failures in full-connect, with round extrusions
• rectangle-selection of some components (rollers, wheels and other compound parts) doesn't work
• need to finish supporting some new precision plates etc.
• I'm still seeing strange position locations (insert ball overlays) once in a while
• full-connect after simple move?
• cleanup redundant tla usage (mentioned below also in the context of load)
• simplify load code (factor out the tla stuff, copied 3 times at the moment)
• with 90x45, symmetry doesn't seem to remove some apparently identical options -> investigate
• measure perf of the full scan in computeAllBoltPositions/annotateWithMatchingBolts, for a larger scene
• rename grabLocalMatrices and local_matrix?
• a cons-ing cleanup pass, when time permits
• strong, automatated, offline 1mm spacer checking
• strong STEP check vs mdb.js
• migrate mdb to a more orderly and/or external solution?
• run a web security test suite (applicable to our whole site)
• issue noticed by E, whereby in perspective mode, moving a sub-graph around was strangely offset away from the mouse position