Mapping Competition Requirements $\rightarrow$ System Architecture $\rightarrow$ 3-Week Lookahead Execution
For UBC Rover Operations & Subteam Leads
Target: University Rover Challenge (URC) 2027 & Canadian International Rover Challenge (CIRC) 2027
Traditional engineering project management tools (Gantt charts, basic Excel trackers, or flat task boards) always break down in student robotics teams. Tasks get assigned before prerequisites exist, leads get blocked by silent upstream delays, and nobody knows why a milestone slipped until it's too late.
This project implements an adapted version of the Last Planner System (LPS)—a lean production management workflow originally developed in Lean Construction—customized specifically for the realities of UBC Rover.
- The "Critical Web" vs. Linear Gantt Charts: Engineering a Mars rover isn't a straight line. It is an interconnected web where a single physical subsystem (like the Chassis Baseplate) has 10+ subteams and modules feeding into it, and capabilities (like "Stable while drilling") cannot be solved until upstream tools (like the Rover Lab sampling tool) have settled their reaction forces and mounting geometry.
- Busy Student Volunteer Bandwidth: Leads should never be handed a blank template or forced to do complex dependency mapping in huge spreadsheets with 120+ competition requirements.
- The Go-Ahead Checklist (Go / No-Go Decision):
- A common mistake in project management is requiring "CAD Frozen" before starting a task, when a huge portion of the team's work is design itself!
- In our system, the Soundness check is a Go / No-Go Checklist needed to actually start making, purchasing, or committing to execution:
- Upstream / Interface Defined: The upstream task is done OR its physical/electrical interface is defined (we know what it connects to / looks like; physical prototypes do not have to be done to take on a downstream task).
- Design & Scope Defined: Clear concept, dimensions, and understanding of what we are creating.
- Exact Parts & Budget Specified: Bill of materials and exact parts specified for budget approval and purchasing.
- Tooling & Equipment Ready: We know which tools, machines, or 3D printers we need first and have confirmed access.
- When all 4 checks pass, the task is Go-Ahead Ready (Sound for the 3-week rolling lookahead). If any check fails, it is marked No-Go / Blocked, prompting for an explanation and a standardized Reason for Variance (RFV 1–6) code.
Everything in our rover connects in a closed loop from competition points down to weekly work:
┌────────────────────────────────────────────────────────┐
│ URC 2027 Competition Rules │
│ (ASTRO-##, DELIV-##, EQUIP-##, AUTO-##) │
│ Scored capabilities rated: YES / TOUGH / NO │
└──────────────────────────┬─────────────────────────────┘
│ fulfills
▼
┌────────────────────────────────────────────────────────┐
│ Multi-Level System Architecture │
│ (draw.io System Block Diagrams) │
│ Competition -> Rover -> Chassis/Arm/Compute... │
└──────────────────────────┬─────────────────────────────┘
│ implemented by
▼
┌────────────────────────────────────────────────────────┐
│ Subteam Deliverables (TSK-###) │
│ 120 Atomic Subteam Engineering Tasks │
└──────────────────────────┬─────────────────────────────┘
│ screened by
▼
┌────────────────────────────────────────────────────────┐
│ LPS 3-Week Rolling Lookahead & Constraints │
│ [CAD Frozen?] [Parts Here?] [Tools Ready?] [Handoff?]│
│ Sound -> Committed in Weekly Plan │
│ Blocked -> Tagged with Variance Code (RFV) │
└────────────────────────────────────────────────────────┘
We are executing this rollout in strict phase gates with our AI assistant:
-
STEP 1: Define Setup Plan & Architecture (Completed ✅)
- Established the relational schema linking Requirements
$\rightarrow$ Architecture Nodes$\rightarrow$ Atomic Tasks. - Extracted and saved the team context, anchor dates, subteam lead ownerships, and tie-breaker protocols.
- Established the relational schema linking Requirements
-
STEP 2: Bulk Ingestion & Intermediary Triage Micro-App (CURRENT PHASE 🔄)
- Parsed raw input files (
system_architecture.xml,subteam_requirements.xlsx,budget_and_schedule.xlsx, andurc_rules_2027.pdf). - Built this disposable local micro-app (
app.py) so Andres, Rowan, and the Leads can rapidly link multi-input/output dependencies and screen constraints via UI buttons instead of walls of text chat or fragile spreadsheets.
- Parsed raw input files (
-
STEP 3: Production Deployment & Critical Web Engine (Next Phase 🚀)
- Feed the triaged dataset (
lps_data.json) back to the AI agent to compute the full Critical Web, generate automated dependency layers, and deploy the final production tracking system (integrated with MS Planner / GitHub or a lightweight cloud dashboard).
- Feed the triaged dataset (
Rowan: "On first look it looks really busy."
Andres: "It is, this is an intermediary tool... just to define and interlink all of our requirements. I think all the links will be difficult in Excel."
This micro-app is not the final weekly dashboard for the general team. It is a specialized workbench designed for Operations (Andres, Rowan, Noah, Anika, Ben) to sit down and rapidly interlink the messy baseline data.
- Multi-Input & Multi-Output Linking:
- Architecture Nodes: Select 1 Primary node, and click
+ Add Nodeto attach secondary interfacing nodes. - Competition Requirements: Attach multiple URC point-scoring capabilities (
ASTRO-##,DELIV-##, etc.) to a single task. - Upstream Prerequisites: Select which predecessor tasks block this work from starting.
- Downstream Handoffs: Designate which subteams receive the output when complete.
- Architecture Nodes: Select 1 Primary node, and click
- The Go-Ahead Checklist (Go / No-Go Decision):
- Before any task is considered "Ready" for the 3-week lookahead, it must pass 4 gates:
1. Upstream / Interface Defined(Know what it connects to / looks like; physical prototype not required)2. Design & Scope Defined(Clear concept & dimensions of what we are creating)3. Exact Parts & Budget Specified(BOM & part costs specified for budget & purchasing)4. Tooling & Equipment Ready(Know which tools/machines/printers are needed & have them)
- Before any task is considered "Ready" for the 3-week lookahead, it must pass 4 gates:
- Reason for Variance (RFV) Logging:
- If a task is blocked or misses its commitment, leads tag it with one of the team's standardized variance codes:
RFV-1: Upstream Handoff DelayRFV-2: Parts / Machine / Printer BottleneckRFV-3: Scope / Integration SurpriseRFV-4: Academic Crunch (Midterms/Finals)RFV-5: Overcommitment / Bandwidth ExceededRFV-6: Funding / Budget Hold
- If a task is blocked or misses its commitment, leads tag it with one of the team's standardized variance codes:
- Tie-Breaker Rule:
- Noah (Systems Lead) breaks cross-subteam integration and dependency ties, with final sign-off by Captains Andres & Rowan.
The micro-app is completely self-contained in a single lightweight Python script with an embedded UI. No complex web servers or Node.js required.
- Python 3.8+ installed on your machine.
- Open your terminal in this repository folder:
# 1. Install Flask (one-time) pip install -r requirements.txt - Start the micro-app:
# 2. Run the server python app.py - The app will automatically open your browser to:
http://localhost:5000 - Click through your subteam tab, link prerequisites/nodes, toggle constraints, and click Save Changes.
- All updates write directly to
lps_data.json. Commit and push your changes to GitHub!
ubc-rover-lps/
├── app.py # The Step 2 Flask triage micro-app & embedded UI
├── lps_data.json # Master working dataset (Tasks, Nodes, Reqs, Constraints)
├── requirements.txt # Dependencies ('flask')
├── README.md # This document
│
├── raw_inputs/ # Source of truth raw files from the team
│ ├── system_architecture.xml # 10-page draw.io system block diagram export
│ ├── subteam_requirements.xlsx # Anika's URC capabilities sheet (Yes/Tough/No)
│ ├── budget_and_schedule.xlsx # PATH funding budget & legacy preliminary schedule
│ ├── urc_rules_2027.pdf # Official URC 2027 competition rulebook
│ └── brain_dump_2026-11-15.md # Andres's raw operations context & lead notes
│
├── parsed_data/ # Parsed intermediate JSON exports from AI extraction
│ ├── architecture_raw.json # 86 nodes, 106 edges, 10 pages
│ ├── requirements_raw.json # 109 scored requirements, 90 overview tasks
│ ├── budget_schedule_raw.json # 137 budget items, 90 schedule items
│ └── urc_rules_raw.json # 4 missions, 17 general rules, 9 key dates
│
└── GEMINI.md # Agent constitution & strict phase-gate protocol
TSK-###— Atomic Subteam Engineering Deliverables (e.g.TSK-012: [Equipment Servicing] arm reach covers ground to 1.5m).ASTRO-##/DELIV-##/EQUIP-##/AUTO-##— Specific URC Competition scoring rules from the official rulebook.- Architecture Nodes — Blocks from the draw.io system diagram (e.g.
Rover,Chassis,PTZ,Compute,Death Ray). - Subteams:
ARM: Arm & End Effector (Leads: Will, Matt)CHAS: Chassis (Lead: Eugene)ELEC: Electrical (Official Lead: Darwyn; Acting Lead: Rowan)RLAB: Rover Lab (Lead: Jennifer)SCI: Science (Lead: Jennifer)SW: Software (Leads: Riddhima, Ben, Rowan)OPS: Operations (Andres, Rowan, Anika, Noah, Ben)BIZ: Business (Lead: Angelica)
Once Step 2 triage is complete, the AI agent will compile the linked data into our Step 3 Production System.
In our weekly leads meetings:
- No Finger-Pointing: If a task was missed, we don't argue—we log its RFV Code so Operations can kill the root bottleneck (e.g. order more 3D printer filament, adjust for midterm schedules, or fix cross-team handoffs).
- Rolling 3-Week Commitment: Each subteam lead commits only to tasks in the 3-week lookahead that are 100% constraint-free (sound).
- Visible Critical Web: When a task on the critical path is checked off, its dependent branches automatically unlock across the rover, giving everyone total clarity on what needs to be built next.