110 engineering problems

This is the season’s engineering register: 110 numbered problems, each with the requirement that creates it, the experiment that produces the missing number, and the criterion that closes it. Every problem traces to at least one 2027 requirement, and all 155 requirements are covered by at least one row. The rows are ordered by dependency wave rather than by subsystem, because the order in which the work becomes possible is not the order in which subsystems are usually written down.

What a row contains

Priority is deliberately coarse. 29 rows are Critical: open at their gate, with the season at risk through a pass-or-fail rule, a milestone, or a measurement that three waves depend on. 72 are High: they block three or more later problems or a whole mission capability. 8 are Medium and 1 is Low, recoverable later without a rule breach.

The fine ordering is the unlock count, and the two are meant to be read together. A High row with eight dependents is better value than a Critical row with none, which is the case for a May submission deadline.

The five waves

The five waves sum to 110. The PDR is due 2026-12-02 S-URC27-RULES §3.a.ii, the SAR 2027-03-05 S-URC27-RULES §3.a.iii, and the finals run 2027-06-02 to 06-05 S-URC27-RULES §3.a.iv. The communications documentation is due 2027-04-30, in the last week of W3, so its substance has to be finished before that week opens.

All 110 problems

The table is searchable and sortable, and it filters by wave and by priority. Dependency counts are in the Unlocks column; a row with a high count is a row that frees design work elsewhere.

The dependency spine

110 problems connect through 253 dependency edges. Thirteen rows have no prerequisite, because they are pure measurements, and twenty-one are leaves, because they are submissions, rehearsals or drills. Thirteen rows carry five or more dependents, and they are the rows that decide when the rest becomes possible.

Six of the thirteen rows are in the measurement wave. That is the register’s central claim about a first build: the numbers that size the machine are not published, so the early season is spent producing them.

Working the register every week

A register that is read once is a document. One that is reviewed weekly is a management system, and the minimum useful ritual is 45 minutes, five questions, in this order.

1. What closed this week, and what artefact proves it? A row closes only when its success criterion is met and the record is filed where the SAR evidence can be found. A measurement either exists or the row is still open.
2. What is now unblocked? Every closed row carries the number of later problems it unlocks. Those are the candidates for next week, and they are the only rows that should be scheduled without an argument.
3. Which Critical row is nearest its gate? The PDR gate set and the SAR gate set are fixed, so anything in them that cannot close in the remaining weeks is a scope decision to take now.
4. What did a measurement contradict? A number that disagrees with a subsystem study is a finding: record it in the row and in the study, and re-run the rows that used the old number.
5. Which question now blocks more than one row? If a written question has grown dependents, escalate the email rather than guessing in two places.

A worked row

The wheel-diameter row, W1-03, shows what a well-formed entry looks like. The problem is stated as a decision rather than a preference: fix wheel diameter, wheelbase and track. It depends on four measurements: the mass frame (W0-01), traction (W0-09), the step rig (W0-10) and the arm mass class (W1-01). Its first experiment is a packaging study of 250, 280 and 330 mm wheels inside the cube gauge, plus a tip-angle calculation at each diameter. Its success criterion is a diameter that fits the cube with 20 mm per side, keeps the static tip angle at or above 55° as a derived target, and has a bearing and shaft design at the measured torque. The artefact is a signed envelope drawing, the packaging calculation and a photograph in the gauge. It unlocks seven rows: gear ratio, suspension, mast height, chassis, the mass and centre-of-gravity document, kinematics, and the structural load cases.

That shape is the test for every row: a decision, its inputs, one experiment, one number, one artefact, and the list of what it frees. A row with no number in its success criterion is a topic, not a task.

Coverage of the 155 requirements

Every one of the 155 requirements is traced by at least one problem: 155 cited, zero uncovered, zero unknown identifiers. The distribution is itself a check on the analysis.

The drone is the thinnest area, at eleven requirements carried by four problems and one of them Low priority, which matches the absence of any strong drone evidence in the record. Safety has eight requirements and eight problems, three of which are tests, the right ratio for a pass-or-fail clause.

Twenty-one requirements are carried by exactly one row, and twelve of those are mandatory. Two are worth watching, because a single scheduled item is load-bearing for something scored. The panorama and close-up requirements, REQ-SCI-007 and REQ-SCI-008, both rest on the Astrobiology rehearsal W3-08, so a documentation-only dry run is added earlier. REQ-SAF-004, that every mobile deployable carries its own E-stop, rests on W2-24 and is correct only while nothing is deployed.

Twenty questions the register gates

Twenty questions in the register are the ones no amount of engineering can settle. Each is a written question to the URC Director, each gates named rows, each has an answer date, and each has a conservative fallback so that an unanswered question never stops work. The four with the largest design consequences are below; the open questions page carries the wider set of 77.

Each question goes as one numbered email that quotes the clause, states the interpretation the team intends, and asks for confirmation or correction, with a request that the answer be mirrored onto the public Q&A page, which is itself a binding rule source. The reply is recorded against the rows it gates.

What would make this register wrong

• The 2027 numbers move. The 50 kg, 70 kg and 1.2 m limits, the five milestone dates and the mission structure are confirmed for 2027. A rules revision would force every row that cites the mass, the timeline or the mission structure to be re-read before the next review.
• A task weight changes the order, not the content. The Astrobiology, Delivery and Equipment Servicing weights are unpublished. When they arrive, the sequence inside the last two waves changes, because the highest-points task is attacked first; the problems themselves do not.
• A measurement can overrule a study. Every value carried over from a subsystem study is an estimate until the measuring row closes. Where a measurement disagrees, the measurement wins and the row records the difference.
• The plan assumes one rover. Two mission configurations with a real payload change would grow W1-19 and W2-05 and turn the 10-minute transition into a rehearsal item with its own test.

Further reading

• 70 risks points every risk at a problem row, so a risk with no row is a missing row here.
• 77 open questions carries the full question set and the meanwhile assumption for each.
• The first 90 days reads these dependencies forwards, week by week.
• 155 engineering requirements is the table every row traces into.
• Coupling, conflicts and twelve trade studies settles the design decisions the architecture wave tests.
• Interfaces and couplings supplies the interface rows the detailed-design documents must contain.