Evidence coverage and known limits

The analysis can only be as good as what URC and the teams publish, and that record is uneven. This appendix states, topic by topic, where the evidence is strong enough to design against and where it is thin enough that a measurement has to replace a citation. It also sets out what the 2027 rules leave unpublished, because those omissions decide the order of the team’s own test work.

What the record covers well

The rules are the strongest part of the record. The 2027 rulebook was published in September 2026 and the Q&A was current to 11 October 2026, and both can be read against the archived copies S-URC27-RULES S-URC27-QA. Every hard limit, mission task and deadline in this book traces to a clause, and the 155 requirements give each trace a line with the check that confirms it S-URC27-REQ. The Q&A is a living document that changed twice in its first month, so a ruling carries the date it was read, and the same questions have to be re-read before the PDR on 2 December 2026 and again before the SAR on 5 March 2027 S-URC27-QA S-URC27-RULES §3.a.ii–iii.

The results are nearly complete. The five official score sheets for 2022 to 2026 hold 187 team-years, 162 of them with finals scores, and every mean, zero rate and variance share in the statistics part is computed from them rather than estimated S-URC-RESULTS S-CENSUS. Two limits survive even here: a rank is not a column in a score sheet but a derivation from row order, and a team with a review score and no mission scores has no rank at all.

Team architecture is the third strong area, and its depth depends on the team. The 51 profiles are built from published reports, repositories, review films and papers, and the Tier A pages can follow a suspension or an arm down to a part S-CENSUS. The tier is a documentation test as well as a form guide: Tier A holds eight teams with a podium finish or an exceptional public record, Tier B holds 24 with a consistent finals presence, and Tier C holds the remaining 19 profiles, most of which rest on a website and a result.

Where the evidence thins out

Mass, centre of gravity and energy

Mass is the number that most constrains a first build and the number the field publishes least. The coverage counts are the finding.

Across the 17 configurations the minimum is 33.0 kg, the maximum 53.0 kg, the mean 47.9 kg and the median 49.1 kg, and four of them sit within 0.2 kg of the limit, one at 49.8 kg and three at exactly 50.0 kg. Those numbers size a mass budget well. They do not predict a weigh-in, because the one measurement in the record comes from a 2021 virtual event, and the rest are the teams’ own statements or deductions from them.

Task objects, forces and the course

The servicing tasks are described in words and scored on completion, but URC publishes no mass, dimension or force for the objects involved: the sample tube, the cache, the key and lock, the hose fitting, the valve, the buttons, switches and knobs, the hammer, or the astronaut’s suit S-URC27-RULES S-URC27-QA §1.d.i–ii. A gripper torque budget, a lift rating and a typing force therefore rest on a mock that the team builds, not on a published value, and every such target in this book says so.

The course is described by terrain type rather than by numbers. The Q&A gives no design value for slope or drop: inclinations vary by route and are not provided S-URC27-QA S-URC27-QA AUQ17 D1. Sign-reading distance and camera resolution are left to performance rather than stated as distances. The response in this book is a self-chosen obstacle set, measured against a local rock pile and the previous season’s course film, and the result is a design target rather than a rule.

The 2027 rubrics and weights

Three 2027 documents are promised and not yet available. The SAR page says the 2027 requirements will be posted in early 2027 S-URC27-SAR, the Science Plan page says the same S-URC27-SCIENCEPLAN, and the Delivery course description is promised before the competition rather than with the rulebook S-URC27-RULES §1.c.i.

The weights are the second half of the problem. Each mission carries 100 points, but per-task values are published only for the Autonomy Mission S-URC27-RULES §1.a.i §1.e.iv–xviii. Astrobiology names four components without weights S-URC27-RULES §1.b.xi, and Equipment Servicing scores each task without publishing its value S-URC27-RULES §1.d.i–ii. A comparison between two sub-tasks in those three missions is therefore an ordering, not a price.

Where the book needs a price it uses the 2026 rubric, which is the most recent published one, and labels it as a proxy S-URC26-SAR. That rubric is a floor rather than a guarantee: 2027 adds an astronaut-assistance autonomy mission, a fine-manipulation key task and a display interface, and none of them is priced in the 2026 document.

Delivery and drones: one recorded scope limit

Delivery is the only subsystem without a dedicated study of its own, and the reason lies in the record rather than the mission. No comparable published account of the 2027 Delivery course, its stage tasks or its drone use exists, so the chapter is a synthesis from the Delivery rules S-URC27-RULES §1.c, the drone rules S-URC27-RULES §2.b.v–vii, the Delivery Q&A and the arm, astronaut and communications material. Container and stowage detail is thinner than in the other eleven chapters, and there is no delivery-specific end-effector comparison table.

Two rule questions in this area remain open and each one changes a design: whether a drone counts toward the 70 kg fielded total, and whether the Equipment Servicing cache is the team’s own astrobiology cache S-URC27-RULES §2.b.vii. Neither is settled by the rulebook, and both are worth a written answer from the judges before the preliminary design review.

From a gap to a measurement

A gap is only useful if it names the test that closes it. The 77 open questions collect the ones that change a design decision, grouped by what would settle each.

The 110 engineering problems carry the same gaps into the build plan, each with an acceptance criterion and a place in the dependency order. A handful of them close in a week and turn several rows on this page from a limit into a number: a weigh-in of the completed rover, one logged discharge under a measured load, a lock torque on a matching cylinder, and one climb test on a 30° slope of known soil.

What stays open after the measurement

• Rank is a derivation. A score sheet is sorted, not ranked, so a one-place difference between two teams is not a published fact.
• A photograph is not a datasheet. A motor, gearbox or camera identified from a review film is recorded as an inference, with the visual basis stated, until a document confirms the part.
• A winning design does not transfer. Where this book describes what worked for a champion, it describes that team’s choice and the result it produced, not a requirement for anyone else.
• The commanded astronaut interaction is new. Follow, Stay, Fetch, Come and Give, with a beckoning gesture, have no predecessor task, so that part rests on the rules, the Q&A and first principles rather than on observed practice.

Further reading

• About the evidence sets out the four grades, the citation form and the way a number is read.
• Team dossier index lists the 51 profiles; the census holds all 68 teams and the tier rule.
• Mass and power data carries the 103 data rows behind the mass figures quoted here.
• Mass and power shows what the model concludes and where it is sensitive.
• 77 open questions turns each limit into a test, a measurement or a question for the judges.
• Delivery, payload handling and drones is the chapter with the recorded scope limit.