Skip to content
Mission backfit·Chapter 6.3

Equipment servicing

Equipment Servicing is the arm’s report card, and the field’s weakest mission after Autonomy. Nine task families span ground level to 1.5 m, and the smallest targets in the rulebook are 2 × 2 cm ArUco tags at the corners of a vertical keyboard and 1 × 1 cm tags on a key and a keyhole. The difficulty is manipulation itself, not the task list.

100Mission pointsNine task families in 30 minutes, in any order2027S-URC27-RULES
1.5 mWorkspace heightFrom ground level, at an unspecified standoff2027S-URC27-RULES
19.05 mmKeyboard key pitchAbout 0.44 N over 2 mm, with no auto-repeat2027S-URC27-TASKHW
29–35The field’s mean for four seasonsFlat despite four years of rule changes2023–2026S-URC22-SCORES

Key takeaways

  • The flat mean is the finding. The mission mean has stayed between 29 and 35 for four seasons despite rule changes, because the difficulty is manipulation itself rather than the task list.
  • Open-loop precision does not exist at this scale. Planetary joint backlash of 15–18 arcmin is 4.4–5.2 mm at 1 m, wider than a keyway. Vision has to close the loop on typing and on the key.
  • 2027 adds two hard tasks and removes one. The key in the tubular lock has no precedent, the hose is stiffer, and the USB Type-C task is gone. Per-task weights are unpublished in both years, so the five-year mean is the base rate.
  • A mid-cost arm reaches the middle of the field. UIU’s 72.63 came from a 6-DoF arm, an ArUco-aligned typing routine and a rehearsed mock lander S-UIU-2026-SAR.

What the mission physically demands

Demand Number Basis
Workspace Ground level to 1.5 m above ground, from an unspecified standoff; a 100 m traverse; 30 minutes S-URC27-RULES §1.d.i
Task families Nine, in any order: sample tube, cache transport, drawer, latch and panel, typing, key and lock, hose, quarter-turn valve, buttons and switches S-URC27-RULES §1.d.ii
Keyboard Redragon K552, 87-key tenkeyless, 354 × 123 × 37 mm, 875 g, 19.05 mm key pitch, about 0.44 N actuation over 2 mm of roughly 4 mm travel, no auto-repeat S-URC27-TASKHW
Keyboard tags 2 × 2 cm ArUco 4x4_50 markers at the four corners S-URC27-RULES §1.d.ii
Key and lock A flat key with a 1 × 1 cm tag on the bow, a standard 7-pin tubular keyway with 1 × 1 cm tags at the four corners of the face, and a 1–2 mm keyway to align to S-URC27-RULES §1.d.ii
Accuracy budget implied Backlash of 15–18 arcmin is 4.4–5.2 mm at 1 m Datasheet values, by inference
Time per task implied 30 minutes minus a 100 m traverse and nine families leaves about 2–3 minutes each, including approach and retreat By inference; Monash 2022 spent about 4 minutes on the door latch alone across two attempts S-MONASH-ES22
Contact rule Every task must be doable with one gloved hand, and the keyboard must not be touched by hand S-URC27-QA S-URC27-RULES §1.d.ii EQ1
The 2027 keyboard with ArUco markers at the four corners
The keyboard is a standard 354 × 123 × 37 mm tenkeyless unit with 19.05 mm pitch. The tags at the corners give the arm a visual reference, but the press still has to land on a key that is 19 mm wide S-URC27-TASKHW.The Mars Society / URC, 2026. Competition rulebook; reproduced for reference. Original source Full resolution
§1.d.ii." aria-label="Open full-size figure">Two keys carrying 1 × 1 cm ArUco markers
The key task is the smallest target in the rulebook: a flat key, a 1 × 1 cm tag on the bow and a keyway to align to. Removal, transfer between two locks and a turn are three separate failure points in one task S-URC27-RULES §1.d.ii.The Mars Society / URC, 2026. Competition rulebook; reproduced for reference. Original source Full resolution

What the best teams did

The record is consistent: a long tail of zeros, a handful of 93–100 runs, and a mean that does not move.

Year Teams Mean Median Exact 0 Scores ≥80 Maximum
2022 26 21.8 19.5 6 0 58 (Missouri S&T and IMPULS)
2023 31 35.3 35.0 7 3 93 (AGH)
2024 34 32.0 26.3 5 3 95 (BYU)
2025 35 31.5 27.0 4 1 93 (Missouri S&T)
2026 36 29.1 25.0 5 1 100 (Missouri S&T)

Derived from the official mission column in the five result sheets S-URC22-SCORES S-URC23-SCORES S-URC24-SCORES S-URC25-SCORES S-URC26-SCORES.

Equipment Servicing scores 2022–2026 with mean and zero counts
Five seasons of Equipment Servicing scores. The top of the distribution moves between teams, but the mean and the zero count do not, which is what a mission limited by manipulation quality looks like.Source material, 2026. S-URC22-RESULTS S-URC26-RESULTS Full resolution
Team 2022 2023 2024 2025 2026 Architecture and lesson
Missouri S&T 58 68 94 93 100 A hybrid of one prismatic and five rotary axes, mounted on the top deck and chosen for precision on the panel face
BYU 18 58 95 59.2 22.11 A wrist sensor head with ViSP visual servoing; bimodal, 95 then 22
AGH 16 93 74 did not enter did not enter 93 in 2023 was the field maximum
West Virginia 25.2 85 84 43 38.95 An over-torque cascade in 2023 and a damaged USB-C cable in 2026 that took out cameras and two motors
Monash 53 82 18 54 58.95 Every arm redesign scored worse than the design it replaced, 2023 excepted
UIU 23 50 72 42.4 72.63 A 6-DoF arm, ArUco typing and a rehearsed mock lander: the most replicable result in the record
MIST 2021: 100 29 30 41 23.68 The only complete judged run on video, from 2021; hobby servos once the tasks hardened

Official Equipment Servicing column S-URC22-SCORES S-URC26-SCORES, with MIST’s 2021 result from its own competition record S-MIST-VURC2021.

The per-task pattern. Tasks where the object comes to the end effector are routine: buttons, switches, the drawer handle and the latch release. Tasks where the end effector must arrive at a precise pose in free space fail: typing, key insertion, the hose push and re-opening a panel. 2027 adds two of the second kind.

What 2027 changes

2027 change Verdict
Key in a tubular pin-tumbler lock added No precedent. 1 cm tags, a 1–2 mm keyway, and insertion and turn torque not yet measured
USB Type-C task removed Invalidated. Fixtures, poses and rehearsal built for USB are obsolete, and one bankable task disappears
Hose: GatorLock with no gasket becomes a 1.5-inch cam lock with a half-thickness gasket Invalidated hardware. 2026’s fitting was the easier push, so any 2026 tuning is to the wrong part
Display formally linked, at 4.2 inches on e-paper while the Q&A says 4.3 inches Raised bar. A no-touch zone, the launch key read from the display, and judges who reset it
Buttons, switches, knobs, quarter-turn valve, latch, drawer, tube and cache Carry forward
Per-task weights Not published in either year, so no pricing of tasks is possible
§1.d.ii." aria-label="Open full-size figure">The panel-mounted tubular lock with 1 × 1 cm ArUco markers
The 7-pin tubular keyway at the centre of the four alignment tags. The arm has to hold a key on a target this small while applying turning torque, which is why the task carries the highest uncertainty in the competition S-URC27-RULES §1.d.ii.The Mars Society / URC, 2026. Competition rulebook; reproduced for reference. Original source Full resolution

The net effect is not simply “harder”. Two hard tasks were added, one was removed, and the weights are unpublished. Anyone who assumes the 2026 mean carries over exactly is guessing in both directions.

Which subsystem options serve it

Decision Option A applies when Option B applies when
Arm architecture Five rotary axes plus one prismatic axis, with the rail parallel to the lander face, when Equipment is the priority: the best scores in the field use this layout Six rotary axes for the least-surprising path with in-house machining; five axes plus a 400 mm lateral stage when the arm would otherwise be too long to stow
Rail caution A prismatic axis must be designed for driving shock: Missouri S&T’s own vertical axis failed under driving in 2024 S-MST-NEWS2024 A pure rotary arm if the rail cannot be made shock-proof in one season
Actuators Integrated brushless modules with CAN and torque mode, to buy past driver development Harmonic or cycloidal drives at the wrist when typing or the key task must work open-loop
End effector A two-finger 100–110 mm jaw with a compliant typing tip, a chamfered key funnel and a hook or pin A turret of several end effectors, never on a first build
Perception A wrist camera at 1080p, 60–70° field of view, fixed focus at 0.2–0.5 m, for the 2 cm keyboard tags A 5 MP camera if the 1 cm key tag needs margin; the mast camera cannot resolve 1 cm at task range
Typing Two-stage visual servo: coarse inverse kinematics, then tag-based fine alignment, with each key confirmed Open-loop typing only if wrist backlash is 0.5 mm or less
Order The cheap tasks banked first, because any order is allowed The hardest task first only if its weight dominates the gate, and that weight is unpublished
Transition One arm configuration for both Autonomy and Equipment, so no re-weigh is triggered S-URC27-QA EQ4 A tip swap inside the 10-minute window, never a module swap

Minimum viable versus stretch

  • The minimum that scores: one completed manipulation task. The cheapest are buttons, switches, knobs, the quarter-turn valve and the hinged latch.
  • The mid-field: four to six tasks banked. UIU’s 72.63 in 2026 shows this band is reachable with a mid-cost 6-DoF arm and a rehearsed panel.
  • The stretch: all nine families, including typing, the key transfer and the hose. One team did it for 100 in 2026, with a prismatic-axis arm and a dedicated typing tip.

Risks to retire early

Risk Why it is real The test that retires it Lead time
Key dropped or lost The task is new and a dropped key ends it, possibly for both locks A tethered key, a mechanical stop in the jaw and a drop test over a padded tray: 50 of 50 grasps, retained through 20 shakes and a 1 m drop 5 months
Key will not insert or turn A 1–2 mm keyway against 4.4–5.2 mm of open-loop backlash at 1 m The same class of 7-pin lock, with insertion force and turn torque measured and the keyhole tags used for visual servoing: 20 of 20 insertions 5 months
Hose will not seat The gasket is now half thickness, and 2026 tuning was to a fitting with no gasket An exact 1.5-inch cam-lock mock, with force logged against depth over 50 cycles and a retention pull to confirm the connected state 4 months
Over-torque cascades to other subsystems West Virginia in 2023 lost its arm cameras through a USB hub, and in 2026 lost cameras and two motors to a damaged cable S-WVU-YT-DOORFAIL S-WVU-NEWS2026 Per-joint current limits and fault-isolated camera paths, so that no single cable takes out more than one camera and no motor 4 months
Typing misalignment or a double press Backlash, motion blur and no auto-repeat One hundred automated presses at 3 mm offsets, each key confirmed by camera: 95 % correct on 3–6 letter keys, including a backspace correction 4 months
Panel slips closed Monash 2022 lost about 9 minutes to two closures S-MONASH-ES22 A prop or hook that holds the panel, rehearsed as its own step: the panel stays open hands-off for 60 seconds 3 months
Drawer jam or the rover pulled The insertion is tight-fitting and the pull force is unmeasured A mock drawer across a range of clearances: ten insertions at the tightest clearance with no rover motion 3 months
Time runs out About 2–3 minutes per task and a 100 m traverse inside 30 minutes A timed full run on a mock panel in the real order: six tasks banked in 20 minutes 3 months
The 10-minute transition Any change in mass or configuration forces a re-weigh S-URC27-QA EQ4 Autonomy, weigh-in and Equipment rehearsed inside the window: 8 minutes end to end, five runs in a row 2 months
Wrong button or switch One mis-press can fail a task A mock labelled exactly like the rulebook figures: 20 of 20 correct on a randomised sequence 2 months

What is not published

  • The per-task weights, which is why tasks can be ranked by risk but not by points per minute.
  • Every force, torque and travel for the key, the lock, the hose, the valve, the drawer, the latch and the switches. A mock and a load cell are the only route to a number.
  • Whether the cache in this mission is the team’s own Astrobiology cache or a judge-supplied one, so the interface must work for both.
  • Whether digits can appear in the typing sequence, since the rule says 3–6 letters.
  • The 2027 partial-credit split for typing corrections.

Further reading

Sources for this page

  1. S-UIU-2026-SARUIU Mars Rover, SAR 2026 (video; AURION, 6-wheel to 4-wheel rocker transition, spring pivot, carbon-core wheel)UIU Mars Rover Team, 2026-02-28 (upload).Original
  2. S-URC27-RULESUniversity Rover Challenge 2027 – Requirements and GuidelinesThe Mars Society / URC.Original Archived copy
  3. S-URC27-TASKHW2027 task hardware resolved from the rulebook's own hyperlinks (keyboard, display, hammer, hose)Robotics Club @ Mississippi State University, 2026-10-25.Print copy
  4. S-MONASH-ES22Monash Nova Rover, Equipment Servicing Task, field footage, URC 2022 (chaptered)Monash Nova Rover, 2022 (uploaded Apr 2023).Original
  5. S-URC27-QAURC2027 Questions & Answers (updated through 2026-10-11)URC Director / judges, 2026-10-11.Original Archived copy
  6. S-URC22-SCORESURC2022 Official ScoresURC, 2022-06.Original
  7. S-URC23-SCORESURC2023 Official ScoresURC, 2023-06.Original
  8. S-URC24-SCORESURC2024 Official ScoresURC, 2024-06.Original
  9. S-URC25-SCORESURC2025 Official ScoresURC, 2025-06.Original
  10. S-URC26-SCORESURC2026 Official ScoresURC, 2026-05/06.Original Archived copy
  11. S-MIST-VURC2021"Virtual URC 2021: MIST Mongol Barota, Session 1" on the official University Rover Challenge channel (1:20:42, uploaded 2021); full judged run: weigh-in 42.1 kg, box check 1.047 m, Equipment Servicing 100/100, Autonomous Navigation failure + manual intervention; auto-caption transcript retrieved with yt-dlp,write-auto-subUniversity Rover Challenge (Mars Society), 2021.Original
  12. S-MST-NEWS2024MRDT Newsletter Winter 2024–25 (PDF, 7.5 MB), URC 2024 review with failure forensics, 2025 design update, 88 active membersMRDT (CEO Abigail Frerking), 2025-01.Original
  13. S-WVU-YT-DOORFAILFailure: Wanderer Struggled to Close a Door with a Broke Arm and Under Strong Wind During URC 2023WVU Team Mountaineers / WVU IRL, 2024-07-13 (event: URC 2023).Original
  14. S-WVU-NEWS2026WVU robotics team places fourth overall, earns top marks at international University Rover ChallengeWVU Statler College (Erik Rudolph), 2026-06-03.Original