PaperlayerView exact revision on arXiv

Source-linked AI summary

Designing an Efficient Excavator Bucket for Lunar ISRU: A Comparative Study with Vision-Based Fill and Displacement Analysis

Abdulla Hil Kafi, Tomoki Koshi, Jorge Casir, Kenji Nagaoka

arXiv:2609.25724v1cs.RO

TL;DR

Lunar excavation faces constraints from reduced gravity, abrasive regolith, and mission limits on mass and energy. This paper evaluates a spiral-cavity wheel using lightweight vision- and log-based tools; it reports higher fill ratios and excavation rates than two baseline wheels.

  • Problem

    Lunar excavation faces reduced-gravity soil dynamics, abrasive regolith, and mission constraints on system mass and energy.

    Source: §I · p. 1

  • Method

    The paper experimentally evaluates a spiral-cavity regolith-collection wheel against two baselines using lightweight vision- and log-based tools.

    Source: §I · p. 1

  • Results

    The spiral-cavity wheel achieved higher fill ratios and excavation rates than two baseline designs, with improved regolith collection per unit energy and penetration depth.

    Source: §VIII · p. 6

  • Takeaways & Limitations

    The proposed geometry is presented as a promising candidate for integrated excavation systems in ISRU missions.

    Source: §VIII · p. 6

Abstract

from arXiv · show

This paper present a spiral-cavity wheel for lunar regolith excavation and a sensor-light evaluation stack that jointly estimates fill ratio (vision), sinkage (vision), and specific energy from actuator logs. In benchtop tests (four revolutions at 5, 10, and 15~RPM) against two literature baselines, the proposed wheel achieved higher excavated mass and fill ratio, delivering 2.2-3.0 times higher excavation rate while reducing specific energy by 29 % relative to a bucket-drum baseline. Normalized sinkage (mm/kg) was also lower, indicating stable traction without bogging. Effort-time traces show a steady torque envelope with repeatable cut-carry-dump cycles across speeds. We provide a retention index $\eta$ that correlates with fill ratio and a DEM setup that reproduces experimental trends with low error. Results suggest spiral-cavity wheels can replace heavier multi-actuator diggers when mass, simplicity, and energy efficiency are mission drivers.

I. INTRODUCTION

Lunar excavation must operate under reduced gravity, abrasive regolith, and mission constraints on mass and energy. The paper proposes a spiral-cavity wheel and lightweight vision- and log-based tools to evaluate excavation against two baselines.

I. INTRODUCTION

  • Lunar excavation supports oxygen production, shielding, and construction, but reduced gravity, abrasive particles, and mission constraints challenge system design.Source: §I · p. 1
  • Bucket drums add deployment mass and complexity, while robotic arms impose high power and control demands, motivating simpler integrated solutions.Source: §I · p. 1
  • A wheel combining mobility and regolith collection could reduce subsystem redundancy, but efficiency, energy use, and sinkage require experimental quantification.Source: §I · p. 1
  • A spiral-cavity wheel was designed for regolith retention and experimentally benchmarked against two literature-derived baselines.The evaluation uses vision- and log-based tools without external instrumentation.Source: §I · p. 1
  • The paper also identifies excavation-specific energy analysis as a contribution.Source: §I · p. 1

II. MOTIVATION

The paper motivates integrating excavation with rover mobility and addresses gaps in wheel excavation, DEM validation, and excavation-specific measurement. It presents a wheel design and lightweight evaluation tools in that context.

II. MOTIVATION

  • Integrating regolith collection into a rover wheel is proposed as an alternative to a dedicated excavation mechanism, with the aim of reducing system mass, power consumption, and mechanical complexity.Source: §II · p. 1

III. RELATED WORK

  • Wheel excavation remains underexplored as a design objective, while prior wheel studies examine traction, retention, torque, and sinkage.Prior studies include grousered wheels, drum-type wheels with discrete scoops, and single-wheel tests on cohesive soils.Source: §III · p. 2
  • DEM can reproduce granular-flow and excavation trends when calibrated, but wheel excavators remain under-validated across geometries and operating conditions.Source: §III · p. 2
  • Vision and current-based methods estimate mobility-related quantities, but excavation-specific metrics and energy are seldom quantified.Source: §III · p. 2
  • The paper combines a wheel design that integrates excavation into rover architecture with lightweight vision-based evaluation tools that require no external instrumentation.Source: §III · p. 2

IV. WHEEL DESIGNS

Three wheel designs were evaluated, including a proposed spiral-cavity wheel intended to retain regolith during excavation. Its geometry and retention index are described, alongside its laboratory-scale fabrication.

IV. WHEEL DESIGNS

  • The study evaluated three wheel designs under controlled conditions to compare excavation performance.Source: §IV · p. 2

A. Design A (Proposed Wheel)

  • Design A uses a central spiral cavity as a passive regolith holder, intended to reduce losses during lifting compared with the open centers of Designs B and C.Its dimensions were selected to match rover suspension geometry and actuator torque limits; the cavity curves use tangential continuity for smooth transitions.Source: A. Design A (Proposed Wheel) · p. 2
  • The spiral profile completes one turn, places the inner scoop end at 0.2 times the wheel radius, spaces scoops uniformly, and connects curves tangentially.Source: A. Design A (Proposed Wheel) · p. 2
  • The soil-holding ratio η compares approximate cavity volume with the wheel’s bounding cylindrical volume and serves as an index of retention potential, not an exact volumetric measurement.Its ranking is reported to correlate with experimentally measured fill ratios.Source: A. Design A (Proposed Wheel) · pp. 2–3
  • The final geometry was modeled in CAD and fabricated from PLA using fused deposition modeling for laboratory experiments.Source: A. Design A (Proposed Wheel) · p. 3

B. Design B (Bucket Drum-Inspired Wheel):

The study compares a reconstructed bucket-drum-inspired wheel with a Keio wheel design, producing prototypes at a scale matched to the proposed design.

B. Design B (Bucket Drum-Inspired Wheel):

  • Design B distributes uniformly spaced buckets around the wheel periphery, with an aggressive cutting profile that imposes higher torque demand.Its CAD geometry was reconstructed from published dimensions and fabricated as a PLA prototype.Source: B. Design B (Bucket Drum-Inspired Wheel) · p. 3

C. Design C (Keio Wheel):

  • Design C uses shallower scoops and surface features intended to improve retention while keeping torque requirements moderate, representing a hybrid of cutting and lifting.Its reconstructed CAD and PLA prototype matched the scale of Designs A and B for comparison.Source: C. Design C (Keio Wheel) · p. 3
  • Table I is identified as presenting the wheel geometry used in the experiments.Source: Table I · p. 3

A. Testbed-based excavation experiment

The experiments used a custom testbed to compare wheel excavation, while DEM simulations reproduced experimental mass trends with close agreement.

A. Testbed-based excavation experiment

  • A HEBI X8-16 smart actuator supplied real-time torque, angular velocity, and energy data without external load cells.Wheels were fixed in silica sand with bulk density 1282 kg/m³ and mean particle size 510 µm.Source: A. Testbed-based excavation experiment · p. 3
  • Each wheel was tested at 5, 10, and 15 RPM, with three repetitions per condition and collected sand weighed after each trial.The wheel ran for 48, 24, and 12 seconds at those respective speeds.Source: A. Testbed-based excavation experiment · p. 3

B. DEM Simulation

  • The simulation’s predicted mass trends closely matched observed excavation outcomes; the experimental post-excavation mass was 1.192 kg, and bulk density was set to 1282 kg/m³.Source: B. DEM Simulation · p. 4
  • The DEM setup depicts an initially embedded cylindrical tool and its post-excavation state, with 1.192 kg total excavated mass shown.Simulation parameters followed Nakano et al., except for bulk density, set to match the experimental material.Source: Figure 5 · p. 3, B. DEM Simulation · p. 4

VI. DATA ACQUISITION AND ANALYSIS

The vision-based method segments regolith within the wheel’s circular image region, then uses pixel counts and optional calibration to estimate fill and mass.

A. Vision-Based Fill Ratio and Mass

  • A circular region of interest is segmented into a binary regolith mask using HSV thresholding, CIE–Lab k-means, or interactive GrabCut.The region is defined by the wheel circle’s center and radius.Source: A. Vision-Based Fill Ratio and Mass · p. 4
  • The uncalibrated fill fraction is the regolith pixel count divided by the wheel-circle area, with optional empty/full two-point calibration.The pixel count is Pt and circle area is A; calibrated fill ratio is then used with inner diameter, axial depth, and bulk density to estimate mass.Source: A. Vision-Based Fill Ratio and Mass · p. 4

B. Video-Based Vertical Displacement (Sinkage)

The displacement method locates the wheel center in start and end frames, converts pixel distances to metric units, and reports vertical displacement as sinkage.

B. Video-Based Vertical Displacement (Sinkage)

C. Actuator-Log Power, Energy, and Rate

Actuator logs provide energy and rate measures, while torque–angle loops expose cut–carry–dump cycles. The comparison covers three wheel designs and speeds, with Design A reported as superior in retention and throughput.

C. Actuator-Log Power, Energy, and Rate

  • Specific energy is total actuator energy divided by collected mass, and excavation rate is collected mass divided by test duration.Actuator effort and angular velocity are integrated over the trimmed interval; available position data also supports revolution segmentation and torque–angle loops.Source: C. Actuator-Log Power, Energy, and Rate · p. 4

VII. RESULTS AND DISCUSSION

  • The results use silica sand as a regolith simulant, and Table II summarizes four-revolution tests across three wheel designs and speeds.The paper notes that silica sand differs from lunar regolith in mineral composition and particle angularity but is used to capture wheel–soil interaction trends and benchmark DEM predictions.Source: §VII · p. 4, Table II · p. 5
  • Design A shows superior soil retention and throughput, with favorable efficiency relative to Designs B and C across 5, 10, and 15 RPM.The comparison includes fill ratio, excavation rate, specific energy, and sinkage per unit mass.Source: Figure 8 · p. 5

A. Excavation Capacity

Design A achieved higher fill ratios and excavation rates than the two baseline designs, with its advantage strongest at higher speeds. It used less energy per kilogram than Design B and required less sinkage per kilogram than either baseline.

A. Excavation Capacity

  • 37.3% mean fill ratio for Design A, versus 32.4% for Design B and 29.0% for Design C; at 10 RPM, its excavation rate was 3.36 kg/min.Source: A. Excavation Capacity · p. 5

B. Energy and Efficiency

  • 59.1 J/kg specific energy for Design A was 29% lower than Design B’s 83.5 J/kg, but higher than Design C’s 44.2 J/kg.Source: B. Energy and Efficiency · p. 5

C. Sinkage Behavior

  • 72.6 mm/kg normalized sinkage for Design A was 42% lower than Design B and 31% lower than Design C.Design A had greater absolute sinkage than either baseline.Source: C. Sinkage Behavior · p. 5

D. Torque Observations

Design A’s spiral cavity improved soil retention and excavation efficiency, while torque traces showed repeatable engagement across tested speeds. DEM simulations reproduced measured mass trends, with the reported prediction error varying by speed.

D. Torque Observations

  • Across speeds, Design A maintained a consistent torque envelope with repeatable cut–carry–dump cycles and no stall-like spikes.The traces provide sensor-level evidence of wheel–regolith interaction; the time-integrated signal relates to mechanical energy.Source: D. Torque Observations · pp. 5–6, Figure 9 · p. 6

E. Dem Simulation Insight

  • 1.06% prediction error at 15 RPM; DEM predicted 1.439 kg against the measured mass, with 9.48% MAPE across tested speeds.The figure shows mass accumulation over time at 5, 10, and 15 RPM; the table compares DEM and experimental excavated mass per four revolutions.Source: E. Dem Simulation Insight · p. 6, Figure 10 · p. 6, Table III · p. 6

F. Overall Discussion

  • Design A delivered more regolith per unit sinkage and energy than baseline wheels despite greater torque and absolute sinkage.The soil-holding ratio η was reflected in measured fill ratios.Source: F. Overall Discussion · p. 6

VIII. CONCLUSIONS AND FUTURE WORK

The spiral-cavity wheel collected more regolith than two baseline designs, with improved collection per unit energy and penetration depth despite greater torque and absolute sinkage. Future work proposes broader environmental, mobility, and system-level testing.

VIII. CONCLUSIONS AND FUTURE WORK

  • The spiral-cavity wheel achieved higher fill ratios and excavation rates than two baseline designs, with improved regolith collection per unit energy and penetration depth.The authors describe its geometry as a promising candidate for integrated ISRU excavation systems.Source: §VIII · p. 6
  • Future experiments and DEM simulations will cover denser and cohesive simulants, while evaluating reduced gravity, vacuum, and regolith abrasiveness.Source: §VIII · pp. 6–7
  • The retention index η is planned for further development into a predictive framework across wheel geometries and soil types.Source: §VIII · p. 7
  • Planned development includes a double-bucket configuration, horizontal mobility tests, and integration with a four-wheeled rover for system-level testing.Source: §VIII · p. 7
Loading 2609.25724v1…