pith. sign in

arxiv: 2510.18766 · v2 · pith:YA2XPKYOnew · submitted 2025-10-21 · 💻 cs.RO

Sharing the Load: Autonomous Multi-Rover Cargo Transport

Alexander Krawciw , Luka Antonyshyn , Sven Lilge , Nicolas Olmedo , Faizan Rehmatullah , Maxime Desjardins-Goulet , Pascal Toupin , Timothy D. Barfoot This is my paper

Pith reviewed 2026-05-18 04:56 UTC · model grok-4.3

classification 💻 cs.RO
keywords multi-rover coordinationdistributed model predictive controlshared cargo transportlunar utility vehiclesteach and repeat navigationautonomous roverskinematic decoupling
0
0 comments X

The pith

Two rovers can carry shared cargo while holding a mean relative separation error of 9.2 cm using a distributed controller and custom coupling.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops a distributed model-predictive controller that lets two vehicles transport one shared cargo load. A custom coupling supports the full cargo mass while decoupling the kinematics so each rover can track its own path independently. Field tests on two 800 kg path-to-flight rovers carrying 475 kg cargo kept the average separation error at 9.2 cm and the maximum at 33.4 cm. The system preserves the accuracy of each rover's lidar teach-and-repeat navigation. This setup offers a way for smaller lunar utility vehicles to move large habitat modules over multi-kilometer distances.

Core claim

The paper establishes that a distributed model-predictive controller, used together with a custom cargo coupling that decouples each vehicle's kinematics while fully supporting the cargo mass, enables two rovers to share a load and follow repeated paths. In field tests the rovers maintained a mean relative separation error of 9.2 cm and a maximum error of 33.4 cm while each retained high-quality lidar teach-and-repeat path tracking, and the kinematic freedom of the vehicles supports mission improvements in other coordinated operations.

What carries the argument

Distributed model-predictive controller combined with a custom cargo coupling that decouples vehicle kinematics while supporting full cargo mass.

If this is right

  • Each rover retains its individual high-accuracy lidar teach-and-repeat path tracking during shared cargo transport.
  • The vehicles maintain low relative separation error (mean 9.2 cm) while carrying a 475 kg load in outdoor field tests.
  • Kinematic freedom from the coupling allows the same controller architecture to support other multi-rover operations.
  • Smaller utility rovers gain the capacity and flexibility to move large payloads together for repeated long-distance routes.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The approach could scale to three or more rovers for even heavier or bulkier loads without redesigning the controller.
  • The same distributed coordination might transfer to terrestrial autonomous logistics tasks that require vehicles to stay in formation over repeated routes.
  • Testing the system at the full 5 km distances cited for lunar habitat logistics would reveal whether terrain variation affects the separation error over longer durations.

Load-bearing premise

The custom cargo coupling successfully decouples the kinematics of each vehicle without introducing unmodeled dynamic coupling that would invalidate the independent path tracking assumed by the distributed controller.

What would settle it

A field test on similar terrain in which the mean relative separation error between the two rovers exceeds 10 cm or the lidar teach-and-repeat tracking error for either rover rises above its standalone performance while carrying the 475 kg cargo would show the claim does not hold.

Figures

Figures reproduced from arXiv: 2510.18766 by Alexander Krawciw, Faizan Rehmatullah, Luka Antonyshyn, Maxime Desjardins-Goulet, Nicolas Olmedo, Pascal Toupin, Sven Lilge, Timothy D. Barfoot.

Figure 1
Figure 1. Figure 1: A coupled robot convoy navigating the Woody Loop while [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: The four controllers share similarities but are divided by how the inter-robot distance is measured and what variables are free for the MPC to [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 4
Figure 4. Figure 4: A three-robot convoy repeating in unison using the Distributed [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
read the original abstract

A future lunar habitat, as part of the Artemis program, will require a significant amount of logistics infrastructure. Cargo that is transported to the Moon will need to be moved from a landing site to other key locations that may be up to 5 km away. Teach and repeat navigation is well suited to this task as utility rovers will need to repeat these cargo routes many times. One of the most significant challenges involves the modules that will be assembled together to form the habitat. Canada is studying potential Lunar Utility Vehicle (LUV) designs to carry these large payloads between the landing site and the location of the habitat. As the details of the cargo continue to evolve, using two, smaller LUVs to carry cargo together would provide high capacity and mission flexibility. In this paper, we develop and implement a distributed model-predictive controller that allows vehicles to carry cargo that is shared between them. The algorithm is compared to baselines in small-scale before being implemented onboard two 800 kg path-to-flight rovers and field tested carrying a 475 kg cargo between them. A custom cargo coupling decouples the kinematics of each vehicle while fully supporting the cargo's mass. In our field test, the rovers maintain a relative separation error of 9.2 cm and maximum error of 33.4 cm. This multi-vehicle control architecture retains the high-quality path tracking of lidar teach and repeat for each rover. We demonstrate that kinematic freedom of the vehicles allows a single controller to provide mission improvements for other operations as well.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 1 minor

Summary. The paper develops and implements a distributed model-predictive controller enabling two rovers to cooperatively transport shared cargo. A custom coupling is introduced to support the full cargo mass while decoupling each rover's kinematics, allowing independent path tracking. The approach is first compared to baselines in small-scale tests and then deployed on two 800 kg path-to-flight rovers carrying 475 kg cargo in field trials, where mean relative separation error is 9.2 cm (max 33.4 cm) and lidar teach-and-repeat path tracking quality is retained. The work targets lunar logistics applications under the Artemis program.

Significance. If the central results hold, the demonstration of distributed MPC on representative hardware with substantial payload (475 kg on 800 kg rovers) offers concrete evidence for multi-rover cargo transport in unstructured environments. The retention of high-quality teach-and-repeat performance alongside cooperative cargo handling is a practical strength that could inform flexible logistics architectures for future planetary missions.

major comments (1)
  1. [Abstract / Field-test description] Abstract and field-test description: the claim that the custom cargo coupling 'decouples the kinematics of each vehicle while fully supporting the cargo's mass' is load-bearing for the distributed MPC formulation, which treats the rovers as kinematically independent. No interaction-force measurements, torsional-load data, or sensitivity analysis to unmodeled compliance or lateral forces are referenced, so the reported 9.2 cm mean / 33.4 cm max separation errors cannot yet be unambiguously attributed to successful decoupling rather than controller compensation or test-specific conditions.
minor comments (1)
  1. [Abstract] Abstract: limited detail is provided on controller tuning parameters, exact baseline implementations, and disturbance-handling strategies during the 475 kg cargo runs; expanding this would improve reproducibility.

Simulated Author's Rebuttal

1 responses · 1 unresolved

We thank the referee for their constructive and detailed review. The major comment identifies a valid point about the strength of evidence supporting the cargo coupling's decoupling properties. We respond below and will revise the manuscript to address the concern.

read point-by-point responses

  1. Referee: [Abstract / Field-test description] Abstract and field-test description: the claim that the custom cargo coupling 'decouples the kinematics of each vehicle while fully supporting the cargo's mass' is load-bearing for the distributed MPC formulation, which treats the rovers as kinematically independent. No interaction-force measurements, torsional-load data, or sensitivity analysis to unmodeled compliance or lateral forces are referenced, so the reported 9.2 cm mean / 33.4 cm max separation errors cannot yet be unambiguously attributed to successful decoupling rather than controller compensation or test-specific conditions.

    Authors: We agree that direct force or torsional measurements would provide stronger attribution. The custom coupling uses a central load-bearing hitch with universal joints at the rover interfaces. These joints permit relative yaw and roll motion, mechanically decoupling lateral and heading kinematics while transferring the full vertical cargo load. This underpins the independent kinematic models in the distributed MPC. The field results show that individual lidar teach-and-repeat path tracking quality was fully retained alongside the low separation errors, which would be unlikely if significant unmodeled coupling forces were present. We will revise the manuscript to expand the coupling description with a mechanical schematic, add discussion of the design rationale for kinematic independence, and include a limitations paragraph on unmodeled compliance. However, the test hardware was not equipped with force/torque sensors, so quantitative sensitivity analysis to lateral forces cannot be added from existing data. revision: yes

standing simulated objections not resolved
  • Direct interaction-force measurements, torsional-load data, or quantitative sensitivity analysis to unmodeled compliance from the field trials, as the rovers were not instrumented with force sensors during testing.

Circularity Check

0 steps flagged

No circularity: results from direct hardware measurements

full rationale

The paper presents an implemented distributed MPC for shared-cargo rover transport, validated through small-scale comparisons and field tests on two 800 kg rovers carrying 475 kg cargo. Reported metrics (9.2 cm mean separation error, 33.4 cm max) are direct observations from lidar teach-and-repeat path tracking under the custom coupling. No equations, fitted parameters, or predictions are shown to reduce to inputs by construction; the kinematic-decoupling claim is supported by the physical design and external experimental outcomes rather than self-referential fitting or self-citation chains. The derivation chain is therefore self-contained against hardware benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

The approach relies on standard assumptions of model-predictive control and rigid-body kinematics without introducing new free parameters, axioms, or invented entities beyond the custom coupling mechanism whose performance is validated experimentally.

pith-pipeline@v0.9.0 · 5837 in / 1166 out tokens · 31821 ms · 2026-05-18T04:56:33.241051+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

28 extracted references · 28 canonical work pages

  1. [1]

    Industrial Robotics Review,

    I. Ermolov, “Industrial Robotics Review,” inRobotics: Industry 4.0 Issues & New Intelligent Control Paradigms, A. G. Kravets, Ed., Cham: Springer In- ternational Publishing, 2020, pp. 195–204

    work page 2020

  2. [2]

    Slowik and B

    P. Slowik and B. Sharpe,Automation in the long haul: Challenges and opportunities of autonomous heavy- duty trucking in the United States, Mar. 2018

    work page 2018

  3. [3]

    Spatial Model Predictive Control for Smooth and Accurate Steering of an Autonomous Truck,

    P. F. Lima, M. Nilsson, M. Trincavelli, J. M ˚artensson, and B. Wahlberg, “Spatial Model Predictive Control for Smooth and Accurate Steering of an Autonomous Truck,”IEEE Trans. Intell. V eh., vol. 2, no. 4, pp. 238– 250, Dec. 2017

    work page 2017

  4. [4]

    Robotics in Construction,

    K. S. Saidi, T. Bock, and C. Georgoulas, “Robotics in Construction,” inSpringer Handbook of Robotics, B. Siciliano and O. Khatib, Eds., Cham: Springer International Publishing, 2016, pp. 1493–1520

    work page 2016

  5. [5]

    Robotics in Mining,

    J. A. Marshall, A. Bonchis, E. Nebot, and S. Sched- ing, “Robotics in Mining,” inSpringer Handbook of Robotics, B. Siciliano and O. Khatib, Eds., Cham: Springer International Publishing, 2016, pp. 1549– 1576

    work page 2016

  6. [6]

    Autonomous underground tramming for center-articulated vehi- cles,

    J. Marshall, T. Barfoot, and J. Larsson, “Autonomous underground tramming for center-articulated vehi- cles,”J. Field Robot., vol. 25, no. 6-7, pp. 400–421, 2008

    work page 2008

  7. [7]

    Au- tonomous Driving in Underground Mines via Parallel Driving Operation Systems: Challenges, Frameworks and Cases Study,

    B. Tian, J. Yang, C. Zhang, X. Hao, S. Meng, S. Wang, Z. Yang, L. Chen, Y . Zhao, and S. Ge, “Au- tonomous Driving in Underground Mines via Parallel Driving Operation Systems: Challenges, Frameworks and Cases Study,”IEEE Trans. Intell. V eh., vol. 10, no. 5, pp. 3268–3277, May 2025

    work page 2025

  8. [8]

    Visual teach and repeat for long-range rover autonomy,

    P. Furgale and T. D. Barfoot, “Visual teach and repeat for long-range rover autonomy,”J. Field Robot., vol. 27, no. 5, pp. 534–560, 2010

    work page 2010

  9. [9]

    Radar Teach and Repeat: Architecture and Initial Field Testing,

    X. Qiao, A. Krawciw, S. Lilge, and T. D. Barfoot, “Radar Teach and Repeat: Architecture and Initial Field Testing,” in2025 IEEE Int. Conf. Robot. Autom. ICRA, May 2025, pp. 13 021–13 027

    work page 2025

  10. [10]

    Are We Ready for Radar to Replace Lidar in All-Weather Mapping and Localization?

    K. Burnett, Y . Wu, D. J. Yoon, A. P. Schoellig, and T. D. Barfoot, “Are We Ready for Radar to Replace Lidar in All-Weather Mapping and Localization?” IEEE Robot. Autom. Lett., vol. 7, no. 4, pp. 10 328– 10 335, Oct. 2022

    work page 2022

  11. [11]

    Autonomous Convoying: A Survey on Current Research and Development,

    S. Nahavandi, S. Mohamed, I. Hossain, D. Nahavandi, S. M. Salaken, M. Rokonuzzaman, R. Ayoub, and R. Smith, “Autonomous Convoying: A Survey on Current Research and Development,”IEEE Access, vol. 10, pp. 13 663–13 683, 2022

    work page 2022

  12. [12]

    An Overview on Ap- proaches for Coordination of Platoons,

    V . Lesch, M. Breitbach, M. Segata, C. Becker, S. Kounev, and C. Krupitzer, “An Overview on Ap- proaches for Coordination of Platoons,”IEEE Trans. Intell. Trans. Sys., vol. 23, no. 8, pp. 10 049–10 065, Aug. 2022

    work page 2022

  13. [13]

    Design of leader’s path following system for multi-vehicle au- tonomous convoy,

    X. Zhao, W. Yao, N. Li, and Y . Wang, “Design of leader’s path following system for multi-vehicle au- tonomous convoy,” in2017 IEEE Int. Conf. Unmanned Syst. ICUS, Oct. 2017, pp. 132–138

    work page 2017

  14. [14]

    Leader-follower func- tion for autonomous military convoys,

    L. Vasseur, O. Lecointe, J. Dento, N. Cherfaoui, V . Marion, and J. G. Morillon, “Leader-follower func- tion for autonomous military convoys,” inUnmanned Ground V eh. Technol. VI, vol. 5422, SPIE, Sep. 2004, pp. 326–337

    work page 2004

  15. [15]

    A Perception-Based Architecture for Autonomous Con- voying in GNSS-Denied Areas,

    A. Bienemann, L. Beer, A. Reich, T. Steinecker, B. Forkel, A. Backhaus, P. Berthold, J. D. Gonz ´alez, P. Mortimer, T. Luettel, and M. Maehlisch, “A Perception-Based Architecture for Autonomous Con- voying in GNSS-Denied Areas,”IEEE Trans. Field Robot, pp. 1–1, 2025

    work page 2025

  16. [16]

    Co- operative Object Transport in Multi-Robot Systems: A Review of the State-of-the-Art,

    E. Tuci, M. H. M. Alkilabi, and O. Akanyeti, “Co- operative Object Transport in Multi-Robot Systems: A Review of the State-of-the-Art,”Front. Robot. AI, vol. 5, May 2018

    work page 2018

  17. [17]

    A Comprehensive Survey on Vehicu- lar Networking: Communications, Applications, Chal- lenges, and Upcoming Research Directions,

    N. H. Hussein, C. T. Yaw, S. P. Koh, S. K. Tiong, and K. H. Chong, “A Comprehensive Survey on Vehicu- lar Networking: Communications, Applications, Chal- lenges, and Upcoming Research Directions,”IEEE Access, vol. 10, pp. 86 127–86 180, 2022

    work page 2022

  18. [18]

    GEOMOVE: Detached AGVs for Cooperative Transportation of Large and Heavy Loads in the Aeronautic Industry,

    C. Rizzo, A. Lagra ˜na, and D. Serrano, “GEOMOVE: Detached AGVs for Cooperative Transportation of Large and Heavy Loads in the Aeronautic Industry,” in2020 IEEE Int. Conf. Auton. Robot Syst. Compet. ICARSC, Apr. 2020, pp. 126–133

    work page 2020

  19. [19]

    A LiDAR-Based Dual-Vehicle Collaborative Transportation Approach Without Communication,

    F. Xie, H. Liu, M. Li, and M. Xia, “A LiDAR-Based Dual-Vehicle Collaborative Transportation Approach Without Communication,”IEEE Access, vol. 13, pp. 104 406–104 421, 2025

    work page 2025

  20. [20]

    MeWBots: Mecanum-Wheeled Robots for Collaborative Manipulation in an Obstacle-Clustered Environment Without Communication,

    A. Rauniyar, H. C. Upreti, A. Mishra, and P. Sethura- malingam, “MeWBots: Mecanum-Wheeled Robots for Collaborative Manipulation in an Obstacle-Clustered Environment Without Communication,”J Intell Robot Syst, vol. 102, no. 1, p. 3, May 2021

    work page 2021

  21. [21]

    Force Distribution and Estimation for Cooperative Transportation Control on Multiple Unmanned Ground Vehicles,

    F. Huzaefa and Y .-C. Liu, “Force Distribution and Estimation for Cooperative Transportation Control on Multiple Unmanned Ground Vehicles,”IEEE Trans. Cybern., vol. 53, no. 2, pp. 1335–1347, Feb. 2023

    work page 2023

  22. [22]

    Behavior-based multi-robot collab- oration for autonomous construction tasks,

    A. Stroupe, T. Huntsberger, A. Okon, H. Aghazarian, and M. Robinson, “Behavior-based multi-robot collab- oration for autonomous construction tasks,” in2005 IEEERSJ Int. Conf. Intell. Robots Syst., Aug. 2005, pp. 1495–1500

    work page 2005

  23. [23]

    Collaborative Transport Strategy for Dual AGVs in Smart Ports: En- hancing Docking Accuracy in No-Load Formations,

    Q. Zhang, W. Li, L. Guo, and X. Qi, “Collaborative Transport Strategy for Dual AGVs in Smart Ports: En- hancing Docking Accuracy in No-Load Formations,” J. of Marine Sci. and Eng., vol. 13, no. 1, p. 81, Jan. 2025

    work page 2025

  24. [24]

    Kinematic and dynamic vehicle models for autonomous driving control design,

    J. Kong, M. Pfeiffer, G. Schildbach, and F. Bor- relli, “Kinematic and dynamic vehicle models for autonomous driving control design,” in2015 IEEE Intell. V eh. Symp., Jun. 2015, pp. 1094–1099

    work page 2015

  25. [25]

    VT&R3: Generalizing the Teach and Re- peat Navigation Framework,

    Y . Wu, “VT&R3: Generalizing the Teach and Re- peat Navigation Framework,” MASc, University of Toronto, Toronto, Canada, Sep. 2022

    work page 2022

  26. [26]

    CasADi – A software framework for nonlinear optimization and optimal control,

    J. A. E. Andersson, J. Gillis, G. Horn, J. B. Rawlings, and M. Diehl, “CasADi – A software framework for nonlinear optimization and optimal control,”Math. Program. Comput., vol. 11, no. 1, pp. 1–36, 2019

    work page 2019

  27. [27]

    On the implementation of an interior-point filter line-search algorithm for large-scale nonlinear programming,

    A. W ¨achter and L. T. Biegler, “On the implementation of an interior-point filter line-search algorithm for large-scale nonlinear programming,”Math. Program., vol. 106, no. 1, pp. 25–57, Mar. 2006. [28]Zenoh, Eclipse Foundation. https://zenoh.io/

    work page 2006

  28. [28]

    Multi-Robot Systems and Cooperative Object Trans- port: Communications, Platforms, and Challenges,

    X. An, C. Wu, Y . Lin, M. Lin, T. Yoshinaga, and Y . Ji, “Multi-Robot Systems and Cooperative Object Trans- port: Communications, Platforms, and Challenges,” IEEE Open J. Comput. Soc., vol. 4, pp. 23–36, 2023

    work page 2023