PaperlayerView exact revision on arXiv

Source-linked AI summary

Path-Following Control and Terramechanics Analysis for Planetary Rovers Under Wheel-to-Wheel Traction Asymmetry

Ryuya Matsuoka, Keisuke Takehana, Kentaro Uno, Toshinori Kuwahara, Kazuya Yoshida

arXiv:2609.38715v1cs.RO

TL;DR

Path following on loose terrain is challenging because conventional wheel acceleration can trigger slip-sinkage and immobilization. The paper uses a model-free controller that corrects heading through selective deceleration; experiments report reduced lateral drift without sinkage and measured load equalization.

  • Problem

    On loose terrain, conventional outer-wheel acceleration during path recovery can cause shear failure, slip-sinkage, and immobilization.

    Source: A. Related Works · p. 2

  • Method

    A model-free controller caps commanded velocity and corrects heading by selectively decelerating outer wheels, using terrain resistance as an anchor.

    Source: B. Contributions · p. 2

  • Results

    Across velocity regimes up to 0.7 m/s, deceleration suppressed cumulative lateral drift without sinkage and equalized dynamic loads, restoring positive traction to the opposite driving wheel.

    Source: §V · p. 7

  • Takeaways & Limitations

    The results support prioritizing continuous mobility on soil over strict speed tracking as a locomotion strategy for wide-area, high-speed lunar exploration.

    Source: §V · p. 7

    On steep inclinations or transverse slopes, reduced commanded velocity may deplete forward momentum and risk rollback or traction stall; a hybrid deceleration and torque-vectoring framework is proposed for difficult conditions.

    Source: D. Discussion on Kinetic Energy… · p. 7

Abstract

from arXiv · show

This paper proposes a control strategy for path following that is model-free and relies solely on deceleration for skid-steering planetary rovers navigating deformable loose terrain under continuously imposed traction asymmetry. While conventional controllers that are based on kinematics frequently cause slip-sinkage entrapment by accelerating the wheels during path correction, the proposed approach prevents this failure by setting an upper limit on the maximum commanded velocity. Heading correction is achieved solely through the selective deceleration of the outer wheels, which are located on the outside of the turn, driving them into a negative slip regime to act as a mechanical anchor. The system was evaluated using the four-wheel independent-drive rover EX1 under an asymmetric wheel configuration with different left and right grouser heights that induces significant deviations from the path. Experimental results demonstrate that this deceleration-only control successfully suppresses accumulated lateral drift across various velocity regimes up to 0.7 m/s without causing sinkage. Crucially, direct force measurements from onboard multi-axis sensors provide important empirical evidence of the underlying terramechanics, proving that the targeted deceleration establishes dynamic load equalization across the chassis and completely restores the native thrust capability of the opposite driving wheel.

I. INTRODUCTION

Planetary exploration is shifting toward wider-area, higher-speed traverses, while loose-soil mobility remains difficult to predict and control. Prior approaches face challenges from uncertain soil parameters and outer-wheel acceleration on deformable terrain.

I. INTRODUCTION

  • Planetary exploration is shifting toward wide-area, higher-speed traverses, challenging mobility strategies designed for slow rovers.Source: §I · p. 1
  • Loose regolith makes traction, sinkage, resistance, and slip depend strongly on wheel–soil deformation, complicating mobility prediction and control.Source: §I · p. 1

A. Related Works

  • Existing path-following and traction methods include terrain estimation and torque distribution, while outer-wheel acceleration risks shear failure, slip-sinkage, and immobilization on loose ground.Source: A. Related Works · pp. 1–2

B. Contributions

The paper proposes a model-free path-following controller that corrects heading through deceleration rather than wheel acceleration. Its experiments use asymmetric wheel traction and direct force measurements to assess recovery and the underlying terramechanics.

B. Contributions

  • The model-free controller caps commanded velocity and corrects heading only by decelerating outer wheels, avoiding online slip and soil-parameter estimation.Source: B. Contributions · p. 2
  • An asymmetric left-right grouser-height arrangement imposes steady thrust imbalance on loose terrain as an experimental path-recovery benchmark.Source: B. Contributions · p. 2
  • Direct force measurements show selective deceleration equalized dynamic normal loads and restored thrust without sinkage across velocity regimes up to 0.7 m/s.Source: B. Contributions · p. 2

A. The EX1 Rover System

EX1 is a four-wheel independent-drive skid-steering rover designed for high-speed loose-soil locomotion research. The experiments use asymmetric grouser heights to create a repeatable traction imbalance and heading drift.

A. The EX1 Rover System

  • EX1 is a four-wheel independent-drive, skid-steering rover built to investigate high-speed locomotion on loose soil, with a maximum speed of 1.0 m/s.Source: A. The EX1 Rover System · p. 2
  • An onboard IMU supplies real-time yaw rate, while wheel-mounted six-axis force/torque sensors measure each wheel’s drawbar pull and vertical load.Source: A. The EX1 Rover System · p. 2

B. Asymmetric wheel configuration

C. Conventional Path Tracking and its Limitations

Conventional skid-steering path correction accelerates the outer wheel beyond its reference speed. On loose soil, that can trigger a slip-sinkage cycle that degrades traction and risks immobilization.

C. Conventional Path Tracking and its Limitations

III. METHOD

The path-following system feeds back the rover’s measured yaw behavior to command the left and right wheels, while generating corrective yaw through deceleration. It estimates heading from integrated IMU yaw rate and defines error relative to the target heading.

III. METHOD

  • Corrective yaw is generated through deceleration, with the rover’s yaw behavior fed back to calculate left- and right-wheel commands.The primary constraint is intended to prevent slip-sinkage.Source: §III · p. 3

A. Heading Estimation and Error Calculation

  • The onboard IMU supplies yaw rate, which is integrated over time to estimate the rover’s actual heading angle.The estimate is compared with the target heading angle.Source: A. Heading Estimation and Error… · p. 3
  • Heading error θ(t) is defined relative to the target heading angle ψref.The section introduces θ(t) as the difference between actual and target heading.Source: A. Heading Estimation and Error… · p. 3

B. Sensor Noise Mitigation and Proportional Control

The controller filters heading error with a deadband and applies proportional control. Its corrective yaw moment is produced strictly by deceleration, with commanded velocity capped at the reference speed.

B. Sensor Noise Mitigation and Proportional Control

  • The controller sets corrective input to zero when absolute heading error is within θdb, limiting motor chattering and unnecessary soil shear from noisy yaw measurements.The deadband addresses high-frequency components in the measured yaw rate.Source: B. Sensor Noise Mitigation and… · p. 3
  • The control input u(t) is obtained by multiplying filtered heading error θ′(t) by proportional gain Kp.The implementation used Kp = 3.0 m/(s · rad).Source: B. Sensor Noise Mitigation and… · pp. 3–4
  • The corrective yaw moment is generated strictly through deceleration by keeping maximum commanded velocity at or below vref.This is the stated constraint in the control-system block diagram.Source: Figure 3 · p. 4
  • The lower velocity limit is vmin = 0.2 m/s, preventing backward wheel rotation that could excavate regolith.The 50.0 Hz loop used a deadband threshold of θdb = 0.04 rad (≈2.29°), selected to exceed IMU yaw sensor noise while limiting actuator chattering and localized soil shear.Source: B. Sensor Noise Mitigation and… · pp. 3–4

C. Deceleration-Only Distribution with Saturation Limit

The controller distributes correction by selectively decelerating one wheel while capping the maximum commanded velocity at vref. The evaluation imposes steady traction asymmetry with different grouser heights and compares open- and closed-loop traversals at three speeds.

C. Deceleration-Only Distribution with Saturation Limit

  • Wheel commands are bounded below by vmin and subtract only the positive deceleration component of u(t) from the reference velocity.The left and right commands use opposite signs of u(t) to select which wheel decelerates.Source: C. Deceleration-Only Distribution with… · p. 4
  • When drift is rightward (u(t) < 0), only the left wheel is decelerated to generate a leftward yaw moment, while the right wheel maintains reference speed.The maximum commanded velocity therefore never exceeds vref.Source: C. Deceleration-Only Distribution with… · p. 4

A. Experimental Setup and Methodology

  • The asymmetric test configuration used 15.0 mm left-wheel grousers and 5.4 mm right-wheel grousers, inducing thrust imbalance and rightward drift.Open-loop and proposed closed-loop trials were conducted at vref = 0.3, 0.5, and 0.7 m/s, with three independent trials per speed.Source: A. Experimental Setup and Methodology · p. 4
  • Open-loop trials drifted steadily in heading, whereas the proposed closed-loop control achieved path correction during the asymmetric traversal.A total station tracked a prism mounted on the rover chassis to measure trajectories and lateral deviations.Source: A. Experimental Setup and Methodology · p. 4, Figure 4 · p. 4

B. Trajectory Analysis and Path Tracking Performance

With the asymmetric wheel configuration, open-loop trajectories showed substantial rightward drift, while closed-loop control kept the rover near the reference path across tested speeds up to 0.7 m/s without slip-sinkage immobilization.

B. Trajectory Analysis and Path Tracking Performance

  • Table II summarizes steady-state traction performance across independent trials, and Table III compares dynamic normal loads.The supplied captions do not report numerical table values.Source: Table II · p. 5, Table III · p. 5
  • Across velocity regimes up to 0.7 m/s, closed-loop control constrained lateral deviation near Y = 0, while open-loop trajectories drifted rightward.The plotted curves represent median trajectories filtered from three independent trials.Source: B. Trajectory Analysis and Path… · pp. 4–5, Figure 5 · p. 5
  • The closed-loop controller maintained stable path keeping under continuous traction imbalance without triggering slip-sinkage immobilization.The comparison uses the asymmetric wheel configuration on loose soil.Source: B. Trajectory Analysis and Path… · pp. 4–5

C. Dynamic Traction Analysis and Mechanical Insights

Direct force measurements show that selective braking reverses the asymmetric traction state: it balances chassis loads, corrects heading, and improves net traction. Under open-loop control, lateral drift and side-slip concentrate load on the shorter-grouser wheel, which remains in drag.

Control OFF

  • OFF, the shorter-grouser right wheel produced −5.83 ± 3.79 N at 0.7 m/s, with its load at 65.20±6.87 N versus 45.73 ± 6.14 N on the left.The right wheel remained in drag as lateral resistance shifted vertical load toward it.Source: C. Dynamic Traction Analysis and… · pp. 5–6

Control ON

  • With control ON, the left wheel entered braking at −9.70 ± 4.43 N while the right wheel shifted to positive traction at +3.02 ± 3.41 N.The synchronized traction dissociation was accompanied by quasi-symmetric vertical loading and was recorded in the dynamic profiles.Source: C. Dynamic Traction Analysis and… · p. 6, Figure 6 · p. 6

Load equalization and net traction

  • The reported net-traction improvements were +8.13 N, +9.71 N, and +8.85 N across velocity regimes, alongside balanced dynamic loads of 54.43±7.28 N left and 57.52 ± 5.02 N right.The paper attributes the changes to heading recovery and dissipation of lateral-load transfer.Source: C. Dynamic Traction Analysis and… · p. 6, Figure 7 · p. 6

Mechanical interpretation

  • The authors describe the chassis as a seesaw: braking energy from the left wheel is transferred through the chassis, lifting the right wheel from drag and restoring its thrust capability.The reported traction inversion is shown in Fig. 7(b).Source: Figure 7 · p. 6, C. Dynamic Traction Analysis and… · p. 7

D. Discussion on Kinetic Energy Trade-off and Operating Limitations

Deceleration-only steering trades forward drawbar pull during correction for slip-sinkage suppression. The tested flat-regolith setting maintained momentum, but steep inclines, transverse slopes, and high-resistance obstacles may require a hybrid controller.

Kinetic-energy trade-off

  • During steering interventions, selectively decelerating the outer wheel decreases net forward drawbar pull, although the opposite wheel maintained momentum on tested flat regolith with vmin = 0.2 m/s.The authors identify preservation of forward kinetic energy as the physical trade-off.Source: D. Discussion on Kinetic Energy… · p. 7

Operating boundary

  • On steep inclines or transverse slopes, persistent velocity reduction risks depleting momentum and may induce gravitational roll-back or complete traction stall.These scenarios are outside the flat-regolith benchmark conditions described.Source: D. Discussion on Kinetic Energy… · p. 7

Disturbances and future control

  • The study focuses on continuous heading disturbance from grouser asymmetry; abrupt localized disturbances and extreme slopes or high-resistance obstacles motivate a hybrid anchoring and torque-vectoring framework.The proposed framework would balance slip prevention against climbability.Source: D. Discussion on Kinetic Energy… · p. 7

V. CONCLUSION

The paper presents a model-free, deceleration-only path-following strategy for skid-steering rovers on loose deformable terrain. EX1 experiments with asymmetric wheels showed reduced lateral drift up to 0.7 m/s, while force measurements link deceleration to load redistribution and recovered traction.

V. CONCLUSION

  • The controller caps commanded velocity and corrects heading by selectively decelerating the outer wheels, avoiding the slip-sinkage cycle associated with velocity and acceleration tracking on sand.The approach is model-free and intended for skid-steering rovers on loose deformable terrain.Source: §V · p. 7
  • EX1 experiments with asymmetric wheels showed that the controller suppressed cumulative lateral drift across velocity regimes up to 0.7 m/s.The asymmetric wheel arrangement induced heading disturbances.Source: §V · p. 7
  • Force measurements indicate that deceleration uses terrain resistance as an anchor, equalizes dynamic normal loads, and restores positive traction to the opposite driving wheel.The conclusion describes future evaluation on curved paths and sloped, heterogeneous terrain with non-uniform soil shear strengths.Source: §V · p. 7
Loading 2609.38715v1…