Tension Tells the Tail: Cable-Obstacle Interaction Sensing for Tethered Mobile Robot Navigation

Ryuki Shimada; Koki Suzuki; Genya Ishigami

cs.ROOct 17, 2026

Tension Tells the Tail: Cable-Obstacle Interaction Sensing for Tethered Mobile Robot Navigation

Authors: Ryuki Shimada, Koki Suzuki, Genya Ishigami

Organizations: The authors are with the Space Robotics Group, Department of Mechanical Engineering, Keio University, Yokohama, Kanagawa 223-8522, Japan.

Abstract

Tethered mobile robots can sustain long-duration operation through continuous power supply and stable communication via their cables. However, cable-environment interaction causes unintended obstacle displacement, cable snagging or entanglement, and even cable damage, leading to mission failure. To assess the risk of individual cable-obstacle contacts, this study investigates cable tension as a sensing modality for estimating whether a contacted obstacle is movable or non-movable. We conducted experiments with a tethered mobile robot and collected tension data while varying robot velocity, contact angle between the obstacle and cable, obstacle mass, and obstacle side-surface material. The results show that obstacle movability is characterized by the initial tension rise and, more importantly, by whether the tension relaxes or remains high after its peak. Based on these observations, we select six features from the tension profiles and train a logistic-regression classifier using 117 labeled trials. The proposed method distinguishes movable and non-movable cases with a balanced accuracy of 95.5% after 30 cm of post-contact travel. These results demonstrate the potential of cable tension for risk assessment of cable-obstacle interactions and provide a step toward online, cable-aware autonomous navigation.

Figures & tables

Figure 1: Overview of the cable–obstacle interaction sensing task. As the robot moves, the cable contacts the obstacle and generates tension; the goal is to estimate whether the obstacle is movable or non-movable from the tension measured within a short travel distance after contact.

Figure 2: Cable friction around a cylindrical obstacle. Capstan equation governs the tension distribution along the cable–obstacle contact region.

Panel of Figure 3 — Figure 3: Experimental setup. (A) Jackal robot with a force/torque sensor at its rear; the sensor $z$ -axis is aligned with the robot’s longitudinal axis. (B) Test field with the cable base, whose constant-force spring (close-up in the red box) keeps the cable pretension constant; the red arrow shows the robot’s direction of travel. (C) Steel cans with three side-surface materials (bare steel, sandpaper #400, and rubber sheet), loaded with weights (top right) to set the obstacle mass.

Category	Model	Size	Additional specifications
Robot	Clearpath Robotics Jackal	580 $\times$ 430 $\times$ 250 mm	Mass 17 kg, Max speed 2.0 m/s
Cable	Blue Robotics Fathom ROV Tether	$\varnothing$ 4 mm $\times$ 3 m	—
Obstacle	20 L steel can	$\varnothing$ 306 mm $\times$ 367 mm	Mass 1.25 kg (empty)
Force/torque sensor	Leptrino PFS080YS102U6	$\varnothing$ 80 mm $\times$ 35 mm	Max ∗ $\pm$ 1000 N ( $F_{z}$ ), Resolution 0.25 N, Frequency $\leq$ 1.2 kHz
Constant-force spring	MISUMI CFS 1.4	$\varnothing$ 25.6 mm	Pretension 13.73 N, Stroke 1000 mm

Table I : Hardware specifications.

Figure 4: Experimental conditions: (1) robot velocity, (2) contact angle, (3) side-surface material, and (4) obstacle mass.

Panel of Figure 5 — Figure 5: Representative examples of cable–obstacle interaction behaviors. (A) Rotation and translation (movable): the obstacle rotates (red arrow) and translates (green arrow) on the ground. (B) Rotation only (non-movable): the obstacle rotates in place without translating. (C) Stationary (non-movable): the obstacle stays in place while the cable slides along its side surface.

Figure 6: Behaviors and tension profiles on the bare surface. (A) Interaction behavior in each condition; numbers denote trial counts. Green thick lines mark the movable/non-movable boundary, and red dashed boxes mark the conditions in which velocity changes the behavior. (B) Representative

\Delta F_{z}

profile of the majority behavior in each condition. Time axes cover similar robot travel distances, and green arrows at

\theta=45^{\circ}

and

v=0.10

m/s mark the three phases used for feature design.

Figure 7: Mean tension profiles over three trials for different obstacle side surfaces at

v=0.10

m/s and

\theta=45^{\circ}

. (A) 16.25 kg. (B) 31.25 kg.

Figure 8: Pipeline for movability classification. For each tension signal, we detect the change point

t_{\mathrm{cp}}

and extract three features from each of two post-contact windows with equal duration. The resulting six features are used in a logistic regression classifier to classify the contacted obstacle as movable or non-movable.

Window	No.	Feature name	Notation	Definition	Unit
Early	1	Peak force rise	rise_to_peak	$f_{\mathrm{peak}}-f_{\mathrm{base}}$	N
	2	Normalized peak timing	peak_time_ratio	$t_{\mathrm{peak}}/$ $\Delta t$	–
	3	Post-peak force drop	drop_ratio	$(f_{\mathrm{peak}}-f_{\mathrm{end}})/$ $(f_{\mathrm{peak}}-f_{\mathrm{base}})$	–
Late	4	Late-to-peak force ratio	late_mean_over_peak	$(\bar{F}_{z}^{\mathrm{late}}-f_{\mathrm{base}})/(f_{\mathrm{peak}}-f_{\mathrm{base}})$	–
	5	Late-phase coefficient of variation	wlate_cv	$\sigma_{z}^{\mathrm{late}}/(\bar{F}_{z}^{\mathrm{late}}-f_{\mathrm{base}})$	–
	6	Late-phase zero-crossing density	wlate_zc_detrended	Sign changes per meter after linear detrending	1/m

Table II : Six features extracted from cable tension

F_{z}

. Definitions are relative to the contact onset

t_{\mathrm{cp}}

, where early and late denote two consecutive windows after

t_{\mathrm{cp}}

, each of length

\Delta t=D/v

. The baseline force

f_{\mathrm{base}}

is the mean of

F_{z}

over

[t_{\mathrm{cp}}-\Delta t,t_{\mathrm{cp}})

; the peak force

f_{\mathrm{peak}}

and the peak time

t_{\mathrm{peak}}

are the maximum of

F_{z}

in the early window and its time from

t_{\mathrm{cp}}

; and the end force

f_{\mathrm{end}}

is the last sample of the early window.

Dropped Feature	$D=0.05$ m	$D=0.10$ m	$D=0.15$ m	$D=0.20$ m	$D=0.25$ m	$D=0.30$ m
(post-contact travel $2D$ )	(0.10 m)	(0.20 m)	(0.30 m)	(0.40 m)	(0.50 m)	(0.60 m)
rise_to_peak	$-0.006$	$-0.061$	$-0.053$	$\mathbf{-0.080}$	$\mathbf{-0.044}$	$\mathbf{-0.024}$
peak_time_ratio	$\phantom{+}0.000$	$-0.011$	$+0.007$	$\phantom{+}0.000$	$+0.004$	$\phantom{+}0.000$
drop_ratio	$\phantom{+}0.000$	$\phantom{+}0.000$	$+0.007$	$-0.004$	$\phantom{+}0.000$	$-0.010$
late_mean_over_peak	$\phantom{+}0.000$	$\mathbf{-0.083}$	$\mathbf{-0.076}$	$-0.046$	$-0.021$	$\phantom{+}0.000$
wlate_cv	$\mathbf{-0.068}$	$-0.007$	$-0.014$	$+0.007$	$\phantom{+}0.000$	$\phantom{+}0.000$

Table III : Drop-one feature ablation across prediction distances

D

(stratified 5-fold cross-validation). Each value is the change in mean balanced accuracy when that feature is dropped from the full six-feature set,

\Delta\overline{\mathrm{BA}}=\overline{\mathrm{BA}}_{\mathrm{drop}}-\overline{\mathrm{BA}}_{\mathrm{full}}

; more negative values indicate greater importance. Bold marks the largest negative

\Delta\overline{\mathrm{BA}}

in each column.

Figure 9: Early and late feature-extraction windows for different travel distances

D

v=0.10

m/s, where each window has duration

\Delta t=D/v

. Bold marks the selected

D

Figure 10: Logistic-regression decision slice at

D=0.15

m, with the remaining four features fixed at their dataset medians. The solid line is the decision boundary, and circles denote misclassified trials, numbered as in Section V-D .

Figure 11: Distributions of the six features for movable (Mov.,

n=71

), rotation-only (Rot.,

n=20

), and stationary (Stat.,

n=26

) interactions at

D=0.15

m. The green dashed line marks a late-phase tension equal to the peak rise.

Safe Robot Navigation

Safe robot navigation develops methods for reaching goals while avoiding collisions and managing uncertainty, dynamic obstacles, and environmental hazards.

See all 139 papers

Entanglement-Free Trajectory Planning for Tethered Mobile Robots with a Slack Tether
Aug 10, 2026Gianpietro Battocletti, Dimitris Boskos, Bart De SchutterConstrained Motion Planning Kinodynamic Motion Planning
Delft Center for Systems and Control, Delft University of Technology, Delft, The Netherlands
Topological Motion Planning Diffusion: Generative Tangle-Free Path Planning for Tethered Robots in Obstacle-Rich Environments
Mar 16, 2026Yifu Tian, Xinhang Xu, Thien-Minh Nguyen +1Motion Planning Robot Motion Planning
Trajectory Optimization for Self-Wrap-Aware Cable-Towed Planar Object Manipulation under Implicit Tension Constraints
Mar 10, 2026Yu Li, Amin Fakhari, Hamid SadeghianTrajectory Optimization Constrained Motion Planning
Munich Institute of Robotics and Machine Intelligence, Technical University of Munich, Germany. · The author is with the Department of Mechanical Engineering, Stony Brook University, Stony Brook, NY 11794, USA.

Aug 10, 2026J/K move · Enter open · S save

Entanglement-Free Trajectory Planning for Tethered Mobile Robots with a Slack Tether

Gianpietro Battocletti, Dimitris Boskos, Bart De Schutter

Delft Center for Systems and Control, Delft University of Technology, Delft, The Netherlands

Constrained Motion Planning Kinodynamic Motion Planning Trajectory Generation Safe Motion Planning Robot Motion Planning

Open paper AskPDF arXiv

In motion planning algorithms for tethered mobile robots, the entanglement state of the tether is a critical aspect to consider during the planning phase. This is particularly important in case of a slack tether, where the shape of the tether is not determined solely by the geometry of the environment and the location of the obstacles, but also by the dynamics of the tether, by the trajectory followed by the robot, and possibly by exogenous forces. In this scenario, preventing entanglement requires planning a robot trajectory that accounts for the entanglement definition and for the dynamics of the robot and of the tether. In this work, we propose a motion planning algorithm for tethered mobile robots with a slack tether that computes dynamically feasible entanglement-free trajectories to navigate through an environment with static obstacles. By considering the entanglement state during all the stages of the planning pipeline, we are able to compute safer trajectories that avoid entanglement during the motion of the robot. We achieve this through a three-step pipeline, which includes (i) the construction of a topological model of the entanglement-free configuration space of the tethered robot, (ii) the generation of a set of candidate paths using this model, and (iii) the computation of a dynamically feasible entanglement-free trajectory by solving a homotopy-constrained trajectory generation problem. The resulting trajectory can then be executed to lead the robot to its target location, while maintaining the tether in an entanglement-free configuration. We demonstrate the benefits of this algorithm in simulations, where we show how the planning algorithm avoids violations of the entanglement constraints, resulting in safer and more reliable trajectories.

Abstract

Safe Robot Navigation

Explore similar work

Entanglement-Free Trajectory Planning for Tethered Mobile Robots with a Slack Tether

Topological Motion Planning Diffusion: Generative Tangle-Free Path Planning for Tethered Robots in Obstacle-Rich Environments

Trajectory Optimization for Self-Wrap-Aware Cable-Towed Planar Object Manipulation under Implicit Tension Constraints