A modeled welding cable can change the loads calculated at a robot’s joints, even when the cable weighs far less than the robot itself, according to a new preprint. In a planar simulation, including the umbilical produced non-negligible cable-induced torques and substantially changed the robot’s dynamic response.
The study is a deterministic planar simulation with prescribed robot motion, and the authors identify experimental validation as future work.
How the cable is represented
The paper proposes a constrained multibody model of a welding umbilical. The cable is represented as a chain of rigid bodies connected by passive joints. Joint stiffness stands in for elastic bending, while friction represents dissipative effects. Motion at the distal anchor is constrained.
The equations are projected onto the set of physically allowed velocities. That removes explicit constraint-force multipliers from the reduced calculation while preserving the ability to recover the reaction wrench at the anchor. A wrench is the combined force and turning effect acting at a point.
Newton-Euler recursion is used to calculate the system’s inertia matrix and internal force terms. The resulting linear system is solved iteratively with a Krylov-subspace method, and the coupled motion is integrated with a predictor-corrector scheme using a fixed time step of 0.5 milliseconds.
A deliberately narrow test case
For the case study, the simulated umbilical had 10 revolute joints, segments 0.14 metres long and a total length of 1.54 metres. Its linear mass density was 2 kilograms per metre. The model used a bending stiffness of 5 newton-metres per radian and viscous friction of 0.1 newton-metre seconds per radian.
The cable’s distal end was attached to a planar robot with three revolute joints. The robot’s three links were each 0.4 metres long, and its total mass was 36 kilograms. The end effector was prescribed to follow a circle with a radius of 0.3 metres while keeping a constant orientation for five seconds.
The model first computes the umbilical’s joint accelerations and the reaction wrench at its anchor. That wrench is then supplied to the robot’s inverse-dynamics calculation, which estimates the additional joint torques associated with the cable.
Why the torque result matters
The with-versus-without comparison showed non-negligible umbilical-induced torques and a substantial modification of the robot’s dynamic response, even though the umbilical mass was more than one order of magnitude smaller than the robot’s.
The authors say that 10 joints offered a balance between computational cost and the resolution used to represent deformation. They also report that increasing the number of joints did not significantly alter the qualitative results.
What the simulation cannot establish
The planar case considers only in-plane bending. It neglects axial extension, compression and transverse shear, and treats the links as inextensible.
The parameters were chosen for consistency with industrial robot welding umbilicals, but no experimental identification was performed. Experimental validation is listed as future work, so the study does not establish how closely the calculated anchor wrench or torque changes match a real setup.
The authors also identify mechanical-parameter identification, intermediate-link attachments, and mobile or sliding anchors as future work.
The manuscript is a preprint identified as arXiv:2608.25509v1 [cs.RO] in the supplied front matter and dated 26 August 2026. The research was supported by ANRT CIFRE grant n°2023 /1565, which funded the first author’s doctoral studies.
Paper data and sources
Original title: Dynamic Modeling of a Welding Torch Umbilical and Its Impact on Robot Dynamics
Authors: Nicolas Gautier, Yves Guillermit, Mathieu Porez et al.
Journal/Repository: 20th International Symposium on Advances in Robot Kinematics, Jun 2026, Barcelone, Spain
Status: Preprint, not yet peer-reviewed
First online: 2026-08-26
DOI: Not available
Original paper · Full text