A preprint reports a mathematical route toward stable bilateral teleoperation—the two-way control of a master robot and a remote slave robot—when communication between them is delayed and the remote arm sends back direct contact-force information. The central result is conditional: under the theorem’s assumptions and the required communication-parameter inequalities, every closed-loop signal remains bounded and the master and slave velocities converge to zero.
That does not amount to a demonstration on a working robot. The study combines analysis of nonlinear multi-degree-of-freedom models with numerical simulation; it reports no human-operator trial or physical-robot experiment. Its stated target is a link that can preserve stability, position synchronization in free motion and force synchronization during hard contact despite constant communication delays.
A stability result with conditions
The proposed link uses a vector-valued, upper strictly passive wave-based communication law. In the paper’s control framework, passivity is an energy-balance property: the law is intended to compensate for the passivity shortage identified in the remote subsystem. An LMI, short for linear matrix inequality, provides the matrix-based check used to characterize that shortage and guide the communication-law design.
The theorem’s stability condition is not a blanket guarantee. It requires the assumptions stated for the model and communication law, including γl ≤ 0 and γr < −α. Within those conditions, the analysis says all signals stay bounded and both robot velocities approach zero. The authors also describe the result as independent of the size of a constant delay: larger delays are said to reduce transparency and worsen transient behavior without changing stability.
Stable does not mean perfectly matched
Position and force synchronization are more qualified than the stability result alone might suggest. In free motion, the theorem permits the master and slave positions to settle with a constant error inherited from their initial conditions. Exact asymptotic position matching requires the initial positions to match as specified by the theorem. In other words, velocities ending at zero does not alone ensure that the two robots occupy the same position.
For hard contact, the analytical result is different. If the environment stiffness is continuously differentiable and the master acceleration tends to zero, the force-tracking error tends to zero. That is a model-based asymptotic conclusion, not a measurement taken from a physical contact experiment. The supplied analysis also notes that the paper’s delay claim was not backed by a separately reported controlled comparison across delay settings.
The evidence comes from models
For its numerical illustration, the paper used the same nonlinear two-degree-of-freedom planar manipulator structure for the master and slave. The simulation therefore asks whether the proposed law behaves as predicted in a defined computer model; it does not sample different robots, operators or hardware conditions. The analytical and numerical pieces are connected through the passivity-shortage calculation and the communication-law design.
In the grid-based LMI calculation, the authors set λ to 0.001 and obtained α = 5.7709. They call this an approximate certificate because checking grid points does not necessarily establish the inequality over the full set of model conditions. That qualification matters: the number is evidence for the reported calculation, not an uncertainty interval or a guarantee outside the checked region.
The communication settings were then tuned with a simulation-based search whose cost put zero weight on position error, with wq = 0, and full weight on force error, with wf = 1. Each optimization iteration used a 120-second closed-loop simulation. The reported selection was b = 0.06 with γl = −20, producing γr = −69.4444 within the stated search bounds.
What the computer test showed
In that modeled scenario, the closed-loop position trajectories remained stable. During hard contact, position errors approached constant offsets, while the slave’s velocity magnitude stayed below 0.06 m/s. Those observations fit the theorem’s distinction between bounded motion and exact position matching: stable behavior did not imply exact agreement of the positions.
The reported root-mean-square force-tracking error—the study’s summary of force mismatch over the run—was 1.2993 N over the simulation horizon. No uncertainty interval, replicate count or comparator value is reported for that figure. The abstract separately reports improved transparency against classical lossless wave-transformation schemes, but the supplied numerical results do not give a separately tabulated controlled comparison.
A framework, not a field test
The paper is marked arXiv:2608.20043v1 and dated 20 Aug 2026. It is therefore a preprint. Its evidence remains analytical and simulation-based: no human users or physical robotic platforms were evaluated.
The guarantees also depend on assumptions built into the theorems, including bounded trajectories, regular model matrices, damping and bounded initial histories. The LMI result was checked on a grid rather than necessarily across the full parameter-and-state set. The free-motion theorem requires an initial-position match for exact asymptotic synchronization, while the hard-contact simulation retained constant position offsets. The work does not establish human usability, physical-robot performance or real-world haptic realism.
The supplied analysis leaves several practical questions open. It asks whether the framework can extend to master and slave systems with different dimensions, whether richer neural-network-inspired communication channels can transmit additional information such as velocities, and whether the method performs similarly on physical teleoperation platforms with human operators. A matched, reproducible comparison would also be needed to assess the reported transparency improvement against classical lossless wave-transformation schemes.
Paper data and sources
Original title: Wave-Based Bilateral Teleoperation between Nonlinear Manipulators with Direct Contact Force Feedback
Authors: G. Q. Bao Tran, Takanori Miyoshi, Ho Duc Tho
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-20
DOI: Not available
Original paper · Full text