Work

Digital-Twin-Enabled Synchronous Teleoperation of Robotic Arms Using BiLSTM-KF Hybrid Model for Time Delay Compensation

Year: 2026
Type: article
Abstract: Communication delay in digital-twin teleoperation causes temporal mismatch between local virtual arm motion feedback and remote physical arm execution, particularly under time-varying public-network conditions. This study proposes a prediction-informed synchronous framework that combines one-step-ahead Bidirectional Long Short-Term Memory (BiLSTM) delay prediction with recursive Kalman Filter (KF) correction. Historical delay measurements are used to predict the subsequent communication delay, and the corrected estimate is incorporated into a feedforward speed-compensation law for the local virtual arm without modifying the low-level controller of the remote physical arm. The framework therefore adopts corrected delay estimation directly to virtual-arm motion adaptation within the digital-twin teleoperation system. On the evaluating test sequence, the BiLSTM-KF model achieved an RMSE of 5.6498 ms, an MAE of 3.5455 ms, and an MAPE of 7.84%. In teleoperation experiments, BiLSTM-KF-based compensation yielded an average normalized position deviation rate of 2.427 ± 0.569%, corresponding to reductions of 53.63% relative to no compensation, 46.42% relative to historical-mean delay compensation and 35.83% relative to LSTM-based compensation. These results indicate that prediction-informed virtual-side adaptation can reduce virtual–physical mismatch under the tested network and motion conditions.
Source: Symmetry
Language: English
Open Access status: gold
Estimated APC paid: $2,896
OpenAlex ID: W7235308726
FWCI:
Cites: 20
Cited by:
Related to: 10