Reliable transfer of control policies from simulation to real-world robotic systems remains a central challenge in robotics, particularly for car-like mobile robots. Digital Twin (DT) technology provides a robust framework for high-fidelity replication of physical platforms and bi-directional synchronization between virtual and real environments. In this study, a DT-based testbed is developed to train and evaluate an imitation learning (IL) control framework in which a neural network policy learns to replicate the behavior of a hybrid Model Predictive Control (MPC)–Backstepping expert controller. The DT framework ensures consistent benchmarking between simulated and physical execution, supporting a structured and safe process for policy validation and deployment. Experimental analysis demonstrates that the learned policy effectively reproduces expert behavior, achieving bounded trajectory-tracking errors and stable performance across simulation and real-world tests. The results confirm that DT-enabled IL provides a viable pathway for Sim2Real transfer, accelerating controller development and deployment in autonomous mobile robotics.
Path planning in cluttered environments constitutes a critical challenge for mobile robotics. Although optimal solutions can be obtained by classical methods such as A*, they have the disadvantage of being computationally expensive in complex environments. In this paper, we propo…
Energy consumption optimisation has emerged as a critical need in Autonomous Mobile Robots (AMRs). Conventional A* implementations typically minimise path distance, neglecting energy-relevant factors such as directional changes and trajectory smoothness that significantly impact…
Localization and mapping remain critical challenges for Unmanned Ground Vehicles (UGVs) operating in unstructured natural environments, such as forests and agricultural fields. While Visual SLAM (VSLAM) and Visual–Inertial SLAM (VI-SLAM) have matured significantly in structured a…
This study presents an assessment scheme for haptic interaction systems based on Hamiltonian energy prediction, which contributes to procedures applied to neurorehabilitation. It focuses on robotic systems involving human participation in the control loop, where uncertainty may c…
Supervised training of neural networks is time consuming, and the scenarios required for obtaining a representative dataset must be carefully considered for each task. Applying an unsupervised training approach can greatly simplify this data collection aspect. This paper explores…
Packing food into lunch boxes requires the correct portion to be selected. Food items such as fried chicken, eggs, and sausages are straightforward to manipulate when packing. In contrast, deformable objects like spaghetti can give challenges to lunch box packing due to their fra…