Dependable Humanoid Robots: Fault Diagnosis, Fault-Tolerant Control and Safe Physical Interaction

Humanoid robots are increasingly expected to operate outside controlled laboratory conditions, where sensor degradation, actuator faults, calibration drift, unexpected contact and changing environments can compromise stability and safety. A humanoid platform combines many actuated joints with force and torque sensors, inertial units, cameras and complex drive electronics. A fault in any of these components can propagate through the system, leading to degraded motion, loss of balance, unsafe contact forces or a fall. These risks are especially important when humanoids operate close to people or interact physically with their surroundings.Many methods for fault diagnosis, condition monitoring and fault-tolerant control have been developed for electrical drives, industrial machinery and other mechatronic systems. However, their transfer to floating-base, contact-rich humanoid robots remains challenging. Humanoid systems combine high-dimensional dynamics, changing contacts, coupled joints, uncertain sensing and strict real-time constraints. This creates a clear research gap: how can a humanoid robot detect and interpret degradation, adapt its behaviour after a fault, and continue operating safely rather than failing abruptly?This Research Topic will examine methods for improving the dependability of humanoid robots across the complete fault-management cycle: detection, diagnosis, decision-making, adaptation and recovery. It will bring together data-driven, model-based and hybrid approaches for monitoring joints, actuators, drives and sensors; identifying the source and severity of faults; estimating the robot’s state under uncertainty; and adapting control or interaction strategies to preserve stability and safety. Particular attention will be given to explainable diagnosis, robust perception and sensor fusion, graceful degradation, and the transfer of established fault-tolerant methods from industrial drives and mechatronic systems to humanoid platforms.The topic will also consider how learning-based and hybrid control methods can support post-fault locomotion, balance recovery and manipulation, provided that the work addresses realistic degradation, uncertainty or physical safety. Contributions should be grounded in humanoid robots or closely related legged and human-centred robotic platforms through hardware experiments, physics-based simulation, fault-injection studies or reproducible benchmark scenarios. Submissions should clearly report how they evaluate diagnostic accuracy, detection latency, false alarms, recovery performance, post-fault stability, interaction safety or generalisation across operating conditions.To maintain a clear common thread, this Research Topic focuses on the dependability of humanoid robots and related mechatronic systems. Generic machine learning, computer vision or fault-diagnosis studies that are not connected to humanoid or legged embodiment are outside the scope. Likewise, humanoid control studies that do not address faults, degradation, uncertainty, robustness or physical safety are not the primary focus.We welcome articles addressing, but not limited to, the following themes: • Fault detection, isolation and diagnosis in humanoid joints, actuators, drives and transmissions •Condition monitoring, degradation modelling and predictive maintenance for humanoid platforms •Sensor faults, calibration drift, actuator degradation and robust state estimation •Explainable and interpretable diagnosis for humanoid condition assessment •Uncertainty-aware perception and sensor fusion for reliable humanoid operation •Safe physical human–robot interaction, contact-force regulation and collision handling under abnormal conditions •Mechatronic and kinematic design for fault tolerance, compliance and graceful degradation •Learning-based, model-based and hybrid methods for fault-aware humanoid control •Fault-tolerant whole-body control, balance recovery and locomotion under partial actuation or sensing loss •Post-fault manipulation, recovery behaviours and adaptive task execution •Simulation frameworks, fault-injection benchmarks, datasets and reproducible evaluation protocols •Transfer and adaptation of fault-diagnosis and fault-tolerant control methods from industrial drives and mechatronic systems to humanoid robots •Hardware validation, long-term reliability and deployment of dependable humanoid systemsWe welcome original research articles, review papers, mini reviews, methods papers, perspective articles and technology reports presenting theoretical developments, diagnostic and control algorithms, simulation frameworks, benchmark resources, mechatronic designs and experimentally validated applications. Topic editors: Emrah ASLAN, Aysegul Ucar, Sergiu-Dan Stan, Yıldırım ÖZÜPAK Source: https://www.frontiersin.org/research-topics/85260/dependable-humanoid-robots-fault-diagnosis-fault-tolerant-control-and-safe-physical-interaction