Next Article in Journal
Adaptive Backstepping Control with Time-Delay Compensation for MR-Damper-Based Vehicle Seat Suspension
Previous Article in Journal
Design of Experiments Approach for Structural Optimization of Urban Air Mobility Vehicles
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Review

Hybrid Fault-Tolerant Control in Cooperative Robotics: Advances in Resilience and Scalability

by
Claudio Urrea
Electrical Engineering Department, Faculty of Engineering, University of Santiago of Chile, Las Sophoras 165, Estación Central, Santiago 9170020, Chile
Actuators 2025, 14(4), 177; https://doi.org/10.3390/act14040177
Submission received: 12 March 2025 / Revised: 27 March 2025 / Accepted: 2 April 2025 / Published: 4 April 2025
(This article belongs to the Section Actuators for Robotics)

Abstract

:
Cooperative robotics relies on robust fault-tolerant control (FTC) to maintain resilience in dynamic environments, where actuators are pivotal to system reliability. This review synthesizes advancements in hybrid FTC, integrating mechanical redundancy with electronic adaptability and learning-based techniques like deep reinforcement learning and swarm-optimized control, drawing from interdisciplinary evidence across manufacturing, healthcare, agriculture, space exploration, and underwater robotics. It examines how these approaches enhance uptime, precision, and coordination in multi-robot systems, reporting significant improvements despite physical validation being limited to approximately one-quarter of strategies. Addressing gaps in prior work by overcoming limitations of traditional methods, it extends to Construction 5.0, supporting human–robot collaboration (HRC) through scalability and adaptability. Future efforts will prioritize broader testing, standardized benchmarks, safety considerations, and optimization under uncertainty to align theoretical gains with practical outcomes, enhancing resilient automation across domains.

1. Introduction

Cooperative robotic systems, where multiple robots collaborate to achieve collective goals, depend on actuators to translate control commands into physical actions, making their reliability under faults—such as mechanical wear, electronic failures, or environmental stressors—crucial for performance in applications like manufacturing, healthcare, and space exploration. As system complexity grows, advanced fault-tolerant control (FTC) strategies become essential to ensure resilience in unpredictable environments. Unlike prior reviews that narrowly focus on single manipulator fault management [1] or diagnostics [2], and contrasting with analytical approaches like observer-based methods for linear parameter-varying systems (e.g., fault-tolerant observer-based leader-following consensus control, International Journal of Systems Science (IJSS)), which falter with nonlinearity and unpredictability, this review synthesizes advancements in hybrid FTC for cooperative robotics. It integrates passive robustness, active adaptability, and learning-based techniques—such as bioinspired and data-driven strategies that excel in dynamic settings like underwater locomotion, robotic arm control, and gait simulation—addressing a gap in understanding multi-robot resilience across domains. Despite their computational complexity and limited physical validation (approximately 23% [3]), these methods outperform traditional model-based techniques, driven by the superior efficiency, flexibility, and robustness of cooperative robotics over single-robot systems, evidenced by improved task completion rates and dynamic task reassignment [4,5,6]. These benefits shine in fields like Construction 5.0, where human–robot collaboration (HRC) boosts productivity via multi-agent control [7,8], and smart agriculture, where blockchain-enhanced Digital Twins (DTs) optimize harvesting [9]. Yet, achieving these gains demands overcoming challenges in fault tolerance, synchronization, and real-time adaptability [10,11], as seen in manufacturing’s reduced assembly times [12], healthcare’s neural-adaptive precision with a three-layer Recurrent Neural Network (RNN) [13], space exploration’s swarm-based reinforcement learning (RL) [6], and agriculture’s sustainable yields [14]. Bioinspired innovations, like magnetic valve systems with millisecond responses [15], and learning-based FTC also enhance educational outcomes [16]. However, scalability issues—e.g., performance drops in large fleets [4]—and the validation gap underscore the need for this synthesis, which examines actuator redundancy, active recovery, and learning-based responses across diverse applications, highlighting bioinspired potential [15,17,18,19]. This paper contrasts passive and active FTC methods, analyzes redundancy and learning-based innovations, explores hybrid integrations, addresses challenges, and offers practical implications, concluding with insights and future directions for broader validation and standardization, including large-scale fleet testing.

2. Fault-Tolerant Control: State of the Art

FTC is a cornerstone of cooperative robotic systems, ensuring operational continuity amidst component failures, sensor degradation, or environmental disturbances. Unlike traditional robust control, which mitigates bounded uncertainties, FTC embeds resilience mechanisms tailored to the multi-robot coordination demands of applications such as manufacturing, healthcare, and space exploration [2]. This section synthesizes foundational principles and recent advancements in FTC, focusing on passive, active, and hybrid strategies as reported in the literature. While prior reviews, such as Lei et al. [2], emphasize diagnostic techniques, this work broadens the scope to examine how these approaches enhance multi-robot resilience across diverse domains, highlighting their strengths, limitations, and emerging trends.

Key Technical Methods in Hybrid FTC

This subsection outlines key hybrid FTC methods, contrasting their data-driven adaptability with traditional analytical approaches like fault-tolerant observers for linear systems (e.g., IJSS), which rely on predefined models and struggle with nonlinearity. Deep Reinforcement Learning with Central Pattern Generators (DRL-CPG) uses a multi-layer perceptron (MLP) with 256 hidden units, a 0.1 learning rate, a 0.9 discount factor, and Rectified Linear Unit (ReLU) activation, converging via prioritized experience replay to adapt gaits dynamically, as seen in underwater locomotion [17]. The Improved Sparrow Search Algorithm (ISSA) employs a 30-particle swarm with logarithmic Cauchy perturbation and a 0.8 inertia weight, achieving precision through a 20-iteration global search for robotic arm control [18]. Multibody Dynamics-Inching Locomotion Adaptive Robustness (MBD-ILAR) leverages a 12-Degrees-of-Freedom (DoF) model with a 0.3 friction coefficient and velocity constraints, simulating robust gaits at 10 Hz for unstructured terrains [19]. These methods enhance adaptability and uptime beyond analytical baselines, though their computational complexity and limited physical validation—approximately 23% [3]—challenge real-time deployment.
Passive Fault-Tolerant Control (PFTC) relies on robust design to withstand anticipated faults without real-time detection, offering a computationally efficient solution for predictable environments. High-Redundancy Actuation (HRA), for instance, employs parallel actuators with a 2:1 redundancy ratio to maintain functionality under moderate failures [20,21], as seen in manufacturing systems with reported uptime improvements [22]. Sliding-mode-based Lyapunov redesign with a 0.1 sliding gain ensures stability in such setups [23], while bioinspired tensegrity structures with a 0.5 stiffness coefficient enhance energy efficiency in agricultural robots [14]. Similarly, bionic fish designs sustain propulsion under partial failure through flexible fin dynamics [24]. However, PFTC’s static nature limits its adaptability to unforeseen faults, a drawback noted in dynamic scenarios like underwater robotics [25]. In contrast, Active Fault-Tolerant Control (AFTC) dynamically compensates faults using sensor-driven Fault Detection and Diagnosis (FDD), excelling in unpredictable multi-robot contexts. Adaptive FTC with Kalman filters, tuned with a 0.05 noise covariance, achieves fixed-time compensation in Unmanned Aerial Vehicle (UAV) swarms [26], while distributed observers operating at a 10 Hz update rate enable rapid recovery [27]. Neural-adaptive control, utilizing a three-layer RNN [28], enhances resilience across various applications [19]. Despite these strengths, AFTC’s high computational demands—often requiring Graphics Processing Unit (GPU) acceleration—pose challenges for real-time deployment [10].
Hybrid FTC strategies integrate PFTC’s durability with AFTC’s responsiveness, addressing complex fault profiles across cooperative robotics [3,29,30]. This synergy is evident in systems combining mechanical redundancy with FDD-driven adaptation, as illustrated in Figure 1, which contrasts PFTC’s robust design with AFTC’s dynamic management. Recent advancements incorporate bioinspired and learning-based methods, detailed in Section Key Technical Methods in Hybrid FTC, to further enhance versatility. These hybrid approaches have demonstrated applicability in agriculture with improved harvesting efficiency [9], construction with enhanced HRC [12], and healthcare with precise fault management [31], while magnetic valve systems with a 1 kHz actuation frequency improve responsiveness in medical simulations [15]. Table 1 summarizes the trade-offs between PFTC and AFTC, with hybrid strategies offering greater adaptability, though often at the cost of increased complexity. Compared to traditional model-based methods, such as fault-tolerant observers for linear systems (e.g., IJSS), hybrid FTC’s data-driven nature better handles nonlinearity, albeit with higher resource demands.
Table 1 summarizes PFTC and AFTC trade-offs, with hybrid approaches outperforming in versatility, validated by a sub-0.017 s recovery [18] and 98% resilience [17].
Despite these advances, physical validation remains limited, with only about 23% of reported strategies tested beyond simulations [3]. This gap undermines the generalizability of reported outcomes, such as high resilience in swarms [30] or efficiency in manufacturing [12], and contrasts with the extensive simulation-based evidence in the literature. Figure 2 illustrates an HRA architecture with torque limiters, validated in domains like agriculture and medicine with up to 95% efficiency [15,24], while Figure 3 depicts a hybrid system integrating redundancy and neural control. Table 2 compares AFTC methodologies, highlighting neural-adaptive control’s precision (up to 94% accuracy [11]) and gain scheduling’s broad applicability (≥90% accuracy [33]). Looking forward, the field would benefit from standardized benchmarks to validate scalability, particularly for fleets exceeding small sizes, a challenge addressed in Section 6. This synthesis underscores FTC’s evolution from diagnostic-focused approaches [2] to comprehensive, bioinspired solutions, paving the way for resilient multi-robot systems across Construction 5.0, smart agriculture, and beyond [9,14,32].

3. Redundancy Strategies in Cooperative Robotic Systems

Redundancy strategies are critical for sustaining operations in cooperative robotic systems, particularly in large-scale applications where failures can disrupt performance. This section reviews the state of the art in redundancy approaches—mechanical, algorithmic, and hybrid—as reported in the literature, focusing on their role in enhancing resilience across domains like manufacturing, healthcare, and space exploration. Unlike prior studies that address single manipulators [30] or diagnostics [1], this synthesis examines how these strategies collectively tackle multi-robot challenges, such as scalability limitations observed beyond small fleet sizes [4]. The literature highlights applications in HRC for assembly [12], sustainable disassembly [34], and agricultural harvesting [9], extending relevance to fields like Construction 5.0 and smart agriculture [9,14,32]. However, physical validation remains limited, with only a fraction of these approaches—approximately 23%—tested beyond simulations [3], a gap explored further in Section 6.
Mechanical redundancy leverages physical duplication to redistribute loads post-failure, enhancing system continuity. For instance, dual or multiple actuator configurations with a 2:1 redundancy ratio maintain capacity during faults, as demonstrated in manufacturing HRC with reduced downtime [35] and agricultural systems with reliable uptime [9]. Parallel kinematics can increase stiffness by up to 3.5 times in coupled systems [36], while dual-drive mechanisms ensure stable torque output [22,37]. High-redundancy actuators, often bioinspired, retain functionality with a 0.3 failure threshold [3], with tensegrity-enhanced designs using a 0.5 stiffness coefficient improving energy efficiency in corn farming [14]. Modular structures, employing interlocking joints, offer fault tolerance in manufacturing [21] and precise kinematics in continuum robots [32], though integration mismatches can reduce efficiency [29]. These approaches, while effective in predictable settings, face limitations such as a 20% mass increase affecting mobility [3] or complex fabrication needs [14]. Bioinspired enhancements, detailed in Section Key Technical Methods in Hybrid FTC, further bolster uptime, achieving up to 85% in uneven environments, and bionic fish designs sustain resilience under partial failure [24], broadening applicability to underwater systems. Table 3 compares mechanical redundancy strategies, detailing their advantages and limitations across applications, with technical foundations outlined in Section Key Technical Methods in Hybrid FTC.
Algorithmic redundancy enhances fault tolerance through predictive and adaptive control, offering dynamic responses to failures. Model Predictive Control (MPC) predicts states with high accuracy using a five-step horizon, as seen in underwater robots [38], and reduces HRC times [12], though its GPU dependency limits real-time scalability [29]. Fuzzy logic controllers, with a three-rule inference system, minimize performance drops in agriculture [30], yet scalability diminishes beyond 50 robots [39]. Neural-adaptive control, utilizing a 128-node RNN, delivers rapid responses in surgery [11] and UAV swarms [27], achieving high accuracy in construction HRC [40], despite significant computational demands [10]. Reinforcement learning with a 0.1 learning rate stabilizes systems in harvesting [28] and minimizes construction errors [34], though exploration risks persist [13]. Hybrid Artificial Intelligence (AI) methods integrate fuzzy logic, neural networks, and RL with a three-layer fusion model, improving efficiency in multi-agent HRC [8,41], with moderate overhead [42]. Compared to traditional analytical methods like robust control, these data-driven approaches excel in handling nonlinearity, supported by innovations detailed in Section Key Technical Methods in Hybrid FTC. Table 4 summarizes these trade-offs, reflecting their task-specific strengths and computational challenges.
Hybrid redundancy combines mechanical and algorithmic strengths, balancing robustness and adaptability. Integrated fault response systems reroute tasks using redundant pathways, with Variable Stiffness Actuators (VSAs) featuring a 0.2 stiffness range enhancing safety in continuum robots [32] and uptime in construction [7]. Distributed control manages fleets of 5–50 robots with efficiency [4], achieving rapid coordination in UAV swarms [27], though latency increases beyond 50 units [44]. Digital Twins, operating at a 10 Hz simulation rate, reduce testing cycles and improve HRC planning accuracy [45], despite minor synchronization lags [11]. Bioinspired designs further enhance adaptability: soft robotics ensures continuity with compliant materials [20], artificial tendons lower power consumption in agriculture [46], arachnid mechanisms maintain capacity with a six-leg configuration [36], and tensegrity structures with a 0.5 stiffness ratio reduce downtime in farming [14], with potential for energy harvesting via piezoelectric elements [47]. Figure 4 illustrates these hybrid systems, validated in aerial and agricultural contexts [3,14]. Blockchain-based data management with a 256-bit hash complements these approaches by reducing latency in multi-agent systems [38]. While effective, the limited physical validation—around 23% [3]—highlights a need for standardized benchmarks to assess scalability, a topic explored in Section 6.

4. Innovations in Learning-Based Fault Tolerance

Learning-based fault tolerance strategies have emerged as a significant advancement in cooperative robotic systems, offering adaptability to complex dynamics and unexpected faults beyond the capabilities of static controllers. This section reviews the literature on neural-adaptive control and reinforcement learning, synthesizing their theoretical foundations, technical implementations, and applications across domains such as manufacturing, surgery, and education [48]. Unlike earlier reviews that focus on specific robotic implementations, this synthesis examines how these approaches enhance multi-robot resilience, drawing from evidence in construction, agriculture, and space exploration [9,12,40]. The literature highlights their potential to improve fault compensation and recovery, though physical validation remains limited to approximately 23% of reported strategies [3], a challenge explored further in Section 6.
Neural-adaptive systems leverage neural networks to adapt to faults in real time, offering precision in dynamic environments. These systems, unlike analytical observers requiring precise models (e.g., IJSS), leverage neural networks with architectures detailed in Section Key Technical Methods in Hybrid FTC to achieve high accuracy under actuator faults, as demonstrated in multi-joint tracking with up to 94% accuracy [11]. Radial Basis Function (RBF) networks with a 0.01 learning rate approximate uncertainties with Lyapunov stability, excelling in UAV swarms with rapid recovery [10,27]. Compared to traditional static controllers like Proportional-Integral-Derivative (PID), which struggle with nonlinearity, neural-adaptive methods better handle unpredictable conditions, though their reliance on GPU-intensive training limits edge deployment [29]. Applications include surgical precision with dynamic neural models [21], construction HRC with enhanced detection accuracy [39], and space inspection using Probabilistic Differential Quadrature (PDQ) kinematics with a 4th-order solver [32]. Recent advancements, detailed in Section Key Technical Methods in Hybrid FTC, integrate optimization techniques to refine control precision, while magnetic valve systems with a 0.5 T field extend responsiveness to medical simulations [15]. Fault-Tolerant Neural Control Barrier Functions (FT-NCBFs) with a three-layer safety network further ensure diagnostic accuracy, supporting underwater resilience [49].
Reinforcement learning provides a model-free approach to fault tolerance, autonomously optimizing control policies without predefined models. RL typically employs Q-learning with a 0.9 discount factor and a 0.1 learning rate, achieving robust recovery in manipulators [28] and efficient harvesting [9]. As detailed in Section Key Technical Methods in Hybrid FTC, advanced RL techniques accelerate convergence for underwater tasks, while active FTC switches policies with a 0.05 threshold for rapid swarm responses [27]. Unlike analytical observers, which require accurate system models, RL adapts to unstructured settings, as seen in construction HRC with reduced errors [34] and quadruped locomotion over rough terrain with a 6-DoF model [3]. However, exploration risks necessitate extensive validation [13], a concern given the limited physical testing in the field [3]. Enhancements, also outlined in Section Key Technical Methods in Hybrid FTC, improve uptime, and blockchain-based data management with a 256-bit hash mitigates delays in large systems [38]. Figure 5 contrasts neural-adaptive precision with RL adaptability, illustrating their complementary strengths across domains.
The literature underscores the computational feasibility and practical implications of these strategies, with neural-adaptive control excelling in high-stakes precision and RL thriving in flexible, unstructured scenarios. Table 5 compares their trade-offs, showing neural-adaptive systems achieving high accuracy in surgery [11] and RL offering robust recovery in swarms [8]. Innovations detailed in Section Key Technical Methods in Hybrid FTC enhance performance, while Digital Twins with a 10 Hz simulation rate support scalable resilience [38]. Validation remains a key limitation, with only about 23% of strategies physically tested [3], contrasting with simulation-based outcomes like improved harvesting efficiency [9] or construction time savings [12]. Compared to static baselines, these methods offer superior adaptability, yet their computational demands highlight a need for efficient algorithms and standardized benchmarks, as discussed in Section 6. Safety considerations, such as ensuring fail-safe operation in high-risk settings like surgery, and ethical deployment, including equitable access to automation benefits, are critical for practical adoption, particularly in human-centric applications like Construction 5.0. This synthesis bridges theoretical advancements with practical applications, providing a foundation for future research in multi-robot fault tolerance.

5. Hybrid Mechanical-Electronic Systems

Hybrid mechanical–electronic systems integrate physical durability with electronic adaptability to enhance fault tolerance in cooperative robotics, as evidenced across domains like manufacturing, surgery, and space exploration. This section synthesizes the literature on these systems, examining how they combine mechanical redundancy with electronic control to sustain operations under failure. Unlike studies focused on isolated mechanical or electronic approaches [30], this review explores their combined application in multi-robot contexts, drawing from reported outcomes in agriculture, construction, and underwater robotics [9,12,31]. The literature highlights their potential to balance strength and flexibility, though physical validation remains limited to approximately 23% of reported strategies [3], a challenge addressed in Section 6.
The integration of mechanical and electronic components in hybrid systems leverages High-Redundancy Actuation and sensor-driven diagnostics to maintain performance under partial failure. HRA, detailed in Section Key Technical Methods in Hybrid FTC, retains capacity in UAV swarms when paired with sensor redundancy, such as Inertial Measurement Units (IMUs) at 100 Hz and Light Detection and Ranging (LiDAR) for rapid fault detection [3,27]. Tensegrity structures with a 0.5 stiffness coefficient improve uptime in agricultural robots [14], while Probabilistic Differential Quadrature kinematics with a 4th-order solver ensure precision in space inspection [32]. Adaptive collaborative control, using algorithms with a 0.1 gain adjustment, enhances robustness in human–robot collaboration, reducing task times in construction [12,43]. Distributed self-tuning with a 0.05 tuning step further supports multi-agent coordination, minimizing latency in fleets of up to 50 robots [4]. Compared to purely mechanical systems, which lack adaptability, or electronic-only approaches prone to hardware limits, hybrid systems offer a synergistic advantage, as seen in manufacturing with dual-drive redundancy reducing motor power [41] and surgery with efficient laparoscopic outcomes [31]. Figure 6 illustrates this integration, combining HRA, sensors, and blockchain-enhanced data flow with a 256-bit hash for secure coordination [38].
Bioinspired and learning-based innovations enhance the capabilities of hybrid systems across diverse applications. As detailed in Section Key Technical Methods in Hybrid FTC, these methods ensure uptime in rough environments, support resilience in underwater tasks, and refine control precision in manipulators, achieving significant outcomes like up to 85% uptime in agriculture [19]. Magnetic valve systems with a 1 kHz actuation frequency improve responsiveness in medical simulations [15]. In construction, hybrid systems achieve high detection accuracy in HRC [39], while Digital Twins with a 10 Hz simulation rate optimize planning and reduce downtime in automotive manufacturing [37,41]. Swarm RL with a 0.9 discount factor enhances efficiency in space exploration [6], demonstrating the versatility of these approaches. Table 6 summarizes trade-offs, showing mechanical components offering capacity and uptime, electronic elements providing precision, and coordination balancing efficiency with latency challenges [4,35,40].
The benefits of hybrid systems include reliability, flexibility, and longevity, though challenges persist in complexity and validation. Reported outcomes include robust uptime in surgery [50], efficient harvesting [9], and cost-effective construction HRC [34], with modular designs extending wear reduction [29]. However, integration increases costs [41], and neural processing can spike computational demands by up to 300% [10], though blockchain mitigates latency [38]. Physical validation, limited to about 23% [3], contrasts with simulation-based evidence, necessitating a benchmarking framework for standardized testing, as proposed in Section 6, to quantify scalability and resilience trends across domains [29,40]. Compared to isolated approaches, hybrids better address dynamic faults, yet their computational overhead and testing gaps suggest a need for efficient designs and standardized benchmarks, as explored in Section 6. This synthesis underscores the role of hybrid systems in bridging mechanical strength with electronic agility, supporting resilient multi-robot applications in Construction 5.0 and beyond [12,14].
In Figure 7, it can be observed how the hybrid approach enhances efficiency and maintains high levels of availability even under faults, confirming its applicability in various industrial and service domains.

6. Challenges and Future Directions

Fault-tolerant cooperative robotics faces significant barriers in scalability, computational complexity, and real-world validation, as highlighted across the literature. This section synthesizes these challenges and explores emerging research directions to enhance resilience and efficiency in multi-robot systems, drawing from applications in manufacturing, agriculture, and Construction 5.0 [9,12,14]. Unlike prior reviews that focus on specific implementations, this synthesis examines broader trends affecting fault tolerance, integrating evidence of bioinspired and learning-based approaches [15,17,18,19]. The field remains constrained by limited physical validation—approximately 23% of reported strategies [3]—prompting a need for future efforts to bridge simulation and practice.
Scalability poses a critical challenge as multi-robot systems grow beyond small fleets, with efficiency often declining due to coordination overhead and fault propagation [4]. The literature reports performance disruptions in centralized architectures, with efficiency dropping significantly in fleets exceeding 100 robots [4], while decentralized control paradigms reduce latency using distributed tuning, as seen in UAV swarms with rapid recovery [27,44]. Swarm-based and bioinspired approaches enhance connectivity, leveraging distributed coordination to sustain operations, as evidenced in agricultural systems with improved uptime [14]. Techniques detailed in Section Key Technical Methods in Hybrid FTC support resilience in underwater robotics and refine precision in larger fleets [17,18]. Table 7 compares these architectures, showing decentralized and swarm-based systems managing larger scales with moderate to high overhead [44,51].
Figure 8 illustrates these scalability trends, emphasizing advantages beyond small fleet sizes [44]. Addressing scalability requires architectures that balance coordination and autonomy, a direction underscored by Construction 5.0 applications with enhanced human–robot collaboration [12].
Computational complexity escalates with sophisticated fault-tolerant control algorithms, particularly those relying on neural processing, limiting their deployment on resource-constrained platforms [52]. Studies have reported significant demands from learning-based methods, such as neural-adaptive control in manipulators [10], though event-triggered designs with a 0.1 trigger threshold reduce loads in noisy environments [53]. Hardware acceleration, including GPUs with a 2 GHz clock and FPGA with a 100 MHz clock, enables rapid responses in swarms [27,29], while magnetic valve systems with a 0.5 T field optimize energy efficiency in simulations [15]. Blockchain-based data management with a 256-bit hash further mitigates overhead in multi-agent systems [38], as seen in construction applications [12]. Compared to traditional FTC with lower computational needs, these advanced approaches offer adaptability at the cost of increased resource demands, suggesting a future focus on lightweight algorithms and edge computing solutions to enhance practicality across domains.
To bridge this gap, a benchmarking framework is proposed as a standardized platform, integrating multi-layered architectures with hierarchical designs for scalability [29], AI-driven predictive maintenance using three-layer Convolutional Neural Network (CNN) to reduce breakdowns [21], and complementary optimization strategies like quantum-inspired metaheuristics (e.g., Quantum Beetle Antennae Search [53]) efficiently solving constrained optimization problems, paralleling multi-robot coordination challenges [51].

7. Discussion

This section synthesizes theoretical advancements and practical applications of fault-tolerant cooperative robotics as reported in the literature, spanning manufacturing, healthcare, space exploration, and education. The integration of hybrid FTC approaches—combining mechanical redundancy with electronic adaptability—underpins resilience across these domains, as evidenced by improved uptime in manufacturing [41] and precision in surgical robotics [13]. Unlike fragmented studies focused on single aspects, this review examines how these strategies collectively address scalability and validation challenges, extending to applications like Construction 5.0 with enhanced HRC [12,40]. The literature highlights bioinspired and learning-based innovations, though physical validation remains limited to approximately 23% of reported strategies [3], shaping the discussion of current trends and future needs.
Hybrid FTC integrates mechanical and electronic strategies to enhance fault tolerance, with practical outcomes validated across diverse contexts. In underwater systems, robust actuator uptime is achieved using a 6-DoF model [54], while event-triggered control supports rapid recoveries in dynamic environments [55]. Manufacturing benefits from hybrid systems reducing downtime through parallel kinematics and adaptive control [38,41], and healthcare leverages neural synchronization for reliable surgical outcomes [11,13]. Educational applications show improved outcomes through robotic training [48], reflecting the interdisciplinary reach of these approaches. Bioinspired techniques, detailed in Section Key Technical Methods in Hybrid FTC, enhance adaptability in swarms [6] and refine precision in manipulators [18], while supporting uptime in agriculture [8]. These methods outperform isolated approaches by balancing stiffness and flexibility [46]. Construction 5.0 applications demonstrate efficiency gains in HRC [12], though scalability beyond small fleets remains constrained by performance declines [4].
Key challenges persist in scalability, computational complexity, and validation, as noted across the literature. Scalability is limited by coordination overhead in large fleets, with distributed optimization reducing latency in swarms [51], yet efficiency drops remain a concern [4]. Computational demands escalate with neural processing, posing deployment challenges [56], though optimization techniques like ISSA’s adaptive mutation mitigate overhead [18]. The validation gap—only 23% of strategies physically tested [45]—underscores discrepancies between simulation and real-world performance, despite efforts with blockchain-enhanced data flow [37] and terrain-adaptive dynamics [19]. Interdisciplinary integration also faces delays, requiring coordinated efforts across domains. Clarity in performance metrics is another hurdle, with rapid benchmarks improving reliability in construction [12], yet broader standardization is needed. These challenges highlight a tension between theoretical potential and practical deployment, particularly in resource-intensive settings like space exploration [6] and high-stakes healthcare [13].
Future research directions address scalability, efficiency, and validation, incorporating optimization techniques like Non-linear Activated Beetle Antennae Search (NABAS) for robustly solving non-convex tax-aware portfolio problems [57], enhancing resilience under uncertainty [51]. Digital Twins with a 10 Hz simulation rate improve maintenance planning in manufacturing [45], while efficient algorithms enhance energy use in construction [12]. Bioinspired actuators, such as tensegrity designs with a 0.5 stiffness ratio, reduce downtime in agriculture [3], and standardized testing protocols at similar rates increase reliability across domains [50]. Scaling reinforcement learning with a 0.01 exploration rate supports adaptability in large fleets [17], as seen in swarm applications [8]. Figure 9 maps these sectoral impacts, illustrating trends in efficiency and uptime across manufacturing, healthcare, and space exploration [11,13,32].

Safety and Ethical Considerations

Hybrid FTC’s deployment in high-stakes domains like healthcare and Construction 5.0 raises safety and ethical concerns. In surgery, fail-safe operation is critical to prevent harm, with FT-NCBFs ensuring diagnostic accuracy [49]. In construction, equitable access to automation benefits must be balanced against job displacement risks, necessitating transparent protocols. These considerations, integrated with standardized testing [50], ensure safe and ethical adoption across applications.
Table 8 compares key FTC strategies, highlighting their costs and benefits across applications like manufacturing, swarms, and construction [17,35,38]. The literature suggests a shift toward integrated validation frameworks and benchmarks to align simulation-based gains—like those in Construction 5.0 [12]—with physical outcomes, addressing the current 23% validation limit [3]. This synthesis underscores the potential of hybrid FTC to bridge theory and practice, offering a foundation for resilient multi-robot systems across diverse applications.

8. Conclusions and Future Perspectives

This paper consolidates insights on fault-tolerant control in cooperative robotics, emphasizing the role of hybrid systems in enhancing resilience across domains such as manufacturing, agriculture, healthcare, and space exploration. Hybrid approaches that integrate passive robustness, active detection, and learning-based adaptability have demonstrated potential to sustain multi-robot performance under diverse conditions. These systems enable rapid recoveries in industrial settings, consistent uptime in dynamic environments, and efficient coordination in large fleets, reflecting a synergy that extends applicability to fields like Construction 5.0 through improved human–robot collaboration. Despite these advancements, the field faces constraints due to limited physical testing, shaping a discussion that bridges current understanding with future research needs.
Hybrid fault-tolerant control systems combine mechanical redundancy with electronic adaptability to address faults effectively. In manufacturing, dual-drive configurations and parallel kinematics minimize downtime, while event-triggered mechanisms support swift recoveries in manipulators. Agricultural applications benefit from optimized harvesting through neural techniques, and space exploration gains efficiency from reinforcement learning in swarm operations. Construction contexts show reliable performance under varying conditions with modular designs, supported by bioinspired methods like deep reinforcement learning with multi-layer perceptrons for adaptability, optimization algorithms with particle swarms for precision, and multibody dynamics models for uptime in challenging terrains. These approaches outperform isolated strategies by preventing fault cascades and enhancing coordination, offering dependable fault tolerance in healthcare and improved training outcomes in education. Blockchain-enhanced data flow further improves multi-agent efficiency, illustrating how hybrid systems broaden resilience across applications.
Looking forward, scalability, computational complexity, and validation remain key challenges driving research perspectives. Scalability is hindered by coordination overhead in large fleets, where decentralized and swarm-based architectures help maintain performance. Computational demands from advanced algorithms suggest a need for lightweight solutions and hardware acceleration, such as high-frequency processors, to enable practical deployment. The gap between simulation and physical validation underscores the importance of expanded testing to ensure real-world reliability. Future trends emphasize Digital Twins for planning, bioinspired designs like tensegrity structures for uptime, and AI-driven analytics for maintenance, surpassing analytical baselines (e.g., IJSS) through a benchmarking framework that standardizes resilience metrics across domains. Standardized benchmarks could align simulation-based gains with practical outcomes, enhancing adoption across domains like Construction 5.0. These perspectives highlight a path toward integrated frameworks that balance efficiency, resilience, and validation, providing a foundation for advancing cooperative robotics in critical applications.

Funding

This research received no external funding.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

Data are contained within the article. No new experimental data were generated; this review synthesizes existing literature.

Acknowledgments

This work has been supported by the Faculty of Engineering of the Universidad de Santiago de Chile, Chile.

Conflicts of Interest

The author declares no conflicts of interest.

Abbreviations

The following abbreviations are used in this manuscript:
AFTCActive Fault-Tolerant Control
AIArtificial Intelligence
CNN Convolutional Neural Network
CPGCentral Pattern Generators
DoFDegrees of Freedom
DRLDeep Reinforcement Learning
DRL-CPGDeep Reinforcement Learning with Central Pattern Generators
DTsDigital Twins
FDDFault Detection and Diagnosis
FPGAsField Programmable Gate Arrays
FTCFault-Tolerant Control
FT-NCBFsFault-Tolerant Neural Control Barrier Functions
GPUGraphics Processing Unit
HRAHigh-Redundancy Actuation
HRCHuman–Robot Collaboration
IJSSInternational Journal of Systems Science
ILARInching-Locomotion Adaptive Robustness
IMUsInertial Measurement Units
ISSAImproved Sparrow Search Algorithm
LiDARLight Detection and Ranging
MBD-ILARMultibody Dynamics-Inching Locomotion Adaptive Robustness
MLPMulti-Layer Perceptron
MPCModel Predictive Control
NABASNon-linear Activated Beetle Antennae Search
PDQProbabilistic Differential Quadrature
PFTCPassive Fault-Tolerant Control
RBFRadial Basis Function
RLReinforcement Learning
ReLURectified Linear Unit
RNNRecurrent Neural Network
UAVUnmanned Aerial Vehicle
VSAsVariable Stiffness Actuators

References

  1. Milecki, A.; Nowak, P. Review of Fault-Tolerant Control Systems Used in Robotic Manipulators. Appl. Sci. 2023, 13, 2675. [Google Scholar] [CrossRef]
  2. Lei, Y.; Liu, H.; Li, N.; Cao, J.; Qiao, Y.; Wang, H. Condition Monitoring and Fault Diagnosis of Industrial Robots: A Review. Sci. China Technol. Sci. 2025, 68, 1110301. [Google Scholar] [CrossRef]
  3. Iriarte, I.; Gorostiza, J.; Iglesias, I.; Lasa, J.; Calvo-Soraluze, H.; Sierra, B. An Overactuated Aerial Robot Based on Cooperative Quadrotors Attached Through Passive Universal Joints: Modeling, Control and 6-DoF Trajectory Tracking. Robot. Auton. Syst. 2024, 180, 104761. [Google Scholar] [CrossRef]
  4. Cai, W.; Chen, H.; Zhang, M. A Survey on Collaborative Hunting with Robotic Swarm: Key Technologies and Application Scenarios. Neurocomputing 2024, 598, 128008. [Google Scholar] [CrossRef]
  5. Tang, Q.; Ma, L.; Zhao, D.; Sun, Y.; Lei, J.; Wang, Q. A dual-robot cooperative arc welding path planning algorithm based on multi-objective cross-entropy optimization. Robot. Comput. Integr. Manuf. 2024, 89, 102760. [Google Scholar] [CrossRef]
  6. Barth, A.; Ma, O. Cooperative Behavior of a Heterogeneous Robot Team for Planetary Exploration Using Deep Reinforcement Learning. Acta Astronaut. 2024, 214, 689–700. [Google Scholar] [CrossRef]
  7. Dhanda, M.; Rogers, B.A.; Hall, S.; Dekoninck, E.; Dhokia, V. Reviewing Human-Robot Collaboration in Manufacturing: Opportunities and Challenges in the Context of Industry 5.0. Robot. Comput. Integr. Manuf. 2025, 93, 102937. [Google Scholar] [CrossRef]
  8. Gong, J.; Ma, Y.; Jiang, B.; Zhang, Y.; Guo, L. An Adaptive Fault-Tolerant Control Scheme for Heterogeneous Multiagent Systems. IEEE Trans. Syst. Man Cybern. Syst. 2025, 55, 1264–1276. [Google Scholar] [CrossRef]
  9. Tian, Z.; Guo, X.; Ma, W.; Xue, X. Research on Kiwifruit Harvesting Robot Worldwide: A Solution for Sustainable Development of Kiwifruit Industry. Smart Agric. Technol. 2025, 10, 100792. [Google Scholar] [CrossRef]
  10. Wang, Z.; Li, H.; Chen, Z.; Han, Q.-L. A Fault Diagnosis Method for Quadruped Robot Based on Hybrid Deep Neural Networks. IEEE Trans. Ind. Inform. 2024, 20, 1–10. [Google Scholar] [CrossRef]
  11. Le, Q.D.; Yang, E. Adaptive Fault-Tolerant Tracking Control for Multi-Joint Robot Manipulators via Neural Network-Based Synchronization. Sensors 2024, 24, 6837. [Google Scholar] [CrossRef]
  12. Zhao, R.; Tao, S.; Li, P. Safety-Efficiency Integrated Assembly: The Next-Stage Adaptive Task Allocation and Planning Framework for Human–Robot Collaboration. Robot. Comput. Integr. Manuf. 2025, 94, 102942. [Google Scholar] [CrossRef]
  13. Du, W.; Yi, G.; Omisore, O.M.; Duan, W.; Akinyemi, T.O.; Chen, X.; Liu, J.; Lee, B.-G.; Wang, L. Analyzing Surgeon–Robot Cooperative Performance in Robot-Assisted Intravascular Catheterization. IEEE Trans. Hum. Mach. Syst. 2024, 54, 698–710. [Google Scholar] [CrossRef]
  14. Moshayedi, A.J.; Khan, A.S.; Geng, K.; Hu, J.; Kolahdooz, A. Advancing Agricultural Practices: Analyzing the Role of Robotics in Corn Farming. Int. J. Eng. Trans. A Basics 2025, 38, 1517–1532. [Google Scholar] [CrossRef]
  15. Yoo, J.; Chung, G.; Park, Y. Bio-Inspired Miniaturized Magnetic Heart Valve System for Superior Performance. Adv. Mater. 2025, 2419504, in press. [Google Scholar] [CrossRef]
  16. Wang, K.; Sang, G.-Y.; Huang, L.-Z.; Li, S.-H.; Guo, J.-W. The Effectiveness of Educational Robots in Improving Learning Outcomes: A Meta-Analysis. Sustainability 2023, 15, 4637. [Google Scholar] [CrossRef]
  17. Hameed, I.; Chao, X.; Navarro-Alarcon, D.; Jing, X. Deep Reinforcement Learning Enabling a BCFbot to Learn Various Undulatory Patterns. Ocean Eng. 2025, 320, 120322. [Google Scholar] [CrossRef]
  18. Zhang, R.; Choi, H.-S.; Jung, D.; Cho, H.; Anh, P.H.N.; Vu, M.T. Fixed-Time Control with an Improved Sparrow Search Algorithm for Robotic Arm Performance Optimization. Appl. Sci. 2024, 14, 10096. [Google Scholar] [CrossRef]
  19. Cornejo, J.; Sierra, J.E.; Gomez-Gil, F.J.; Gallego, J.A.; Biancardi, C.M.; Weitzenfeld, A. Multibody System Dynamics for Bio-Robotic Design and Simulation Based on Inching-Locomotion Caterpillar’s Gait: MBD-ILAR Method. Bioinspir. Biomim. 2025, 20, 016021. [Google Scholar] [CrossRef]
  20. Huang, Y.; Zhang, S. Cooperative Object Transport by Two Robots Connected with a Ball-String-Ball Structure. IEEE Robot. Autom. Lett. 2024, 9, 4313–4320. [Google Scholar] [CrossRef]
  21. Zhang, Z.; Ma, T.; Zhao, Y.; Yu, S.; Zhou, F. Adaptive Dynamic Programming-Based Multi-Fault Tolerant Control of Reconfigurable Manipulator with Input Constraint. Complex Intell. Syst. 2024, 10, 8341–8353. [Google Scholar] [CrossRef]
  22. Wei, J.; Pan, Y.; Sun, L.; Shang, H.; Chen, X. A Novel Redundant Cooperative Control Strategy for Robotic Pollination. Comput. Electron. Agric. 2024, 220, 108846. [Google Scholar] [CrossRef]
  23. Estrada, M.A.; Fridman, L.; Moreno, J.A. Passive Fault-Tolerant Control via Sliding-Mode-Based Lyapunov Redesign. IEEE Trans. Autom. Control 2024, 69, 6777–6788. [Google Scholar] [CrossRef]
  24. Zhang, D.; Huang, Q.; Xu, H.; Zhu, H.; Cao, Y.; Cao, Y.; Pan, G. Development and Three-Dimensional Modeling of a Bionic Robotic Fish with Pectoral Fins. Ocean Eng. 2025, 321, 120393. [Google Scholar] [CrossRef]
  25. Dong, Z.-Y.; Xinyi, Y.; Wei, Y.; Zhou, L.; Ou, L. A Novel Data-Driven-Based Anti-Disturbance Fault-Tolerant Control Method with Application to Franka-Panda Robot. IEEE Trans. Instrum. Meas. 2024, 73, 3540311. [Google Scholar] [CrossRef]
  26. Mazare, M.; Taghizadeh, M.; Ghaf-Ghanbari, P.; Davoodi, E. Robust Fault Detection and Adaptive Fixed-Time Fault-Tolerant Control for Quadrotor UAVs. Robot. Auton. Syst. 2024, 179, 104747. [Google Scholar] [CrossRef]
  27. Wang, Z.; Wang, T.; Li, T.; Mao, Z. Distributed Observer-Based Close Formation Control for UAV Swarm under Outside Disturbances and Wake Interferences. J. Frankl. Inst. 2024, 361, 106651. [Google Scholar] [CrossRef]
  28. Wang, C.; Yang, L.; Chen, Y.; Lai, G. Adaptive Fault-Tolerant Control for Uncalibrated Camera–Robot System with Multiple Uncertainties. J. Frankl. Inst. 2024, 361, 107048. [Google Scholar] [CrossRef]
  29. Ma, T.; Hu, F.; Su, X.; Shen, C.; Ma, X. Adaptive Neural Cooperative Control of Multirobot Systems with Input Quantization. IEEE Trans. Cybern. 2024, 54, 5518–5528. [Google Scholar] [CrossRef]
  30. Xia, Y.; Liu, Y.; Sun, W. A New Fault Tolerant Strategy Using Adaptive Time Delay Estimation for Robot Manipulators with Actuator Faults. Nonlinear Dyn. 2025, 113, 123–145. [Google Scholar] [CrossRef]
  31. Yamamoto, T.; Fukuda, M.; Yakushigawa, T.; Okuchi, Y.; Taura, K.; Terajima, H. RoboLap Cooperative Technique for Lymphadenectomy in Robotic Surgery for Right-Sided Colon Cancer—A Video Vignette. Color. Dis. 2024, 26, 1779–1781. [Google Scholar] [CrossRef]
  32. Wild, S.; Zeng, T.; Mohammad, A.; Billingham, J.; Axinte, D.; Dong, X. Efficient and Scalable Inverse Kinematics for Continuum Robots. IEEE Robot. Autom. Lett. 2024, 9, 375–381. [Google Scholar] [CrossRef]
  33. Pan, Q.; Ma, B.; An, T.; Dong, B.; Zhu, M.; Li, Y. Design and Experimental Verification: Health Indicator-Based Decentralized Optimal Fault-Tolerant Control for Modular Robot Manipulators via Adjustable Event-Triggered Mechanism. Expert Syst. Appl. 2024, 243, 122944. [Google Scholar] [CrossRef]
  34. Lou, S.; Zhang, Y.; Tan, R.; Lv, C. A Human-Cyber-Physical System Enabled Sequential Disassembly Planning Approach for a Human-Robot Collaboration Cell in Industry 5.0. Robot. Comput. Integr. Manuf. 2024, 87, 102706. [Google Scholar] [CrossRef]
  35. Park, J.H.; Kim, K.; Gong, Y.J.; Yang, S.Y.; Hwang, S.T.; Jung, H.S.; Moon, H.; Koo, J.C.; Rodrigue, H.; Choi, H.R. Cooperative Antagonistic Mechanism Driven by Bidirectional Pneumatic Artificial Muscles for Soft Robotic Joints. Mechatronics 2024, 97, 103099. [Google Scholar] [CrossRef]
  36. Xi, Q.; Chen, Z.; Yin, K.; Liu, X.; Gao, F. Fault-Tolerant Gait for Hexapod Robots with Partial Active Joints Locked. Mech. Mach. Theory 2025, 205, 105895. [Google Scholar] [CrossRef]
  37. Bai, R.; Yang, N.; Qiu, Z.; Johnson, S.; Wu, K.; Li, B.; Chen, G. Achieving High-Quality and Large-Stroke Constant Torque by Axial Force Release. Mech. Mach. Theory 2025, 205, 105869. [Google Scholar] [CrossRef]
  38. Liu, X.; Li, G.; Xiang, F.; Tao, B.; Jiang, G. Blockchain-Based Cloud-Edge Collaborative Data Management for Human-Robot Collaboration Digital Twin System. J. Manuf. Syst. 2024, 77, 228–245. [Google Scholar] [CrossRef]
  39. Guo, J.; Li, Y.; Du, B.; Sun, X.; Wang, K. A Multi-Population Cooperative Coevolution Artificial Bee Colony Algorithm for Partial Multi-Robotic Disassembly Line Balancing Problem Considering Preventive Maintenance Scenarios. Adv. Eng. Inform. 2024, 62, 102750. [Google Scholar] [CrossRef]
  40. Mahmud, K.R.; Wang, L.; Hassan, S.; Zhang, Z. A Knowledge-Driven Framework for Robotic Odor Source Localization Using Large Language Models. Robot. Auton. Syst. 2025, 186, 104915. [Google Scholar] [CrossRef]
  41. Zhang, J.; Jiang, W.; Ge, S.S. Adaptive Fuzzy Fault-Tolerant Control via Integral Terminal Sliding Mode of Robotic Systems with Prescribed Performance. Asian J. Control 2025, 1–11, in press. [Google Scholar] [CrossRef]
  42. Sandini, G.; Sciutti, A.; Morasso, P. Collaborative Robots with Cognitive Capabilities for Industry 4.0 and Beyond. AI 2024, 5, 1858–1869. [Google Scholar] [CrossRef]
  43. Jiang, Y.; Zhang, Q.; Hu, Y. Fault-Tolerant Trajectory Tracking Control of Underwater Salvage Robots Based on Super-Twisting Sliding Mode. Ocean Eng. 2024, 312, 119340. [Google Scholar] [CrossRef]
  44. Ai, X.; Chen, Y.-Y.; Yu, H. Adaptive Fault-Tolerant Formation Tracking Control of Networked Mobile Robots with Input Delays. J. Frankl. Inst. 2024, 361, 248–264. [Google Scholar] [CrossRef]
  45. Lăzăroiu, G.; Gedeon, T.; Valaskova, K.; Vrbka, J.; Šuleř, P.; Zvarikova, K.; Kramarova, K.; Rowland, Z.; Stehel, V.; Gajanova, L.; et al. Cognitive Digital Twin-Based Internet of Robotic Things, Multi-Sensory Extended Reality and Simulation Modeling Technologies, and Generative Artificial Intelligence and Cyber-Physical Manufacturing Systems in the Immersive Industrial Metaverse. Equilibrium. Q. J. Econ. Econ. Policy 2024, 19, 719–748. [Google Scholar] [CrossRef]
  46. Dalla Gasperina, S.; Ferrari, F.; Gandolla, M.; Pedrocchi, A.; Ambrosini, E. Hybrid Cooperative Control of Functional Electrical Stimulation and Robot Assistance for Upper Extremity Rehabilitation. IEEE Trans. Biomed. Eng. 2024, 71, 2642–2650. [Google Scholar] [CrossRef] [PubMed]
  47. Bofill, J.; Abisado, M.; Villaverde, J.; Sampedro, G.A. Exploring Digital Twin-Based Fault Monitoring: Challenges and Opportunities. Sensors 2023, 23, 7087. [Google Scholar] [CrossRef] [PubMed]
  48. Acvi, F. Effects of robotic module-supported experimental activities with the cooperative learning method on student achievement, 21st-century skills, and students’ opinions. J. Comput. Assist. Learn. 2024, 40, 2325–2338. [Google Scholar] [CrossRef]
  49. Song, Z.; Shi, H.; Bai, X.; Li, G. Digital Twin-Assisted Fault Diagnosis System for Robot Joints with Insufficient Data. J. Field Robot. 2023, 40, 258–271. [Google Scholar] [CrossRef]
  50. Yang, Y.; Huang, D.; Ma, L.; Liu, X.; Li, Y. Adaptive Neural Fault-Tolerant Prescribed Performance Control of a Rehabilitation Exoskeleton for Lower Limb Passive Training. ISA Trans. 2024, 151, 143–152. [Google Scholar] [CrossRef]
  51. Carnevale, G.; Mimmo, N.; Notarstefano, G. Nonconvex Distributed Feedback Optimization for Aggregative Cooperative Robotics. Automatica 2024, 167, 111767. [Google Scholar] [CrossRef]
  52. Ebel, H.; Rosenfelder, M.; Eberhard, P. Cooperative Object Transportation with Differential-Drive Mobile Robots: Control and Experimentation. Robot. Auton. Syst. 2024, 173, 104612. [Google Scholar] [CrossRef]
  53. Khan, A.T.; Cao, X.; Li, S.; Hu, B.; Katsikis, V.N. Quantum Beetle Antennae Search: A Novel Technique for the Constrained Portfolio Optimization Problem. Sci. China Inf. Sci. 2021, 64, 152204. [Google Scholar] [CrossRef]
  54. Liu, C.; Filaretov, V.; Zuev, A.; Protsenko, A.; Zhirabok, A. Fault Tolerant Control in Underwater Vehicles. J. Mar. Sci. Eng. 2024, 12, 1836. [Google Scholar] [CrossRef]
  55. Liu, Y.; Wu, J.; Yao, X. Event-Triggered Adaptive Fault-Tolerant Boundary Control for Flexible Bionic Fish Tail with Output Constraint. IEEE Robot. Autom. Lett. 2024, 9, 6384–6391. [Google Scholar] [CrossRef]
  56. Fang, X.; Cheng, R.; Cheng, S.; Fan, Y. Nonsingular Fixed-Time Fault-Tolerant Sliding Mode Control of Robot Manipulator with Disturbance Observer. Int. J. Control Autom. Syst. 2024, 22, 2182–2192. [Google Scholar] [CrossRef]
  57. Khan, A.T.; Cao, X.; Brajevic, I.; Stanimirovic, P.S.; Katsikis, V.N.; Li, S. Non-Linear Activated Beetle Antennae Search: A Novel Technique for Non-Convex Tax-Aware Portfolio Optimization Problem. Expert Syst. Appl. 2022, 197, 116631. [Google Scholar] [CrossRef]
Figure 1. Passive vs. Active FTC diagram—contrasts PFTC’s robust design with AFTC’s dynamic management, enhanced by blockchain, kinematics, DRL-CPG, ISSA, MBD-ILAR, and magnetic control.
Figure 1. Passive vs. Active FTC diagram—contrasts PFTC’s robust design with AFTC’s dynamic management, enhanced by blockchain, kinematics, DRL-CPG, ISSA, MBD-ILAR, and magnetic control.
Actuators 14 00177 g001
Figure 2. HRA architecture—shows parallel actuators (A1–An) with soft elements, MBD-ILAR, and magnetic control.
Figure 2. HRA architecture—shows parallel actuators (A1–An) with soft elements, MBD-ILAR, and magnetic control.
Actuators 14 00177 g002
Figure 3. Hybrid FTC system—integrates redundancy, neural control, blockchain, DRL-CPG, ISSA, MBD-ILAR, and magnetic control.
Figure 3. Hybrid FTC system—integrates redundancy, neural control, blockchain, DRL-CPG, ISSA, MBD-ILAR, and magnetic control.
Actuators 14 00177 g003
Figure 4. Bio-Inspired Hybrid Redundancy—showcases tensegrity, soft robotics, and neural control [3].
Figure 4. Bio-Inspired Hybrid Redundancy—showcases tensegrity, soft robotics, and neural control [3].
Actuators 14 00177 g004
Figure 5. Neural-adaptive vs. RL fault tolerance—compares real-time control with policy optimization, integrating ISSA, DRL-CPG, blockchain, and MBD-ILAR.
Figure 5. Neural-adaptive vs. RL fault tolerance—compares real-time control with policy optimization, integrating ISSA, DRL-CPG, blockchain, and MBD-ILAR.
Actuators 14 00177 g005
Figure 6. Hybrid system diagram—integrates HRA, sensors, blockchain, DRL-CPG, ISSA, MBD-ILAR, and magnetic control, validated across surgery, agriculture, and construction.
Figure 6. Hybrid system diagram—integrates HRA, sensors, blockchain, DRL-CPG, ISSA, MBD-ILAR, and magnetic control, validated across surgery, agriculture, and construction.
Actuators 14 00177 g006
Figure 7. Hybrid performance—compares efficiency and uptime under failure, reflecting trends across domains.
Figure 7. Hybrid performance—compares efficiency and uptime under failure, reflecting trends across domains.
Actuators 14 00177 g007
Figure 8. Scalability comparison—plots efficiency vs. robot count, highlighting trends in decentralized and swarm approaches.
Figure 8. Scalability comparison—plots efficiency vs. robot count, highlighting trends in decentralized and swarm approaches.
Actuators 14 00177 g008
Figure 9. Sectoral impact—charts efficiency and uptime trends across domains.
Figure 9. Sectoral impact—charts efficiency and uptime trends across domains.
Actuators 14 00177 g009
Table 1. Comparative analysis of PFTC and AFTC strategies.
Table 1. Comparative analysis of PFTC and AFTC strategies.
FeaturePFTCAFTC
ComplexityLow computational loadHigh demands due to real-time adaptation
AdaptabilityLimited to predefined faultsDynamic response to unforeseen faults
Implementation CostLower (simpler design)Higher (sensors, algorithms)
Response TimeImmediate, staticSub-0.1 s [27]
Validation85% uptime [19], underwater resilience [24]95% efficiency [32]
Table 2. AFTC performance metrics.
Table 2. AFTC performance metrics.
MethodAccuracyResponse TimeComputational LoadValidation
Neural-Adaptive~94%<50 ms [11], sub-0.017 s [18] High (GPU-accelerated)98% resilience [17]
Gain Scheduling≥90%<100 ms [33], sub-0.1 s [27] Moderate (Field Programmable Gate Arrays
(FPGAs)-based)
98% resilience [8], 95% efficiency [15]
Table 3. Mechanical redundancy strategies.
Table 3. Mechanical redundancy strategies.
StrategyAdvantagesLimitations
Dual Actuators92% capacity [35]Weight, cost, scalability issues [3]
HRA (Tensegrity)70–80% capacity [3], 85% uptime [14]Complex fabrication, energy needs [14]
Modular Structures85% tolerance [21], sub-0.08 mm precision [32]Integration complexity [38]
Table 4. Algorithmic redundancy approaches.
Table 4. Algorithmic redundancy approaches.
ApproachAdvantagesDrawbacks
Model Predictive Control94% accuracy [43]GPU-dependent [29]
Fuzzy Logic91% drop reduction [30], 85% uptime [14]Scalability limits [39]
Neural Adaptive94% accuracy [11]High GPU needs [27]
Reinforcement Learning 90% recovery [28], 98% resilience [17]Safety risks [13]
Hybrid AI95% efficiency [41], 98% resilience [8]20–30% overhead [42]
Table 5. Neural-adaptive vs. reinforcement learning comparison.
Table 5. Neural-adaptive vs. reinforcement learning comparison.
ApproachAdvantagesDrawbacksApplications
Neural-Adaptive94% accuracy [11]GPU-intensive [27]Surgery [13], HRC [40]
Reinforcement Learning90% recovery [28], high resilience [17]Exploration risks [13]Swarms [8], agriculture [9]
Table 6. Multi-agent coordination trade-offs.
Table 6. Multi-agent coordination trade-offs.
AspectAdvantagesChallenges
MechanicalHigh uptime [14], 92% capacity [35]High costs, 20% weight [3]
ElectronicHigh detection accuracy [40]Computational spikes [10]
CoordinationImproved resilience [8]15% latency [44]
Table 7. Control architectures comparison.
Table 7. Control architectures comparison.
ApproachScalabilityOverheadRecoveryComplexity
CentralizedDrops > 100 robots [4]High [39]Moderate [10]Low
DecentralizedUp to 200 agents [44]Moderate [51]Rapid [18]Medium
Swarm-BasedLarge fleets [27]High [44]Rapid [27]Medium–High
Table 8. FTC strategies comparison.
Table 8. FTC strategies comparison.
StrategyCostBenefitApplication
HRAHigh [3]High capacity [35]Manufacturing [7]
RLGPU [28]Improved resilience [17]Swarms [34]
DTsSync [45]Enhanced planning [38]Construction [40]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Urrea, C. Hybrid Fault-Tolerant Control in Cooperative Robotics: Advances in Resilience and Scalability. Actuators 2025, 14, 177. https://doi.org/10.3390/act14040177

AMA Style

Urrea C. Hybrid Fault-Tolerant Control in Cooperative Robotics: Advances in Resilience and Scalability. Actuators. 2025; 14(4):177. https://doi.org/10.3390/act14040177

Chicago/Turabian Style

Urrea, Claudio. 2025. "Hybrid Fault-Tolerant Control in Cooperative Robotics: Advances in Resilience and Scalability" Actuators 14, no. 4: 177. https://doi.org/10.3390/act14040177

APA Style

Urrea, C. (2025). Hybrid Fault-Tolerant Control in Cooperative Robotics: Advances in Resilience and Scalability. Actuators, 14(4), 177. https://doi.org/10.3390/act14040177

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop