Advanced Power Converters and Learning in Diverse Robotic Innovation: A Review

Abstract

This paper provides a comprehensive review of the integration of advanced power management systems and learning techniques in the field of robotics. It identifies the critical roles these areas play in reshaping the capabilities of robotic systems across diverse applications. To begin, it highlights the significance of efficient power usage in modern robotics. The paper explains how advanced power converters effectively control voltage, manage current and shape waveforms, thereby optimizing energy utilization. These converters ensure that robotic components receive the precise voltage levels they require, leading to improved motor performance and enabling precise control over motor behavior. Consequently, this results in extended operational times and increased design flexibility. Furthermore, the review explores the integration of learning approaches, emphasizing their substantial impact on robotic perception, decision-making and autonomy. It discusses the application of techniques such as reinforcement learning, supervised learning and unsupervised learning, showcasing their applications in areas like object recognition, semantic segmentation, sensor fusion and anomaly detection. By utilizing these learning methods, robots become more intelligent, adaptable and capable of autonomous operation across various domains. By examining the interaction between advanced power management and learning integration, this review anticipates a future where robots operate with increased efficiency, adapt to various tasks and drive technological innovation across a wide range of industries.

1. Introduction

Industrial robots have a wide range of applications across various industries, including tasks like handling, painting, assembly, welding and more [1,2,3]. In fields like autonomous vehicles, machine-learning algorithms process sensor data to make real-time decisions, enabling vehicles to navigate and respond intelligently to complex environments [4,5,6]. In healthcare, robots equipped with AI can assist in surgeries, diagnose illnesses and even provide emotional support to patients [7,8,9]. Manufacturing processes are streamlined through AI-driven robotics that optimize production lines and adapt to changing demands [10,11].

Furthermore, in sectors like agriculture, AI-powered drones and robotic systems can monitor and tend to crops with precision, leading to enhanced yields and resource management. Search-and-rescue operations benefit from robots that can analyze disaster scenarios and navigate hazardous environments autonomously [12,13,14,15]. In the service industry, AI-driven robots are employed in customer interactions, from chatbots resolving inquiries to robots providing room service in hotels. Household robotics, such as smart vacuum cleaners, leverage machine learning to map and clean spaces effectively [16,17,18].

The integration of AI and machine learning with robotics also enhance the development of human–robot collaboration. Robots can understand and adapt to human behavior, making them safer and more intuitive to work alongside. The effectiveness and performance of the robot’s manipulative actions in any field of operation primarily depend on the dynamic response of its internal drive system [19,20]. This internal drive system has several impacts on the environment, depending on its design, efficiency and energy source. This concept encompasses the idea of robots being able to adapt, improve and learn from their interactions with their environment, tasks and humans.The control and conversion of electrical energy is responsible for highly efficient robotic systems such as when a robot performs a mix of high-energy and low-energy tasks; learning-based control can allocate energy resources dynamically [21,22,23]. It can optimize power distribution to different actuators based on the priority of tasks. They aim to optimize the efficiency of energy transfer from the power source to the actuators. This is particularly important in robotics, where energy efficiency translates to longer battery life, reduced heat generation and overall better performance [24,25].

Some robots use regenerative braking to capture and store energy generated during deceleration or braking [26,27,28]. Power electronics systems facilitate this process by converting the captured energy into a form that can be stored or reused, thereby increasing energy efficiency. Actuators, like motors, may require different voltage levels and current intensities for different tasks. Power electronics components can regulate these parameters to match the requirements of the specific task the robot is performing [29,30,31]. Another big advancement is use of wireless power transfer in unmanned aerial vehicles as it can transfer the transmission of electrical power from a source to the drone without the need for physical wires or connectors [32,33,34,35]. This technology has the potential to revolutionize the way UAVs are powered and operated, offering benefits such as extended flight times, improved convenience and reduced downtime for recharging.

On the other side, irrespective of the nature of the mission, the primary complexities in the design of multirotor UAVs are shaped by the imperative to optimize payload capacity, maneuverability and flight duration. The capacity to carry payloads and the agility of these UAVs are intertwined with the proportions and mass of their propulsion elements, which directly relate to their power-to-weight ratio. Conversely, the duration of flight hinges on the energy stored and the effectiveness of the propulsion system, encompassing propellers and energy-storage mechanisms [36,37,38,39]. It is widely recognized that efficiency and power density are contrasting prerequisites. To solve this problem, a strong propulsion system has been developed which achieves the high-power density targets for UAVs, while at the same time demonstrating high efficiency and addressing the resiliency and reliability requirements.

In light of all the above developments that are underway, this review addresses the emerging synergy between advanced power management and learning approaches in robotics, highlighting the potential to significantly enhance the capabilities of robotic systems. As industries increasingly rely on robotics, grasping the critical importance of this fusion becomes imperative. Further, with robotics no longer confined to a single industry but spanning across diverse sectors such as manufacturing, healthcare, agriculture and disaster response, this review highlights how these advancements have a far-reaching impact, making it highly relevant to a broad audience. Moreover, in an era of heightened sustainability and environmental consciousness, the optimization of energy utilization in robotics takes center stage. The paper identifies the role of advanced power converters and energy management in promoting energy efficiency—a pressing concern for organizations and governments alike. Lastly, the review explores technologies, including wireless power transfer for UAVs and regenerative braking, addressing how these developments have the potential to disrupt existing paradigms and introduce new possibilities for the field of robotics. In contrast to previous reviews in the field, this paper introduces a fresh perspective on energy harvesting for robotics, emphasizing emerging and innovative solutions that have received limited attention in existing literature. The novel aspects of this review article are as follows:

• The review discusses the integration of various machine-learning and AI methods, including reinforcement learning, supervised learning, unsupervised learning and Bayesian techniques, showcasing the diverse approaches being used to enhance robotic innovation.
• It explores the use of state-of-the-art energy-harvesting technologies, highlighting the latest developments in solar energy for robotic applications.
• The review discusses supercapacitors as a fast-charging alternative to batteries, emphasizing their structural flexibility and the potential for integrating them into robotic systems.
• The review mentions the potential use of polymer electrolyte membrane fuel cells for higher energy density in large robots like UAVs, presenting hydrogen fuel as a promising and economical option for renewable energy in robotics.

The major highlights of this article are as follows:

• Emerging synergy between advanced power management and learning integration in robotics.
• Emphasis on energy efficiency through advanced power converters.
• Learning integration enhancing robotic perception, decision-making and autonomy.

Further, the paper is structured as follows: Section 2 delves into the role of advanced power management systems in robotics, discussing various types of power converters and advancements in power-converter technologies. Section 3 focuses on learning approaches in robotic innovation, covering machine-learning fundamentals, enhancing robotic perception and elevating decision-making processes. Section 4 discusses energy harvesting in robotic applications. Section 5 explores technical advancements observed, emphasizing the influence of integration on robotic applications and highlighting emerging technologies. Finally, Section 6 provides conclusions drawn from the discussions in the paper.

2. Advanced Power Converters in Robotics

Power converters play a pivotal role in robotics by facilitating efficient energy conversion and management [23,40]. In this section, the significance of power converters in robotics is discussed, as well as the various types utilized, recent technological advancements and case studies, highlighting their influence on robotic performance and energy efficiency. Figure 1 illustrates the schematic representation of the robotic system’s partitioning, shedding light on the essential components that collectively enable the robot’s operation.

2.1. Role of Power Converters in Robotics

In the dynamic landscape of robotics, where energy-efficient operation is predominant, power converters serve as the bridge between energy sources and the electrical demands of robotic subsystems [21,41]. The role of power converters encompasses several key aspects:

• Voltage Regulation: Robotic systems incorporate components with diverse voltage requirements. Power converters provide efficient voltage regulation, ensuring that sensors, microcontrollers and other components receive the appropriate voltage levels. This not only prevents potential damage due to overvoltage but also maximizes the efficiency of these components [40,42].
• Current Management: Motors and actuators, crucial for robotic motion and manipulation, often require varying current levels [43]. Power converters enable precise current control, allowing dynamic adjustment to match the specific demands of each task. This adaptability results in optimized motor performance and enhanced control accuracy [44].
• Waveform Shaping: Certain robotic components require specific waveform characteristics, such as sinusoidal signals for AC motors. Power converters facilitate waveform shaping, converting DC power to AC with the desired frequency and amplitude, thereby enabling precise control over motor behavior [45,46].
• Energy Efficiency: By minimizing energy losses during conversion and distribution, power converters contribute significantly to overall energy efficiency in robotic systems. Efficient power conversion reduces wasted energy as heat, leading to prolonged operational times and reduced battery replacements [21,40].

Figure 1. Schematic Representation of Robotic System Partitioning, highlighting the integration of Power Supplies, Converters, Control Box and Sensing Systems for the Robot’s Operation.

Figure 1. Schematic Representation of Robotic System Partitioning, highlighting the integration of Power Supplies, Converters, Control Box and Sensing Systems for the Robot’s Operation.

2.2. Types of Power Converters in Robotics

The complex landscape of robotic applications necessitates a spectrum of power-converter types tailored to specific demands. Figure 2 presents an overview of the power-converter technologies that find applications across various robotic systems.

2.2.1. DC–DC Converters

DC–DC converters are pivotal in robotics, providing voltage step-up or step-down capabilities. For instance, in battery-powered robots, where the available voltage decreases as the battery discharges, DC–DC converters maintain stable voltage levels for critical components. Additionally, as robots integrate sensors and actuators with distinct voltage requirements, these converters ensure compatibility across the system [47].

2.2.2. DC–AC Converters

In robotics, DC–AC converters or inverters play a central role in converting DC power from batteries to AC power for motor-driven systems [48]. Advanced inverter technologies, including advanced modulation schemes like sinusoidal pulse width modulation (SPWM), enable precise control over AC motor characteristics, such as torque, speed and position. This level of control enhances robotic locomotion, manipulation and even aerial operations [49].

2.2.3. AC–DC Converters

AC–DC converters or rectifiers are essential for robotic systems that require power input from alternating current (AC) sources [50]. These converters not only rectify AC power to DC for internal use but also allow robots to draw power directly from AC grids. Charging stations for electric robots, as well as industrial robots operating in environments with readily available AC power, benefit from AC–DC converters.

Figure 2. Power Converters for Diverse Robotic Applications.

Figure 2. Power Converters for Diverse Robotic Applications.

2.2.4. Resonant Converters

Resonant converters utilize resonant components to achieve efficient power conversion with reduced switching losses. This feature makes them valuable for wireless power-transfer applications within robotics, enabling energy delivery without physical connections [51]. Resonant converters are employed in scenarios like charging robotic devices over short distances or even wirelessly powering sensors in remote locations [52,53].

2.2.5. Multi-Level Converters

Robotic systems often demand power converters that can handle high voltages while minimizing harmonic distortion. Multi-level converters achieve this by generating stepped voltage waveforms, reducing stress on components and improving overall system efficiency [54,55]. Applications include high-power robotic arms, where precise motion control requires efficient and high-voltage power conversion.

2.2.6. Soft-Switching Converters

Soft-switching converters focus on reducing switching losses during power conversion. These converters utilize techniques like Zero Voltage Switching (ZVS) and Zero Current Switching (ZCS) to minimize stress on semiconductor devices, leading to improved efficiency and reduced electromagnetic interference (EMI) [56]. In robotics, soft-switching converters find use in high-frequency motor drives and precision robotics that demand minimal energy loss and EMI [57].

2.2.7. Matrix Converters

Matrix converters perform direct AC–AC conversion without intermediate DC links [58]. This feature offers advantages in terms of efficiency, size and reduced components. In robotics, matrix converters can be applied to variable-speed motor drives and actuators [59], facilitating fine-tuned control and efficient power management.

2.2.8. Dual Active Bridge Converters

Dual Active Bridge (DAB) converters provide bidirectional AC–DC conversion, enabling power flow in both directions. This characteristic suits applications where energy regeneration and grid connection are important, such as grid-tied robotics or robots operating in dynamic environments where power needs fluctuate [60].

By understanding and harnessing the capabilities of these advanced power converters, robotics can achieve higher levels of performance, efficiency and adaptability across a wide range of applications.

Table 1. Key Advancements in Power-Converter Technologies for Robotics, along with their Benefits and Applications.

Power Converter	Key Features	Voltage Regulation Range	Efficiency Range	Switching Frequency	Robotics Applications
Buck Converter [61]	Step-down voltage conversion	Narrow to Moderate	High	Medium to High	Battery-powered robots, sensor nodes
Boost Converter [62]	Step-up voltage conversion	Moderate	High	Medium to High	Energy harvesting, charging mobile robots
Buck-Boost Converter [63]	Bidirectional voltage conversion	Wide	High	Medium to High	Battery management, variable power demands
Resonant Converter [52,53]	Zero-voltage switching, reduced EMI	Wide	Moderate to High	Medium to High	Wireless power transfer, contactless charging
Multi-level Converter [54,55]	Reduced harmonics, high voltage capability	Wide	High	Medium to High	High-power robotic arms, electric vehicles
Matrix Converter [58,59]	Bi-directional AC–AC conversion	Wide	High	Medium to High	Variable-speed motor drives, robotic actuators
Soft-Switching Converters [56,57]	Minimal switching losses	Moderate to Wide	High	High	High-frequency motor drives, precision robotics
Dual Active Bridge Converter [60]	Bidirectional AC–DC conversion	Moderate	High	Medium to High	Grid-tied robotics, energy-efficient actuators

provides a comprehensive overview of the key advancements in power-converter technologies for robotics, highlighting the benefits they offer and the specific applications where they find utility. These advancements represent critical milestones in the evolution of robotic systems, enabling enhanced performance, efficiency and adaptability.

2.3. Advancements in Power-Converter Technologies for Robotics

Robotic systems are witnessing transformative impacts due to advancements in power-converter technologies. Table 2 highlights several significant advancements in power-converter technologies for robotics. These advancements are tailored to address the unique challenges and opportunities in robotics.

2.3.1. Integration of Wide-Bandgap Semiconductors

Wide-bandgap materials, notably silicon carbide (SiC) and gallium nitride (GaN), have revolutionized power-converter design. The unique material properties of SiC and GaN enable higher operating temperatures, reduced conduction and switching losses and faster switching speeds [64,65]. In robotics, this translates to increased power-converter efficiency, reduced cooling requirements and improved power density. These benefits are particularly relevant for robots operating in extreme environments, such as industrial automation, space exploration and search-and-rescue missions [66,67].

2.3.2. Enhanced High-Frequency Operation

Robotics often demands compactness and agility. Advancements in high-frequency operation have enabled power converters to operate at frequencies beyond conventional limits [68]. Higher switching frequencies allow for the miniaturization of passive components like inductors and capacitors, resulting in more compact converter designs. This is pivotal in creating lightweight robots that exhibit improved agility, responsiveness and energy efficiency [69,70].

Table 2. Advancements in Power-Converter Technologies for Robotics.

Advancement	Description	Benefits	Applications
GaN Transistors [71,72]	High-efficiency, fast-switching transistors enabling compact power converters.	Reduced power losses, smaller form factors, improved thermal management.	Industrial robots, drones, electric vehicles.
SiC Devices [73]	High-temperature, high-power devices for efficient and reliable converters.	Higher power handling, reduced cooling requirements, better performance in harsh environments.	Electric propulsion, extreme environment robotics.
Digital Power Management [74,75]	Real-time parameter adjustment for adaptable and efficient converters.	Improved adaptability, energy efficiency, remote monitoring.	Mobile robots, medical robots, automation.
Resonant Converter Topologies [76,77,78]	Reduced switching losses, high efficiency, low electromagnetic interference.	Improved efficiency, reduced heat, less EMI.	Renewable energy, wireless charging.
Hybrid and Multilevel Converters [79,80]	Combined topologies for efficiency and voltage control.	Enhanced efficiency, reduced distortion, improved voltage control.	Electric grids, robotic vehicles, renewables.
Advanced Cooling Techniques [81,82,83]	Innovative cooling for efficient operation in confined spaces.	Improved thermal management, higher power handling, compact designs.	High Performance Computing (HPC) clusters, motor drives, confined spaces.
Advanced Control Algorithms [84,85,86]	Precise regulation for changing conditions and loads.	Enhanced accuracy, better response, improved stability.	Prosthetics, haptics, precision control.
Wireless Power Transfer [71,87,88]	Wireless charging for convenience and seamless integration.	Convenience, reduced wear, seamless integration.	Mobile robotics, drones, underwater robots.

Table 2. Advancements in Power-Converter Technologies for Robotics.

Advancement	Description	Benefits	Applications
GaN Transistors [71,72]	High-efficiency, fast-switching transistors enabling compact power converters.	Reduced power losses, smaller form factors, improved thermal management.	Industrial robots, drones, electric vehicles.
SiC Devices [73]	High-temperature, high-power devices for efficient and reliable converters.	Higher power handling, reduced cooling requirements, better performance in harsh environments.	Electric propulsion, extreme environment robotics.
Digital Power Management [74,75]	Real-time parameter adjustment for adaptable and efficient converters.	Improved adaptability, energy efficiency, remote monitoring.	Mobile robots, medical robots, automation.
Resonant Converter Topologies [76,77,78]	Reduced switching losses, high efficiency, low electromagnetic interference.	Improved efficiency, reduced heat, less EMI.	Renewable energy, wireless charging.
Hybrid and Multilevel Converters [79,80]	Combined topologies for efficiency and voltage control.	Enhanced efficiency, reduced distortion, improved voltage control.	Electric grids, robotic vehicles, renewables.
Advanced Cooling Techniques [81,82,83]	Innovative cooling for efficient operation in confined spaces.	Improved thermal management, higher power handling, compact designs.	High Performance Computing (HPC) clusters, motor drives, confined spaces.
Advanced Control Algorithms [84,85,86]	Precise regulation for changing conditions and loads.	Enhanced accuracy, better response, improved stability.	Prosthetics, haptics, precision control.
Wireless Power Transfer [71,87,88]	Wireless charging for convenience and seamless integration.	Convenience, reduced wear, seamless integration.	Mobile robotics, drones, underwater robots.

3. Learning Approaches for Robotic Innovation

In recent years, combining machine learning and artificial intelligence with robots has led to significant progress. This section explores the technical aspects of these learning methods, explaining how they contribute to improving different aspects of robot development [89,90].

3.1. Machine Learning and AI Fundamentals in Robotics

Machine-learning techniques in robotics involve the development of algorithms that enable robots to improve their performance based on data-driven experiences. Reinforcement learning (RL), supervised learning and unsupervised learning serve as the major aspects of these approaches [91,92,93]. RL involves training robots to take actions that maximize a cumulative reward signal. In RL, an agent interacts with an environment and learns to make optimal decisions through trial and error [94]. Techniques such as Q-learning, Deep Q Networks (DQN) [95] and Proximal Policy Optimization (PPO) [96] enable robots to navigate intricate environments and accomplish tasks autonomously. RL algorithms, utilizing exploration and exploitation strategies, have demonstrated exceptional performance in domains like autonomous driving, robotic manipulation and game playing [97,98,99]. Figure 3 provides a visual representation of the profound impact of AI on robotic systems across diverse sectors.

Figure 3. AI in robotic applications across various industries.

Figure 3. AI in robotic applications across various industries.

Further, supervised learning equips robots with the ability to generalize from labeled datasets. Convolutional Neural Networks (CNNs) have proven pivotal for tasks like object recognition and image segmentation. These networks automatically learn hierarchical features from visual data, enabling the identification of objects and scenes [100,101]. Recurrent Neural Networks (RNNs) extend the capabilities of robots to sequential data, enabling tasks like natural language processing and gesture recognition. The supervised learning paradigm, by leveraging large-scale annotated datasets, has led to substantial progress in enabling robots to perceive and interpret their surroundings accurately [102,103,104]. Furthermore, unsupervised learning encompasses diverse techniques that allow robots to uncover patterns and structures within data. Dimensionality-reduction methods, such as Principal Component Analysis (PCA) [105,106] and t-SNE [107], aid in reducing the complexity of sensory data while preserving essential information. Clustering algorithms, like K-means [108,109,110] and Gaussian Mixture Models (GMM) [111,112], facilitate grouping similar data points together. Generative models, including Variational Autoencoders (VAEs) [113,114] and Generative Adversarial Networks (GANs) [115,116], enable robots to learn data distributions and generate novel samples. These unsupervised techniques empower robots to autonomously explore and understand their environments without the need for explicit labels.

3.2. Enhancing Robotic Perception

Robotic perception heavily relies on machine-learning approaches, particularly in the domains of computer vision and sensor fusion. Supervised learning is pivotal for tasks like object recognition, where CNNs excel [117]. CNNs automatically learn hierarchical features from labeled images, enabling robots to detect and classify objects with remarkable accuracy [118,119,120]. Transfer learning, a technique within supervised learning, enables models trained on large datasets (e.g., ImageNet) to be fine-tuned for specific robotic tasks with limited labeled data [121,122,123]. In addition, unsupervised learning techniques play a crucial role in sensor fusion, where data from multiple sensors (such as cameras, LiDAR and IMUs) are combined to create a holistic understanding of the environment [124,125]. By applying unsupervised methods like Independent Component Analysis (ICA) [126,127,128] or Canonical Correlation Analysis (CCA) [129,130], robots can identify relationships between sensor modalities, enhancing their ability to perceive and interpret complex scenes. Figure 4 provides a brief overview of the essential technological clusters that support various aspects of modern robotics.

3.2.1. Object Recognition and Detection

Supervised learning techniques, particularly CNNs, have revolutionized object recognition and detection in robotics. CNNs excel in learning hierarchical features directly from raw pixel data, enabling the extraction of intricate visual patterns [131]. Transfer learning is frequently employed wherein pre-trained CNN models (e.g., ResNet, VGG) on massive image datasets are fine-tuned for specific robotic tasks. This approach allows robots to discern objects even in scenarios with limited labeled data [132,133]. Furthermore, the fusion of object detection frameworks, like Single Shot MultiBox Detector (SSD) [134,135,136] and You Only Look Once (YOLO) [137,138,139], with CNNs enables real-time object localization and tracking. These techniques, driven by machine learning, equip robots to interact with their surroundings intelligently by recognizing and localizing objects of interest.

3.2.2. Semantic Segmentation and Scene Understanding

Semantic segmentation [140], a pixel-level labeling task, is pivotal for robots to understand scene semantics accurately. Deep-learning architectures, such as Fully Convolutional Networks (FCNs) [141,142] and U-Net [143,144], have proven indispensable for semantic segmentation tasks. These networks leverage encoder–decoder architectures with skip connections to capture both local and global contextual information, enabling accurate pixel-wise categorization of objects and regions within a scene. Scene understanding is further enhanced via unsupervised learning methods. By leveraging techniques like Generative Adversarial Networks (GANs) [145,146], robots can perform unsupervised domain adaptation. This enables the model to generalize its perception capabilities from a source domain to a target domain without requiring explicit annotations. Such adaptation ensures robust performance of perception systems in diverse and dynamic environments.

3.2.3. Sensor Fusion for Multi-Modal Perception

Robotics often relies on data from multiple sensors, necessitating robust sensor fusion techniques [147]. Machine-learning approaches, specifically probabilistic methods like Bayesian filtering, play a pivotal role in combining information from various sensors (e.g., cameras, LiDAR, IMUs) to generate a coherent representation of the environment [148,149,150]. Moreover, Kalman Filters and their variants (Extended Kalman Filter, Unscented Kalman Filter) are extensively used for sensor fusion, enabling estimation of the robot’s state and the environment’s structure [151,152,153]. Additionally, Particle Filters provide a non-parametric approach to handle non-linear and non-Gaussian distributions [154,155]. Further, deep-learning techniques have also penetrated sensor fusion, where RNNs [151,156] process sequential sensor data effectively. Long Short-Term Memory (LSTM) [151,157] networks, a type of RNN, are employed to model temporal dependencies in sensor streams, enhancing the robot’s ability to predict dynamic changes in the environment.

3.2.4. Anomaly Detection

Unsupervised learning techniques are harnessed for anomaly detection, allowing robots to identify deviations from expected behavior [158]. One-Class SVMs (Support Vector Machines) [159], Autoencoders [160] and VAEs [161,162] are employed to learn the normal data distribution and subsequently detect anomalies. This is crucial for applications like fault detection in industrial robotics or identifying unfamiliar objects in a scene.

3.3. Elevating Decision-Making Processes

The use of different learning methods, especially reinforcement learning (RL), has brought a big change in how robots make decisions. Reinforcement learning is about training robots to make a series of choices in order to obtain the best overall rewards in their environment. It involves complex algorithms that combine trying out new actions with choosing actions that have worked well before to learn the best ways to do things [163,164,165,166]. Deep Reinforcement Learning (DRL) has become popular for improving robot decision-making by using deep neural networks to handle complicated situations. This is particularly useful in tasks like robotic arm manipulation [92,93,167,168,169,170]. Algorithms like Trust Region Policy Optimization (TRPO) and Soft Actor-Critic (SAC) are good at fine-tuned actions for precise tasks [171,172,173,174,175]. Model Predictive Control (MPC) combined with RL helps robots adjust their actions to deal with unexpected changes in the environment, like in legged locomotion [176,177]. Exploration methods, like Monte Carlo Tree Search (MCTS) and Proximal Policy Optimization (PPO), help robots make good decisions even in complex situations [178,179,180,181,182,183]. Hierarchical RL makes decision-making more efficient by breaking it down into smaller steps. Robots can also learn from human-like behavior and adapt using RL [184,185]. Lastly, handling uncertainty in the environment is important and Bayesian RL helps robots make decisions while considering potential risks [186,187,188,189,190,191,192,193]. All of this has transformed how robots work in different fields, from manipulation to agile movement, making them smarter and more adaptable.

Table 3. Summary of Learning Methods for Robot Decision-Making.

Learning Method	Application	Advantages	Disadvantages	Possible Advancement
Reinforcement Learning	Robot decision-making	Learns from trial and error, Adapts to changing environments	Requires complex algorithms, May converge slowly	Enhanced exploration strategies, Improved sample efficiency
Deep Reinforcement Learning	Robotic arm manipulation	Handles complex situations, Utilizes deep neural networks	High computational cost, Prone to overfitting	Hybrid architectures combining RL and symbolic reasoning, Better regularization techniques
Trust Region Policy Optimization	Precise task execution	Fine-tuned actions for precision, Stable learning	Limited to small-scale problems, Sensitive to hyperparameters	Scalable TRPO variants, Adaptive hyperparameter tuning
Soft Actor-Critic	Fine-tuned actions	Handles precise tasks efficiently, Stable training	Complex to implement, Requires careful tuning	Improved exploration strategies, Real-time implementation
Model Predictive Control combined with RL	Legged locomotion	Adapts to unexpected environmental changes, Predictive control	Computationally intensive, Limited to short planning horizons	Efficient approximations for long planning horizons, Better integration with sensor data
Monte Carlo Tree Search	Complex decision-making	Effective in complex situations, Strong theoretical foundation	Limited to discrete action spaces, Computationally expensive	Hybrid MCTS with RL for continuous action spaces, Parallelization for faster decision-making
Proximal Policy Optimization	Complex decision-making	Stable and straightforward to implement, Good sample efficiency	Can be sensitive to initial conditions, Requires careful hyperparameter tuning	Advanced trust region methods, Adaptive exploration strategies
Hierarchical RL	Efficient decision-making	Breaks down complex tasks into manageable steps, Improved efficiency	Complex to design and train, Hierarchical policies may not generalize well	Better automated hierarchy discovery, Transfer learning between hierarchies
Bayesian RL	Decision-making under uncertainty	Considers uncertainty and risk, Robust decision-making	Requires probabilistic modeling, Computationally demanding	Improved inference algorithms, Incorporation of domain knowledge

offers a concise overview of various learning methods that play a pivotal role in enhancing robot decision-making capabilities. These methods harness the power of AI and machine learning to enable robots to make informed choices and adapt to dynamic environments.

4. Energy Harvesting for Robotic Application

Developments in robotic system over the years have made industrial robots more accessible and user-friendly. However one of the major limitations to the accessibility is related to the self-powered robots, energy storage and conservation between different operating stages [89].

The challenges related to the energy requirement for the operation of the robotics can be easily solved by the energy-harvesting methods. The harvesting can be easily done by generating the power from renewable-energy sources such as thermal, solar and kinetic. Using a renewable alternative to act as a power source for the robotics equipment can lead to a self-powered robotic operation. Hence, it is one of the most innovative research areas for the evolution of robotics [194].

Solar energy is one of the easiest available forms of energy as the global irradiance is somewhere near between 100 to 250 W/m² throughout the year. There has been a lot of development in the solar harvesting as there are studies on thin films which are more convenient and efficient to put on a robotic installation. There are solar cells with dye-sensitized solar cells (DSSCs), copper indium gallium diselenide (CIGS), perovskite solar cells (PSCs) and organic solar cell (OSCs) which are being used widely in the field of robotics [195]. The idea behind the development of different material-based solar cells is to improve the efficiency and reliability of the solar cell along with providing more flexibility to the cell so that it can be installed around the robot [196]. We have covered commercial thin film being placed on a robot to bendable, stretchable and wearable solar film.

Thermoelectrical alternative is also one of the prominent energy-harvesting methods where the change in temperature is used to generates a certain amount of energy [197]. For the generation, p and n of a semiconductor are connected in series electrical and parallel for thermal operation. There are a lot of efforts being devoted to developing a wearable thermoelectrical generation material. The idea is to capture the heat generated during the operation and convert it into energy for further operation, creating a close loop. A wearable thermoelectric unit can generate up to 38 μWcm⁻² from a human body [198]. Innovations have significantly reduced the power requirement by different drivers being implemented, but there still persists a challenge regarding the power sharing between the different drivers as many of the robotic operations are comprised of multiple degrees of freedom during the operation.

Further, developments in battery technology have also impacted the application characteristics of the robots [199]. Previously, the robots needed to have access to power cords for long operational runs or the battery would usually run out faster when the motors operate at their maximum efficiency. There has been a lot of progress in battery technology in the past decade [200]. These changes have resulted in longer-lasting batteries with higher efficiency and robustness. Previously, the batteries used are usually rechargeable and primarily alkaline batteries. Later, for commercial application, batteries such as lead acid, lithium-ion and nickel-metal hydride became popular as they were cheaper for commercial use and can be used in a wide range of operating temperatures [201]. Lead acid batteries presented some challenges with limited life span and low energy density for portable operation. However, few of the issues were overcome by nickel-metal hydride as such batteries have been widely used for autonomous vehicles. Nickel-metal hydride presents advantages with longer durability and faster charging speed. Further, on comparing the nickel-metal hydride battery to lithium-ion, it can be observed that operating temperature still remains an issue as it may cause trouble under extreme conditions. Other than that, issues related to lower specific power and higher operating cost are also present. This forces the modern electric vehicle to switch back to the lithium-ion alternative [202].

Apart from battery technology, there has been a significant development taking place in the field of super capacitors as they provide a fast-charging alternative to batteries [203]. In [204], the authors developed an approach to enhance the performance of Micro Air Vehicles (MAVs) through innovative power-conservation strategies. By employing plasmonics nano-antenna technology, the research had promising results, with a potential mission duration improvement of 16.30 min, offering an insight into the future of MAV capabilities. Predominantly, the supercapacitors can be characterized as either electrochemical pseudo-capacitors or electric double-layer capacitors. The property for charging on the electrical double-layer capacitor is similar to that of the conventional capacitor in which the charge is stored though rapid absorption and desorption of electrons. Because of its unique charge mechanism, the supercapacitors present faster charging and discharging rates [205]. Along with all the advantages, the supercapacitors are deemed to be environmentally friendly when compared to the battery. More then that, because of their stretchability and structural properties, supercapacitors are a great alternative for a wearable robotic system.

Table 4. Energy-Harvesting Methods and Considerations in Robotics.

Method of Energy Harvesting	Description	Advantages and Applications	Challenges and Considerations	Possible Advancements
Solar Energy [89,194,195,196]	Utilizes solar cells (e.g., DSSCs, CIGS, PSCs, OSCs) to convert sunlight into electrical power.	Global availability of sunlight, Suitable for robotic installations, Various material options for flexibility	Efficiency and reliability improvements, Flexible installation options	Enhanced efficiency of solar cells, improved flexibility and advanced installation techniques.
Thermoelectrical Generation [197,198]	Harnesses temperature differentials to generate electrical energy using semiconductor elements.	Harvests heat during operation, Potential for energy capture from human body	Power sharing among multiple drivers, Multi-degree of freedom operation considerations	Development of more efficient thermoelectric materials, advanced power management for multi-driver robots.
Battery Technology [199,200,201,202]	Evolution of rechargeable batteries (e.g., lead acid, lithium-ion, nickel-metal hydride) for improved robot performance.	Longer-lasting batteries, Enhanced efficiency and durability, Various battery chemistries	Issues with limited lifespan (lead acid), Operating temperature limitations (lithium-ion)	Advancements in battery chemistry, increased energy density and improved temperature tolerance.
Super Capacitors [203,205]	Offers fast charging as an alternative to batteries, categorized as electrochemical pseudo-capacitors and double-layer capacitors.	Rapid charge and discharge, Environmentally friendly, Suitable for wearables	Faster charging and discharging, Structural and stretchability advantages	Development of supercapacitors with even faster charging rates and enhanced structural properties.
Polymer Electrolyte Membrane Fuel Cells [206]	Employs fuel cells using chemical properties for energy storage, providing high energy density performance.	High energy density, Potential for green energy transition, Economical hydrogen fuel option	Efficient energy storage, Upscaling potential, Promotes green energy transition	Advances in fuel cell technology for increased energy density, efficiency and scalability.

presents a comprehensive overview of energy-harvesting techniques and the associated considerations that play a pivotal role in powering robotic systems.

It can be deduced from the study that the renewable-energy sources do provide an attractive alternative to the power sources which also results in reduction of size, weight and increase in endurance. The integration of renewables with robot systems accounts for the storage and energy-generation unit. Energy requirements for larger robots such as UAVs or drones will need to have a high energy density power supply for longer operation [206]. Polymer electrolyte membrane fuel cells are mostly preferred for delivering higher energy density performance along with higher energy gravimetrics. Fuel cells use the chemical properties to store the energy instead of wasting heat. Because of the higher efficiency and greater potential to upscale, the fuel cell technology holds the key to the future. The scope of hydrogen fuel is also very economical for the consumer as the green transition for energy generation and storage is promoted by such alternatives.

4.1. Classification of Robots by Power Supply

Robots can be classified into different categories based on their power-supply methods. The choice of power supply is a critical consideration in the design and operation of robots, as it impacts their mobility, energy efficiency and adaptability to specific applications. Here, we categorize robots into four main groups based on their power-supply methods:

4.1.1. Galvanic Contact-Powered Robots

Galvanic contact-powered robots are stationary or fixed in place, receiving power through direct electrical connections, typically via cords or cables. This method offers high energy efficiency as there is minimal energy loss during power transmission. They are commonly used in industrial settings, such as manufacturing lines and CNC machines, where tasks do not require mobility. The advantages of galvanic contact power include high energy efficiency, continuous operation and minimal downtime. However, these robots have limited mobility and range due to their fixed power sources.

4.1.2. Electromagnetic Wave-Powered Robots

Electromagnetic wave-powered robots receive their energy through wireless transfer methods, often using electromagnetic waves like radio waves or microwaves. This approach provides more degrees of freedom for the robot, making it suitable for applications that require mobility, flexibility or remote operation. Drones and autonomous vehicles are examples of robots that often use electromagnetic wave power transfer. While offering enhanced mobility and adaptability to changing environments, this method may have lower energy efficiency compared to galvanic contact. Potential considerations include lower energy efficiency, the risk of interference and limitations in energy-transfer distance.

4.1.3. Battery-Powered Robots

Battery-powered robots are equipped with onboard batteries that store and supply electrical energy. They are highly mobile and versatile, and suitable for a wide range of applications, including drones, robotic vacuum cleaners and autonomous rovers. The key advantages of battery power include mobility, autonomy and the ability to operate in diverse environments. However, they face considerations such as limited battery life, the need for recharging or battery replacement and challenges related to battery weight and size.

4.1.4. Hybrid Systems

Hybrid systems combine multiple power-supply methods to leverage the strengths of each. For example, a robot may have a primary battery power source and employ energy-harvesting techniques or wireless charging to extend operational capabilities. These systems are found in various applications, including medical robots, solar-powered drones and remote environmental monitoring robots. Their advantages include enhanced flexibility, extended operational range and increased energy efficiency. However, they involve complex system integration, potential trade-offs and the need for advanced power management.

4.1.5. Efficiency Considerations

Each power-supply method has its unique advantages and limitations and the choice of power source depends on the specific requirements of the robot’s application. Efficiency considerations play a vital role in this decision-making process. Here are some key efficiency factors to keep in mind:

• Energy Efficiency: Galvanic contact power supply is highly efficient, while electromagnetic wave power-transfer methods may have lower efficiency due to energy losses during wireless transmission.
• Mobility vs. Efficiency: Battery-powered robots offer mobility and autonomy but may have limited operational time between recharges or battery replacements. Electromagnetic wave-powered robots provide mobility but often at the expense of energy efficiency.
• Hybrid Approaches: Hybrid systems allow for a balance between mobility and energy efficiency by combining power-supply methods. For example, a drone with a primary battery source can use solar panels for energy harvesting during flight, extending its operational time.
• Application-Specific Considerations: The choice of power supply should align with the specific requirements of the robot’s intended application, considering factors such as mobility, energy demand and operational environment.

5. Technical Advancements Observed

In this section, the significant effects that come from combining advanced power converters and learning methods in robotic systems are explored. By using these technologies together, researchers and engineers have expanded the possibilities of different robotic applications, bringing in a new era of capabilities and efficiencies.

5.1. Integration’s Influence on Robotic Applications

The convergence of advanced power converters and learning approaches has unleashed a wave of innovation across diverse robotic applications. By seamlessly fusing sophisticated power management techniques with intelligent learning algorithms, robotic systems are becoming increasingly adaptable, efficient and capable of tackling complex tasks [207,208,209]. This section investigates the ways in which this integration impacts several key robotic domains:

• Manufacturing and Automation: The integration of advanced power converters and learning approaches has revolutionized manufacturing processes [210]. Robots equipped with optimized power management systems can dynamically allocate resources, leading to significant energy savings and streamlined operations. Machine-learning algorithms, on the other hand, empower robots to learn from their interactions with the environment, enhancing their ability to handle intricate assembly tasks with precision and adapt to changing production demands [211,212].
• Healthcare and Medical Robotics: The integration has brought about remarkable advancements in medical robotics. Power converters designed for energy efficiency prolong the operation time of medical robots during critical procedures, while machine learning facilitates the development of robots capable of real-time diagnosis and personalized patient care [213,214]. These robots can learn to interpret medical data and collaborate with healthcare professionals, ultimately improving diagnostics, surgical procedures and patient outcomes.
• Agriculture and Environmental Monitoring: In agricultural and environmental settings, the synergy of advanced power converters and learning approaches has led to the creation of autonomous robots that can operate for extended periods in remote locations. These robots harness renewable energy sources through advanced converters, enabling prolonged missions for crop monitoring, soil analysis and wildlife observation [215,216,217]. In [218], the authors highlighted the role of geomatics in Agriculture 4.0, demonstrating the integration of diverse data sources such as satellite imagery, UAVs and autonomous vehicles through advanced data-fusion techniques. By optimizing vineyard management and production with methodologies like Normalized Difference Vegetation Index (NDVI) analysis and sensor-equipped autonomous vehicles, the research offered valuable insights for enhancing precision agriculture in the context of Agriculture 4.0. Further, ML algorithms can empower these robots to navigate challenging terrains, identify anomalies and make informed decisions, contributing to sustainable resource management.
• Search-and-Rescue Missions: The integration plays a pivotal role in enhancing robotic capabilities for search-and-rescue operations. By efficiently managing power resources, robots can operate in disaster-stricken areas for extended durations, maximizing their chances of locating and aiding survivors [219,220]. Learning approaches enable these robots to adapt their search patterns based on evolving conditions and past experiences, significantly increasing their effectiveness in locating and assisting individuals in distress [221,222].

5.2. Emerging Technologies

To provide concrete insights into the transformative potential of integrating advanced power converters and learning approaches, this section highlights the advancements in progress across different domains:

• Smart Factory Optimization: This case study showcases a manufacturing facility that has implemented advanced power converters to optimize energy consumption. By coupling this with machine-learning algorithms, robots within the factory have learned to predict production fluctuations and adapt their energy usage accordingly. The result is a significant reduction in operational costs and improved overall efficiency [223,224,225].
• Minimally Invasive Surgery Assistance: Here, we delve into a medical robotics scenario where power converters with rapid response capabilities enable precise movements of surgical instruments. Paired with learning algorithms, the robot learns to interpret real-time physiological data, adjusting its movements to ensure safe and accurate procedures [226,227]. This integration has led to shorter surgery times and improved patient safety.
• Autonomous Agricultural Monitoring: In the context of agriculture, this case study features autonomous robots powered by renewable energy sources. These robots navigate vast fields, collecting data on crop health and soil conditions. Machine-learning algorithms allow the robots to identify areas requiring special attention and customize their treatment strategies, resulting in higher yields and resource-efficient farming [23,228,229].
• Disaster Recovery with Aerial Robotics: This aspect focuses on the utilization of aerial drones equipped with advanced power systems and learning algorithms in disaster-stricken areas. These drones can fly longer missions, thanks to energy-efficient converters and employ machine learning to rapidly analyze vast amounts of visual and thermal data [230,231,232]. This integration has drastically reduced response times during disaster-recovery efforts.

Table 5. Emerging Technologies Showcasing Integration of Advanced Power Converters and Learning Approaches Across Different Domains.

Emerging Technology	Integration Details	Advancements	Technical Highlights	Innovations
Smart Factory [223,224,225]	Bidirectional converters manage energy based on demand predictions, Robots optimize energy usage via learning.	Reduced costs, efficient production, Adaptive energy allocation.	Real-time power management, Reinforcement learning for optimization.	Agile robotic assembly lines.
Minimally Invasive Surgery [226,227]	High-frequency converters enable precise instrument control, Machine learning adapts movements based on patient data.	Shorter surgery times, improved safety.	Rapid-response power control, Neural networks for patient analysis.	Haptic feedback for surgeon’s perception.
Autonomous Agriculture [23,228,229]	Solar-powered robots monitor crops, Machine-learning processes sensor data for health assessment.	Increased yield, efficient farming.	Photovoltaic energy harvesting, Deep learning for pest detection.	Crop-specific treatment recommendation.
Disaster Recovery with Aerial [230,231,232]	Drones with efficient power systems and learning algorithms assess disasters.	Rapid response, efficient data analysis.	Lightweight power sources for extended flight, Computer vision for disaster assessment.	Collaborative drone swarm coverage.
Ocean Exploration [233]	Underwater robots equipped with power converters and learning algorithms explore ocean environments.	Enhanced data collection, improved navigation.	Advanced underwater power management, Reinforcement learning for underwater navigation.	Real-time analysis of oceanographic data.
Space Exploration [234]	Robots on extraterrestrial missions utilize advanced converters and learning algorithms.	Extended mission duration, autonomous decision-making.	Radiation-resistant power systems, AI for autonomous navigation and exploration.	Self-repair capabilities in extreme conditions.

offers an insight into developments at the intersection of advanced power converters and machine-learning approaches. By examining these aspects and considering their implications across various sectors, it becomes evident that the integration of advanced power converters and learning approaches is a driving force behind the next generation of robotic advancements.

6. Discussion and Future Trends in Robotics

From the above review on various critical aspects of robotics, including the role of advanced power converters, learning approaches and energy-harvesting methods, it is evident that the field of robotics is experiencing significant advancements that are reshaping the landscape of robotic systems. The review on power converters emphasized their pivotal role in robotics. Voltage regulation, current management, waveform shaping and energy efficiency are key aspects of power converters. These capabilities are essential for optimizing the performance and energy efficiency of robotic systems. The trend in power-converter development is towards greater efficiency and adaptability. As robotics applications become more diverse and demanding, the need for advanced power converters that can handle different voltage requirements and provide precise control over current and waveforms is increasing. Additionally, as energy efficiency and sustainability gain importance, power converters are evolving to minimize energy losses and contribute to prolonged operational times. Moreover, various types of power converters used in robotics, from DC–DC converters to matrix converters and soft-switching converters are identified. These diverse converter types cater to different robotic applications and requirements. The trend in this field is towards specialization and innovation in converter design. With the growing demand for robotics in various industries, specialized converters that meet the specific needs of each application are expected to become more prevalent. Further, the review of articles on learning approaches highlighted the crucial role of ML and AI in enhancing robotic perception and decision-making. Reinforcement learning, supervised learning and unsupervised learning are key paradigms in this field. The trend in learning approaches is towards increased autonomy and adaptability. Robots are becoming more capable of perceiving and interpreting their environments, making informed decisions and adapting to dynamic situations. As AI algorithms continue to advance, we can expect further breakthroughs in robotic autonomy and adaptability. Lastly, the review of articles on energy harvesting underlines the importance of energy sources such as solar, thermoelectrical and supercapacitors in powering robotic systems. The trend in energy harvesting is towards sustainability and efficiency. As the world shifts towards greener energy solutions, the integration of renewable energy sources into robotics aligns with environmental and economic considerations. Furthermore, the development of fuel cell technology presents promising opportunities for larger robotic systems like UAVs, offering higher energy density and longer operational times.

6.1. Overall Trends

Driven by the advances in power converters, learning approaches and energy-harvesting methods, some of the key trends can be defined as follows:

• Efficiency and Sustainability: Efficiency and sustainability are at the forefront of robotic system development. Power converters and energy-harvesting methods are being designed to minimize energy wastage and utilize renewable energy sources, contributing to more sustainable and eco-friendly robotics.
• Specialization: With robotics finding applications in a wide range of industries, specialized power converters and learning algorithms are on the rise. These specialized solutions cater to the unique demands of each application, whether it is in healthcare, manufacturing or autonomous vehicles.
• Increased Autonomy: Learning approaches, particularly reinforcement learning and AI, are empowering robots to become more autonomous and adaptable. This trend is particularly evident in fields like autonomous driving, where robots are learning to navigate complex environments with minimal human intervention.
• Integration of Efficient Energy: The integration of energy sources into robotic systems is a growing trend. Solar, thermoelectrical and supercapacitors are increasingly being used to power robots, reducing their reliance on traditional energy sources and contributing to longer operational times.
• Fuel Cell Technology: Fuel cell technology, such as polymer electrolyte membrane fuel cells, holds promise for larger robotic systems. They offer higher energy density, longer operational times and can be a game-changer for applications like UAVs.

6.2. Economic Aspects of Optimal Control in Robotic Applications

The economic aspect of optimal control in robotic applications is a critical consideration that can significantly impact the viability and adoption of robotic systems across various industries. The economic factors to be taken into account when implementing optimal control in robotic applications include initial costs, operational efficiency, maintenance and return on investment.

• Initial Costs: The choice of optimal control structures can have a substantial impact on initial costs. For instance, more complex control algorithms or hardware setups may require a larger upfront investment. However, it is essential to balance initial costs with long-term benefits, such as increased productivity, reduced labor costs and improved product quality.
• Operational Efficiency: Optimal control can enhance operational efficiency by improving accuracy, reducing cycle times and minimizing energy consumption. The economic benefit lies in increased productivity and reduced operating expenses over time. For example, in manufacturing, optimal control can lead to higher throughput, lower scrap rates and energy savings, all of which contribute to cost reduction.
• Maintenance: The choice of control structures can also impact maintenance costs. Complex control systems may require more frequent maintenance and specialized expertise, which can increase operating expenses. Simpler control systems with predictive maintenance capabilities can help reduce downtime and maintenance costs.
• Return on Investment (ROI): The economic feasibility of optimal control largely depends on the ROI it offers. While investing in advanced control structures may have a higher upfront cost, it is crucial to evaluate how quickly these investments will pay off through increased productivity and cost savings. Factors like the expected lifespan of the robotic system, industry-specific demands and potential market growth should be considered when calculating ROI.

Recommendations on the most suitable control structures to be implemented should be made on a case-by-case basis, considering the specific requirements and constraints of the application. However, some general guidelines can be helpful:

• Customization: The optimal control structure should align with the specific needs of the application. Customizing the control system to the unique demands of the task can lead to more efficient and cost-effective solutions.
• Scalability: Consider control structures that can be scaled as needed. This allows for flexibility in adapting to changes in production volume or complexity, ensuring that the investment remains economically viable over time.
• Energy Efficiency: Opt for control structures that prioritize energy efficiency, as this not only reduces operational costs but also aligns with sustainability goals.
• Integration: Ensure that the chosen control structure integrates seamlessly with existing systems and processes. Compatibility can minimize disruptions and reduce integration costs.
• Predictive Maintenance: Implement predictive maintenance capabilities to proactively address issues before they lead to costly downtime. This can extend the lifespan of the robotic system and reduce maintenance expenses.

7. Conclusions

In conclusion, this paper has effectively highlighted the significant impact of advanced power management systems and the integration of learning techniques in the field of robotics. The convergence of efficient power utilization and learning-based control systems has introduced a new era of intelligent, adaptable and innovative robotic systems, revealing notable achievements and innovative research. One of the key aspects identified in the review is the successful implementation of various energy-harvesting methods, such as solar energy, thermoelectrical generation and supercapacitors, in powering robotic systems. These methods have demonstrated their potential to reduce reliance on conventional power sources and increase the endurance and adaptability of robots, making them well-suited for a wide range of applications. Furthermore, the review explored the learning approaches in robotics and how these techniques have transformed the decision-making processes. This has enabled robots to operate autonomously, navigate through intricate environments and interact intelligently with their surroundings, significantly enhancing their capabilities. Further, throughout the review process it is identified that the synergies between advanced power management and learning integration for robotic systems are expected to further revolutionize various sectors, such as industrial automation, healthcare, agriculture and disaster response. The integration of wireless power transfer technology in UAVs is foreseen to hold extended flight times and increased operational efficiency, opening new horizons in aerial robotics. In this developing landscape, the interaction between advancements in the fields of power management, machine learning and robotics will be crucial in realizing the full potential of robotic systems.