Research

Cooperative Control of Multi-agent Systems

Cooperative control of the multi-agent system (MAS) is gaining interest in the research community due to its advantageous applications over a single agent system. The problem of cooperative control has been explored to achieve rendezvous, flocking, formation, attitude synchronization, containment, and target-capturing. In a cooperative mission, agents having different sensing and processing capabilities reach an agreement by interacting with the neighbors. Cooperative algorithms may have either centralized or distributed structures to arrive at the desired objective. My research work primarily focuses on the controller design to achieve common group objectives such as target tracking, consensus, formation, and containment. The aim is to design a controller using complete or incomplete state measurements to reduce the requirement of multiple sensors.

Cooperative Localization using Multiple Unicycles

A group of UAVs is used in different civilian and defense missions, such as fire monitoring, area coverage, search and rescue, and surveillance. During these operations, UAVs need to follow or orbit a specific location or target of their interest. In the case of an unknown target or location, it is essential to localize these unknown states from different measurements. A group of UAVs carrying different sensors can efficiently obtain the information of an unknown target using cooperative localization techniques. Efficacy of localization gets enhanced if the agents form an optimal formation geometry around the target. My research focuses on designing a unified framework of cooperative localization for a group of unicycles that maximizes the information of the target.

Distributed Average Tracking (DAT)

In a DAT problem, the objective is to track the average of multiple references that are possibly time-varying. Each of these references is accessible to only one agent. DAT algorithms have been applied in different distributed estimation methods, such as fusion of multiple sensors, distributed Kalman filter, and merging of feature-based maps. Moreover, such algorithms are also used for controller design, where agents’ physical states have to follow the time-varying references to achieve distributed convex optimization, region following formation, and distributed state coordination. Since a DAT algorithm addresses the convergence of multiple time-varying references, the consensus and distributed tracking algorithms can be considered special cases. My research objective is to develop novel DAT algorithms that work under communication and sensing constraints.

Soft Robotics

Soft robotics is revolutionizing automation by introducing adaptable, flexible systems that interact seamlessly with humans and delicate objects. Our research focuses on developing intelligent soft robotic systems with advanced control, real-time adaptability, and enhanced functionality. We specialize in designing and optimizing soft robotic fingers, integrating flex sensors, servo-driven actuation, and microcontroller-based automation. These innovations contribute to advancements in human-robot interaction, biomedical assistance, and precision handling. By implementing real-time feedback mechanisms, we enhance accuracy and efficiency, enabling soft robots to perform complex tasks with precision. Additionally, we work on exploring cost-effective fabrication techniques and material enhancements to improve durability and responsiveness. These developments aim to make soft robotics safer, more adaptable, and suitable for diverse real-world applications, from assistive healthcare to industrial automation.

Autonomous Navigation and Path Planning

Research in autonomous vehicles focuses on improving navigation through data-driven learning and advanced control techniques. Deep reinforcement learning (DRL) enables autonomous systems to make adaptive, well-ordered control decisions in dynamic, multi-agent environments. Meanwhile, model predictive control (MPC) optimizes real-time decision-making by predicting future states and solving constrained optimization problems. The integration of DRL and MPC enhances the accuracy, flexibility, and robustness of autonomous systems, allowing them to navigate complex, uncertain environments effectively. These advancements have significant implications for real-world applications, including urban mobility, logistics, and industrial automation. By addressing key challenges in autonomous navigation, this research paves the way for safer, more efficient, and intelligent autonomous vehicles capable of operating seamlessly in diverse and dynamic conditions.

Control Systems Design and Applications in Power Systems

Control systems are critical in effectively managing renewable energy-based microgrids, ensuring stability, reliability, and efficient energy distribution. Microgrids are becoming essential for localized energy management with the growing integration of solar, wind, and other renewable energy sources. However, renewable sources' inherent intermittency and unpredictability pose challenges to grid stability. Multi-agent systems (MAS) have emerged as a promising solution to address these challenges. MAS consists of autonomous agents that collaborate, communicate, and make decentralized decisions, enhancing the resilience and flexibility of microgrids. These agents can represent various components, such as energy storage systems, distributed generation units, and loads. They work together to balance energy supply and demand, optimize power flow, and ensure grid stability. Advanced control strategies, including model predictive control, fuzzy logic, and reinforcement learning, are often employed in MAS-based microgrids. These methods enable adaptive decision-making and real-time responses to dynamic grid conditions. Furthermore, MAS can facilitate peer-to-peer energy trading, enhancing economic benefits for prosumers.

Report abuse