Candidate: Ramisha Anjum
Date: Nov 27, 2024
Time: 3:00pm
Location: EIT 3145
Supervisor: Gennaro Notomista
All are welcome!
Abstract:
This thesis presents a novel optimization-based control framework for multi-robot systems interacting with human. This framework is amenable for both centralized and decentralized control strategies and addresses human-robot interaction as well as multi-task allocation and execution. The framework leverages optimization algorithms to manage task allocation while ensuring system passivity, which is critical for maintaining stability and safe interaction between robots and human operators. Both single-integrator and double-integrator robot dynamic models are explored for different control scenarios. While the single-integrator scenario employs decentralized control, ensuring scalability and efficient task execution, the double-integrator scenario typically utilizes centralized control. However, if human-robot interaction constraints are distributed across robots, a decentralized approach can be formulated, allowing for flexible control structures.
Experimental results demonstrate the system’s ability to dynamically allocate tasks, execute them efficiently, and provide optimal haptic feedback to human operators. The human-robot interaction mechanism combines passivity-based control with feedback synthesis, ensuring that the system can adapt to real-time changes in human input while maintaining stability and safety. Additionally, the thesis identifies the limitations of a purely centralized approach, particularly in larger robot teams, and suggests future improvements, including enhanced scalability, improved feedback mechanisms, and adaptations for dynamic, real-world environments.
This research advances the field of human-robot collaboration by offering a robust, adaptable framework capable of handling complex multi-task scenarios. The framework is particularly suited for applications in dynamic and interactive settings, such as telerobotics, industrial automation, and other collaborative operations where human oversight and real-time control play a crucial role