Capability Based

Capability-based approaches aim to formally define and quantify the abilities of agents, whether AI models, robots, or human-robot teams, to improve their performance and interaction. Current research focuses on developing frameworks for representing and utilizing these capabilities, including ontological models for robots, capability-aware task allocation algorithms for multi-robot systems, and methods for enhancing large language models' tool-use abilities without sacrificing general performance. This work is significant for enabling more effective and ethical AI systems, improving human-robot collaboration, and facilitating the development of robust and adaptable autonomous systems across various domains.