Harisankar Babu, Philipp Schillinger, Tamim Asfour
TAPAS is a multi-agent framework that combines language models with symbolic planning to dynamically solve complex tasks without manual environment models.
The TAPAS framework is designed to help computers plan and execute tasks by using multiple agents that communicate with each other. It integrates advanced language models with traditional planning methods, allowing these agents to create and adjust plans without needing a pre-defined model of the environment. This approach enables the system to adapt to new situations and constraints, making it more versatile and capable of handling unexpected changes. TAPAS has been shown to perform well in both standard testing scenarios and in a simulated environment that mimics real-world conditions.
