Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper introduces an agentic Model Predictive Control (MPC) framework that integrates large language model-based agents to enhance control synthesis by incorporating high-level contextual information such as social norms and user intent. The approach allows for dynamic adaptation of control specifications in real-time, addressing the limitations of traditional MPC in handling diverse and structured specifications. The framework's effectiveness is validated through an autonomous driving scenario, showcasing its ability to align with personal preferences and respond appropriately to social situations like yielding to emergency vehicles.
Context-aware MPC can now adapt to social norms and user intent in real-time, transforming how autonomous systems interact with their environments.
While MPC effectively handles structured, diverse, and low-level specifications, it lacks the capability to dynamically incorporate high-level contextual information such as social norms, user intent, or natural language instructions. To address this limitation, this manuscript introduces an agentic MPC framework that enables context-aware, semantically adaptive control synthesis by integrating with large language model-based agents. The agent interprets heterogeneous inputs, including natural language messages, environmental observations, and external knowledge, to resynthesize the control specifications. The effectiveness of the framework is demonstrated in an autonomous driving scenario, where the system aligns with personal preferences or responds to social situations such as emergency vehicle yielding.