World Model-Enabled Causal Digital Twins for Semantic Communications in Physical AI Systems

2026-05-15

AI Summary

This work addresses the limitation of existing semantic communication approaches, which typically optimize instantaneous performance for single tasks and thus fail to meet the demands of long-horizon, closed-loop physical AI systems integrating perception, communication, reasoning, and control. The authors formulate semantic communication as a problem of maximizing long-term reward per transmitted bit under a wireless bit budget. They introduce a causal information value metric to quantify the marginal contribution of semantic tokens to long-term returns and develop a world-model-driven causal digital twin framework that enables counterfactual reasoning and long-term planning. This framework achieves, for the first time, joint optimization of semantic communication and control in physical AI closed-loop systems, significantly outperforming state-of-the-art reinforcement learning methods in AirSim-Sionna drone navigation simulations by improving both reward-per-kilobit efficiency and navigation success rates.

Abstract

Semantic communication has emerged as a promising paradigm for enabling goal-oriented networking. However, most existing semantic communication solutions are tailored to one-shot tasks and optimize instantaneous performance. Hence, they cannot be used to support closed-loop dynamic systems with physical artificial intelligence (AI), in which the transmitted semantics affect not only the current inference outcome but also future control actions, state evolution, and ultimately long-horizon task performance. To address this gap, this paper investigates goal-oriented semantic communications for physical AI systems with closed-loop sensing-communication-inference-control. In particular, the problem of semantic communications is formulated as a long-term return-per-bit maximization under wireless bit-budget constraints while capturing both control efficiency and communication efficiency. To solve this problem, a novel causal information value (CIV) metric is introduced to evaluate the marginal contribution of each semantic token to the expected long-term return by transmission interventions. Then, a world-model-enabled causal digital twin (WM-CDT) framework is proposed to capture the dynamics of closed-loop physical AI systems and enable counterfactual reasoning for long-horizon imagined rollouts. Based on these imagined rollouts, an actor-critic policy is trained for long-horizon agent control with high data efficiency, while the semantic token selector is trained through CIV-per-bit evaluation. Extensive simulations on an AirSim-Sionna-based unmanned aerial vehicle (UAV) navigation simulator show that the proposed WM-CDT framework achieves significant improvement in return-per-kbit and navigation success rate compared to existing reinforcement learning solutions.

Problem

• semantic communication
• physical AI systems
• closed-loop control
• long-horizon task performance
• wireless bit-budget constraints

Innovation

• Causal Information Value
• World Model
• Causal Digital Twin
• Semantic Communication
• Physical AI