Abstract: With the rapid development of artificial intelligence, robotics, and Internet of Things, multi-robot systems are progressively acquiring human-like environmental perception and understanding capabilities, empowering them to complete complex tasks through autonomous decision-making and interaction. However, the Internet of Robotic Things (IoRT) faces significant challenges in terms of spectrum resources, sensing accuracy, communication latency, and energy supply. To address these issues, a reconfigurable intelligent surface (RIS)-aided IoRT network is proposed to enhance the overall performance of robotic communication, sensing, computation, and energy harvesting. In the case studies, by jointly optimizing parameters such as transceiver beamforming, robot trajectories, and RIS coefficients, solutions based on multi-agent deep reinforcement learning and multi-objective optimization are proposed to solve problems such as beamforming design, path planning, target sensing, and data aggregation. Numerical results are provided to demonstrate the effectiveness of proposed solutions in improve communication quality, sensing accuracy, computation error, and energy efficiency of RIS-aided IoRT networks.
Addressing the "Triple Threat" in IoRT
The Internet of Robotic Things (IoRT) envisions fleets of autonomous robots collaborating in complex environments. However, these systems face a "triple threat" of constraints that hinder their deployment in real-world industrial settings:
- Spectrum & Interference: Crowded airwaves lead to signal degradation as robot density increases.
- Physical Obstacles: Warehouses and factories are "cluttered" environments where walls and machinery block high-frequency signals.
- Energy Scarcity: Mobile robots have limited battery life, and high-performance wireless communication is power-hungry.
Traditionally, wireless networks are designed to adapt to a given environment. This research proposes a paradigm shift: using Reconfigurable Intelligent Surfaces (RIS) to program the environment. RIS are thin, smart surfaces that can reflect and steer electromagnetic waves toward moving robots, effectively "bending" signals around corners and creating favorable propagation conditions where none existed before.
AI-Driven Optimization Framework
The paper introduces an RIS-aided framework that optimizes four critical pillars of robotic operation: communication, sensing, computation, and energy harvesting. To handle the dynamic nature of these systems, the authors employ several advanced AI and mathematical tools:
- Federated Deep Reinforcement Learning (FDRL): Used for joint communication and trajectory design. A "Global" stage configures the RIS, while a "Local" stage handles individual robot path planning. This federated approach allows robots to learn from collective experience without sharing raw data, enhancing privacy and scalability.
- Multi-Objective Optimization (MOO): Applied to Integrated Sensing and Communication (ISAC). By using a weighted Tchebycheff-based method, the system finds the "Pareto front"—the optimal balance between high data rates and sensing accuracy.
- Double-Agent Deep Reinforcement Learning (DRL): Employed for "Over-the-Air Computation" (AirComp) and wireless charging. This method addresses "channel aging" by having two agents cooperate to optimize energy beamforming and data aggregation simultaneously.
Performance and Key Insights
Numerical simulations were conducted in a 30m x 30m area cluttered with obstacles, featuring RIS units with 20 to 40 elements. The results demonstrate that the environment is no longer a static constraint but a software-defined variable.
Key findings include:
- Energy Efficiency: The proposed FDRL algorithm significantly outperformed standard Orthogonal Multiple Access (OMA) and Non-Orthogonal Multiple Access (NOMA) setups. RIS-aided path planning proved more efficient than simply increasing transmit power.
- Sensing and Computation: In multi-cell networks, coordinated RIS beamforming achieved a gain ratio of up to 20x in sensing/communication performance. For AirComp, the Double-Agent DRL reduced the Mean Square Error (MSE) by 57.6% compared to single-agent approaches.
- Trajectory Synergy: A robot’s path is no longer just the shortest distance between two points; it is optimized to maintain the best "energy beam" from an RIS, ensuring continuous connectivity and charging.
Future Outlook
While the results are promising, challenges remain in real-time Channel State Information (CSI) acquisition for moving targets and the impact of hardware quantization errors in physical RIS deployments. Future work aims to integrate "Embodied Intelligence," where Large Language Models (LLMs) and multimodal AI help robots collaborate naturally with humans while the RIS infrastructure provides the underlying ultra-reliable, low-latency connectivity fabric.
Summary updated