In different wireless network scenarios, multiple network entities need to cooperate in order to achieve a common task with minimum delay and energy consumption. Future wireless networks mandate exchanging high dimensional data in dynamic and uncertain environments, therefore implementing communication control tasks becomes challenging and highly complex. Multi-agent reinforcement learning with emergent communication (EC-MARL) is a promising solution to address high dimensional continuous control problems with partially observable states in a cooperative fashion where agents build an emergent communication protocol to solve complex tasks. This paper articulates the importance of EC-MARL within the context of future 6G wireless networks, which imbues autonomous decision-making capabilities into network entities to solve complex tasks such as autonomous driving, robot navigation, flying base stations network planning, and smart city applications. An overview of EC-MARL algorithms and their design criteria are provided while presenting use cases and research opportunities on this emerging topic.

Building autonomous --- i.e., choosing goals based on one's needs -- and adaptive -- i.e., surviving in ever-changing environments -- agents has been a holy grail of artificial intelligence (AI). A living organism is a prime example of such an agent, offering important lessons about adaptive autonomy. Here, we focus on interoception, a process of monitoring one's internal environment to keep it within certain bounds, which underwrites the survival of an organism. To develop AI with interoception, we need to factorize the state variables representing internal environments from external environments and adopt life-inspired mathematical properties of internal environment states. This paper offers a new perspective on how interoception can help build autonomous and adaptive agents by integrating the legacy of cybernetics with recent advances in theories of life, reinforcement learning, and neuroscience.

For this reason, their applicability in practice is limited. Growing environmental concerns and advancements in communication The above-mentioned scalability issues can be addressed and monitoring technologies have led to the increased by employing deep RL methods like the Deep Deterministic deployment of Distributed Energy Resources (DERs) Policy Gradient (DDPG) algorithm [10], which utilizes neural in power networks [1], comprising renewable energy sources networks to extend the Q-learning capabilities to continuous and prosumers. The market integration of these units is facilitated state and action spaces. The authors in [11]-[13] propose deep by their large-scale aggregation under financial entities, RL methods for the economic dispatch and market participation commonly known as Virtual Power Plants (VPPs), which have of DERs aggregated in a VPP. The main limitation the capacity for trading in wholesale electricity markets [2], of these works is that they fail to account for the complex [3]. As a self-interested market participant, a VPP aims at internal physical constraints of large-scale VPPs, such as maximizing its own profit generated by its market participation power generation limits and power flow constraints, in order to and the fulfillment of contractual obligations towards its ensure a safe operation.

Model-free Reinforcement Learning (RL) generally suffers from poor sample complexity, mostly due to the need to exhaustively explore the state-action space to find well-performing policies. On the other hand, we postulate that expert knowledge of the system often allows us to design simple rules we expect good policies to follow at all times. In this work, we hence propose a simple yet effective modification of continuous actor-critic frameworks to incorporate such rules and avoid regions of the state-action space that are known to be suboptimal, thereby significantly accelerating the convergence of RL agents. Concretely, we saturate the actions chosen by the agent if they do not comply with our intuition and, critically, modify the gradient update step of the policy to ensure the learning process is not affected by the saturation step. On a room temperature control case study, it allows agents to converge to well-performing policies up to 6-7x faster than classical agents without computational overhead and while retaining good final performance.

Robust Markov Decision Processes (MDPs) are receiving much attention in learning a robust policy which is less sensitive to environment changes. There are an increasing number of works analyzing sample-efficiency of robust MDPs. However, there are two major barriers to applying robust MDPs in practice. First, most works study robust MDPs in a model-based regime, where the transition probability needs to be estimated and requires a large amount of memories $\mathcal{O}(|\mathcal{S}|^2|\mathcal{A}|)$. Second, prior work typically assumes a strong oracle to obtain the optimal solution as an intermediate step to solve robust MDPs. However, in practice, such an oracle does not exist usually. To remove the oracle, we transform the original robust MDPs into an alternative form, which allows us to use stochastic gradient methods to solve the robust MDPs. Moreover, we prove the alternative form still plays a similar role as the original form. With this new formulation, we devise a sample-efficient algorithm to solve the robust MDPs in a model-free regime, which does not require an oracle and trades off a lower storage requirement $\mathcal{O}(|\mathcal{S}||\mathcal{A}|)$ with being able to generate samples from a generative model or Markovian chain. Finally, we validate our theoretical findings via numerical experiments, showing the efficiency with the alternative form of robust MDPs.

In reconfigurable intelligent surface (RIS)-assisted wireless communication systems, the pointing accuracy and intensity of reflections depend crucially on the 'profile,' representing the amplitude/phase state information of all elements in a RIS array. The superposition of multiple single-reflection profiles enables multi-reflection for distributed users. However, the optimization challenges from periodic element arrangements in single-reflection and multi-reflection profiles are understudied. The combination of periodical single-reflection profiles leads to amplitude/phase counteractions, affecting the performance of each reflection beam. This paper focuses on a dual-reflection optimization scenario and investigates the far-field performance deterioration caused by the misalignment of overlapped profiles. To address this issue, we introduce a novel deep reinforcement learning (DRL)-based optimization method. Comparative experiments against random and exhaustive searches demonstrate that our proposed DRL method outperforms both alternatives, achieving the shortest optimization time. Remarkably, our approach achieves a 1.2 dB gain in the reflection peak gain and a broader beam without any hardware modifications.

We introduce a simple but effective method for managing risk in model-based reinforcement learning with trajectory sampling that involves probabilistic safety constraints and balancing of optimism in the face of epistemic uncertainty and pessimism in the face of aleatoric uncertainty of an ensemble of stochastic neural networks. Various experiments indicate that the separation of uncertainties is essential to performing well with data-driven MPC approaches in uncertain and safety-critical control environments.

Parkour is a grand challenge for legged locomotion that requires robots to overcome various obstacles rapidly in complex environments. Existing methods can generate either diverse but blind locomotion skills or vision-based but specialized skills by using reference animal data or complex rewards. However, autonomous parkour requires robots to learn generalizable skills that are both vision-based and diverse to perceive and react to various scenarios. In this work, we propose a system for learning a single end-to-end vision-based parkour policy of diverse parkour skills using a simple reward without any reference motion data. We develop a reinforcement learning method inspired by direct collocation to generate parkour skills, including climbing over high obstacles, leaping over large gaps, crawling beneath low barriers, squeezing through thin slits, and running. We distill these skills into a single vision-based parkour policy and transfer it to a quadrupedal robot using its egocentric depth camera. We demonstrate that our system can empower two different low-cost robots to autonomously select and execute appropriate parkour skills to traverse challenging real-world environments.

Humans throw and catch objects all the time. However, such a seemingly common skill introduces a lot of challenges for robots to achieve: The robots need to operate such dynamic actions at high-speed, collaborate precisely, and interact with diverse objects. In this paper, we design a system with two multi-finger hands attached to robot arms to solve this problem. We train our system using Multi-Agent Reinforcement Learning in simulation and perform Sim2Real transfer to deploy on the real robots. To overcome the Sim2Real gap, we provide multiple novel algorithm designs including learning a trajectory prediction model for the object. Such a model can help the robot catcher has a real-time estimation of where the object will be heading, and then react accordingly. We conduct our experiments with multiple objects in the real-world system, and show significant improvements over multiple baselines. Our project page is available at \url{https://binghao-huang.github.io/dynamic_handover/}.

In this paper, we establish a task-oriented cross-system design framework to minimize the required packet rate for timely and accurate modeling of a real-world robotic arm in the Metaverse, where sensing, communication, prediction, control, and rendering are considered. To optimize a scheduling policy and prediction horizons, we design a Constraint Proximal Policy Optimization(C-PPO) algorithm by integrating domain knowledge from relevant systems into the advanced reinforcement learning algorithm, Proximal Policy Optimization(PPO). Specifically, the Jacobian matrix for analyzing the motion of the robotic arm is included in the state of the C-PPO algorithm, and the Conditional Value-at-Risk(CVaR) of the state-value function characterizing the long-term modeling error is adopted in the constraint. Besides, the policy is represented by a two-branch neural network determining the scheduling policy and the prediction horizons, respectively. To evaluate our algorithm, we build a prototype including a real-world robotic arm and its digital model in the Metaverse. The experimental results indicate that domain knowledge helps to reduce the convergence time and the required packet rate by up to 50%, and the cross-system design framework outperforms a baseline framework in terms of the required packet rate and the tail distribution of the modeling error.