Model-free Reinforcement Learning Research Articles

State Identification via Symbolic Time Series Analysis for Reinforcement Learning Control

Abstract This technical brief makes use of the concept of symbolic time series analysis (STSA) for identifying discrete states from the non-linear time response of a chaotic dynamical for model-free reinforcement learning (RL) control. Along this line, a projection-based method is adopted to construct probabilistic finite state automata (PFSA) for identification of the current state (i.e., operational regime) of the Lorenz system; and a (Q-map-based and model-free) RL control strategy is formulated to reach the target state from the (identified) current state. A synergistic combination of PFSA-based state identification and RL control is demonstrated by simulation of a numeric model of the Lorenz system, which yields very satisfactory performance to reach the target states from the current states in real time.

A homotopy-based reinforcement learning scheme to optimal control for Markov switched interconnected systems

ABSTRACT In this paper, a homotopy-based reinforcement learning optimal control method is developed for Markov switched interconnected systems with unknown system dynamics. By utilising the subsystem decomposition method and parallel learning control method, the solution of the game coupled algebraic Riccati equations with jumping parameters is approximated. To dispense with the requirement of initial stability, a homotopy-based policy iteration is introduced, which can place unstable poles into a stable plane. In this regard, a model-free reinforcement learning method is presented to design the optimal controller for Markov switched interconnected systems. Finally, the effectiveness of the proposed method is verified by a numerical example and a practical example.

Controlling Estimation Error in Reinforcement Learning via Reinforced Operation

In Value-based and Actor-Critic reinforcement learning (RL) methods, both inaccuracy and instability of value estimation will detrimentally affect their performance. Some typical RL methods, such as Maxmin Q-learning and QMD3, are plagued by the underestimation problem while failing to trade off the estimation bias and variance jointly. We propose the Reinforced Operation (RO) to address these shortcomings, which selects the closest median among multiple Q-function. RO is applicable to any model-free RL method. Theoretically, we introduce the Mean Square Error (MSE) to jointly analyze the estimation bias and variance of value estimation methods. We also demonstrate the superiority of RO in MSE reduction and give an upper bound for the estimation bias of value estimation methods with arbitrary distribution to guide the calculation of the estimation bias of value estimation methods. Based on RO, we propose the variants of Q-learning and TD3, Reinforced Q-learning (RQ) and Reinforced Delayed Deep Deterministic policy gradient (RD3), respectively, to tackle different tasks. We empirically demonstrate that our method can reduce estimation error and achieve superior performance on discrete and continuous benchmark tasks.

Reinforcement learning in building controls: A comparative study of algorithms considering model availability and policy representation

Reinforcement Learning (RL) has presented considerable potential as an advanced control technique in building controls to enable buildings operating more energy-efficient. As various types of RL algorithms have been studied on their performance of building controls, benchmarking these algorithms across the entire spectrum of features is essential to provide an overview and deepen the understanding of RL applications. Therefore, this study aims to compare and analyze the effectiveness of various RL algorithms, encompassing the entire RL categories featured by value-based, policy gradient, actor-critic and model-based RL considering model availability and policy representation. To provide a comprehensive analysis, in addition to the control performance quantified by the cumulative rewards based on the cost function of RL, data demand and robustness of hyperparameter tuning were investigated. The open-source Gym-Eplus framework was selected as the virtual environment to train and test different RL agents. The results showed that both model-free and model-based RL agents outperformed the baseline rule-based control in terms of energy consumption and thermal comfort, and RL agents were capable of evaluating both short-term and long-term rewards to achieve adaptive control optimization continuously along with the online control process. Model-based RL agent improved the data sampling efficiency but presented a relatively sacrificed control performance during the tested summer days.

Learning robust autonomous navigation and locomotion for wheeled-legged robots.

Autonomous wheeled-legged robots have the potential to transform logistics systems, improving operational efficiency and adaptability in urban environments. Navigating urban environments, however, poses unique challenges for robots, necessitating innovative solutions for locomotion and navigation. These challenges include the need for adaptive locomotion across varied terrains and the ability to navigate efficiently around complex dynamic obstacles. This work introduces a fully integrated system comprising adaptive locomotion control, mobility-aware local navigation planning, and large-scale path planning within the city. Using model-free reinforcement learning (RL) techniques and privileged learning, we developed a versatile locomotion controller. This controller achieves efficient and robust locomotion over various rough terrains, facilitated by smooth transitions between walking and driving modes. It is tightly integrated with a learned navigation controller through a hierarchical RL framework, enabling effective navigation through challenging terrain and various obstacles at high speed. Our controllers are integrated into a large-scale urban navigation system and validated by autonomous, kilometer-scale navigation missions conducted in Zurich, Switzerland, and Seville, Spain. These missions demonstrate the system's robustness and adaptability, underscoring the importance of integrated control systems in achieving seamless navigation in complex environments. Our findings support the feasibility of wheeled-legged robots and hierarchical RL for autonomous navigation, with implications for last-mile delivery and beyond.

Interactive learning for multi-finger dexterous hand: A model-free hierarchical deep reinforcement learning approach

When a multi-fingered dexterous hand interacts with the external environment, it encounters various challenges, including the utilization of complex control techniques and the intricate coordination of finger motion sequences. Previous studies have primarily concentrated on investigating the interaction between multi-fingered dexterous hands and external objects, usually using model-based control or model-free reinforcement learning techniques. However, during practical implementation, various constraining factors are encountered, such as intricate modeling and limited interaction capabilities. In practical scenarios, the utilization of multi-fingered dexterous hands is imperative for the swift and efficient execution of a wide range of interactive tasks, including but not limited to throwing a ball and playing rock-paper-scissors. These tasks require skilled manual dexterity to demonstrate both precise control and quick responsiveness. To tackle this issue, we propose a hierarchical control approach for multi-fingered dexterous hands with interactive functionalities, utilizing model-free deep reinforcement learning. The complex interaction task is decomposed into simple sub-tasks using hierarchical strategy and action primitive decomposition, which effectively reduces the complexity of the action space, and achieves the motion planning and end finger trajectory control of dexterous hand. In a simulated environment, the aforementioned method has successfully executed interactive tasks, including ball throwing and playing rock-paper-scissors. It achieved a maximum normalized reward of 0.83 and an 84 % success rate. These results are noteworthy in terms of both control accuracy and response speed. This study offers novel insights into the effective resolution of the intricate challenges associated with interactions involving multi-fingered dexterous hands and human-computer interaction.

Real-world humanoid locomotion with reinforcement learning.

Humanoid robots that can autonomously operate in diverse environments have the potential to help address labor shortages in factories, assist elderly at home, and colonize new planets. Although classical controllers for humanoid robots have shown impressive results in a number of settings, they are challenging to generalize and adapt to new environments. Here, we present a fully learning-based approach for real-world humanoid locomotion. Our controller is a causal transformer that takes the history of proprioceptive observations and actions as input and predicts the next action. We hypothesized that the observation-action history contains useful information about the world that a powerful transformer model can use to adapt its behavior in context, without updating its weights. We trained our model with large-scale model-free reinforcement learning on an ensemble of randomized environments in simulation and deployed it to the real-world zero-shot. Our controller could walk over various outdoor terrains, was robust to external disturbances, and could adapt in context.

A variable speed limit control approach for freeway tunnels based on the model-based reinforcement learning framework with safety perception

To improve the traffic safety and efficiency of freeway tunnels, this study proposes a novel variable speed limit (VSL) control strategy based on the model-based reinforcement learning framework (MBRL) with safety perception. The MBRL framework is designed by developing a multi-lane cell transmission model for freeway tunnels as an environment model, which is built so that agents can interact with the environment model while interacting with the real environment to improve the sampling efficiency of reinforcement learning. Based on a real-time crash risk prediction model for freeway tunnels that uses random deep and cross networks, the safety perception function inside the MBRL framework is developed. The reinforcement learning components fully account for most current tunnels' application conditions, and the VSL control agent is trained using a deep dyna-Q method. The control process uses a safety trigger mechanism to reduce the likelihood of crashes caused by frequent changes in speed. The efficacy of the proposed VSL strategies is validated through simulation experiments. The results show that the proposed VSL strategies significantly increase traffic safety performance by between 16.00% and 20.00% and traffic efficiency by between 3.00% and 6.50% compared to a fixed speed limit approach. Notably, the proposed strategies outperform traditional VSL strategy based on the traffic flow prediction model in terms of traffic safety and efficiency improvement, and they also outperform the VSL strategy based on model-free reinforcement learning framework when sampling efficiency is considered together. In addition, the proposed strategies with safety triggers are safer than those without safety triggers. These findings demonstrate the potential for MBRL-based VSL strategies to improve traffic safety and efficiency within freeway tunnels.

Optimal Tracking Control of Heterogeneous MASs Using Event-Driven Adaptive Observer and Reinforcement Learning.

This article considers the output tracking control problem of nonidentical linear multiagent systems (MASs) using a model-free reinforcement learning (RL) algorithm, where partial followers have no prior knowledge of the leader's information. To lower the communication and computing burden among agents, an event-driven adaptive distributed observer is proposed to predict the leader's system matrix and state, which consists of the estimated value of relative states governed by an edge-based predictor. Meanwhile, the integral input-based triggering condition is exploited to decide whether to transmit its private control input to its neighbors. Then, an RL-based state feedback controller for each agent is developed to solve the output tracking control problem, which is further converted into the optimal control problem by introducing a discounted performance function. Inhomogeneous algebraic Riccati equations (AREs) are derived to obtain the optimal solution of AREs. An off-policy RL algorithm is used to learn the solution of inhomogeneous AREs online without requiring any knowledge of the system dynamics. Rigorous analysis shows that under the proposed event-driven adaptive observer mechanism and RL algorithm, all followers are able to synchronize the leader's output asymptotically. Finally, a numerical simulation is demonstrated to verify the proposed approach in theory.

Fuzzy-based collective pitch control for wind turbine via deep reinforcement learning

Wind turbines (WTs) have highly nonlinear and uncertain dynamics due to aerodynamic complexity, mechanical factors, and fluctuations in wind conditions. Turbulence and wind shear add complexity to modelling, especially in constant power region (region 3). Thus, an effective control design demands a deep understanding of the nonlinearities and uncertainties. This paper suggests a novel model-free reinforcement learning (RL) collective pitch angle controller to operate efficiently in region 3. The proposed controller stabilizes generator speed, maximizes power output, and minimizes fluctuations while accommodating system uncertainties, nonlinearity, and pitch limits. The disparity between WT dynamics due to wind speed perturbations and uncertainties is measured using a gap-metric criterion. The controller design adopts a deep deterministic policy gradient (DDPG) algorithm to train six agents in a medium-fidelity WT environment at different mean wind speeds to ensure the controller's robustness. Initially, imitation learning is used for efficient sample collection to fasten training convergence. Afterwards, the agent learns by interacting with the environment. After the training, the pitch control outputs from multi-trained agents are processed by a fuzzy system to have smooth transitions under different operating conditions. The resulting fuzzy DDPG (F-DDPG) controller is deployed to obtain the optimal pitch control action. The performance of the proposed F-DDPG controller is compared to the gain-scheduled PI (GSPI), Linear-Quadratic-Regulator (LQR), and single-DDPG-agent controllers. The controllers are simulated in high-fidelity onshore and offshore 5-MW WT environments using the OpenFAST/MATLAB simulation tools. The results reveal the superiority of the proposed controller in generalizing its optimal performance in different operating conditions.

Model-free reinforcement learning for motion planning of autonomous agents with complex tasks in partially observable environments

Model-free reinforcement learning for motion planning of autonomous agents with complex tasks in partially observable environments

Fractional Deep Reinforcement Learning for Age-Minimal Mobile Edge Computing

Mobile edge computing (MEC) is a promising paradigm for real-time applications with intensive computational needs (e.g., autonomous driving), as it can reduce the processing delay. In this work, we focus on the timeliness of computational-intensive updates, measured by Age-of-Information (AoI), and study how to jointly optimize the task updating and offloading policies for AoI with fractional form. Specifically, we consider edge load dynamics and formulate a task scheduling problem to minimize the expected time-average AoI. The uncertain edge load dynamics, the nature of the fractional objective, and hybrid continuous-discrete action space (due to the joint optimization) make this problem challenging and existing approaches not directly applicable. To this end, we propose a fractional reinforcement learning (RL) framework and prove its convergence. We further design a model-free fractional deep RL (DRL) algorithm, where each device makes scheduling decisions with the hybrid action space without knowing the system dynamics and decisions of other devices. Experimental results show that our proposed algorithms reduce the average AoI by up to 57.6% compared with several non-fractional benchmarks.

Water management scheme based on prioritized deep deterministic policy gradient for proton exchange membrane fuel cells

AbstractTo effectively tackle the intricate and dynamic challenges encountered in proton exchange membrane fuel cells (PEMFCs), this paper introduces a model-free reinforcement learning approach to address its water management issue. Recognizing the limitations of conventional reinforcement learning methods such as Q-learning in handling the continuous actions and nonlinearity inherent in PEMFCs water management, we propose a prioritized deep deterministic policy gradient (DDPG) method. This method, rooted in the Actor-Critic framework, leverages double neural networks and prioritized experience replay to enable adaptive water management and balance. Additionally, we establish a PEMFCs water management platform and implement the prioritized DDPG method using "Tianshou", a modularized Python library for deep reinforcement learning. Through experimentation, the effectiveness of our proposed method is verified. This study contributes to advancing the understanding and management of water dynamics in PEMFCs, offering a promising avenue for enhancing their performance and reliability.

Robust hierarchical games of linear discrete-time systems based on off-policy model-free reinforcement learning

An off-policy model-free reinforcement learning (RL) algorithm is proposed for a robust hierarchical game while considering incomplete information and input constraints. The robust hierarchical game exhibits characteristics of a Stackelberg–Nash (SN) game, where equilibrium points are designated as Stackelberg–Nash–Saddle equilibrium (SNE) points. An off-policy method is employed for the RL algorithm, addressing input constraints by using excitation input instead of real-time update polices as control inputs. Moreover, a model-free method is implemented for the off-policy RL algorithm, accounting for the challenge posed by incomplete information. The goal of this paper is to develop an off-policy model-free RL algorithm to obtain approximate SNE polices of the robust hierarchical game with incomplete information and input constraints. Furthermore, the convergence and effectiveness of the off-policy model-free RL algorithm are guaranteed by proving the equivalence of Bellman equation between nominal SNE policies and approximate SNE policies. Finally, a simulation is provided to verify the advantage of the developed algorithm.

Data-Efficient Reinforcement Learning for Complex Nonlinear Systems.

This article proposes a data-efficient model-free reinforcement learning (RL) algorithm using Koopman operators for complex nonlinear systems. A high-dimensional data-driven optimal control of the nonlinear system is developed by lifting it into the linear system model. We use a data-driven model-based RL framework to derive an off-policy Bellman equation. Building upon this equation, we deduce the data-efficient RL algorithm, which does not need a Koopman-built linear system model. This algorithm preserves dynamic information while reducing the required data for optimal control learning. Numerical and theoretical analyses of the Koopman eigenfunctions for dataset truncation are discussed in the proposed model-free data-efficient RL algorithm. We validate our framework on the excitation control of the power system.

A probabilistic successor representation for context-dependent learning.

Two of the main impediments to learning complex tasks are that relationships between different stimuli, including rewards, can be uncertain and context-dependent. Reinforcement learning (RL) provides a framework for learning, by predicting total future reward directly (model-free RL), or via predictions of future states (model-based RL). Within this framework, "successor representation" (SR) predicts total future occupancy of all states. A recent theoretical proposal suggests that the hippocampus encodes the SR in order to facilitate prediction of future reward. However, this proposal does not take into account how learning should adapt under uncertainty and switches of context. Here, we introduce a theory of learning SRs using prediction errors which includes optimally balancing uncertainty in new observations versus existing knowledge. We then generalize that approach to a multicontext setting, allowing the model to learn and maintain multiple task-specific SRs and infer which one to use at any moment based on the accuracy of its predictions. Thus, the context used for predictions can be determined by both the contents of the states themselves and the distribution of transitions between them. This probabilistic SR model captures animal behavior in tasks which require contextual memory and generalization, and unifies previous SR theory with hippocampal-dependent contextual decision-making. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Fuzzy-based predictive deep reinforcement learning for robust and constrained optimal control of industrial solar thermal plants

Integrating distributed solar fields (DSFs) into conventional heat and power plants (CHPs) of industries is mostly constrained by the availability of a real-time capable control scheme. Safe and efficient operation of industrial DSF requires the supply of a fluctuating and periodically available energy at the required temperature while reducing losses and ensuring operational constraint. Reinforcement learning (RL), specifically Q-learning methods, are being recently applied for such control tasks. However, existing RL approaches cannot be directly applied to continuous domain control tasks or tasks with operation constraints. While using deep learning in model-free RL can remove these limitations, tractability and scalability boundaries on performing the deep learning are becoming significant, with scarce data and multi-objective concerns. In contrast, model-based deep RL schemes can be applied for sample efficiency and satisfying operational constraints, but their modeling framework is not general and accurate. To address these challenges, the work here develops a hybrid fuzzy convolution model (HFCM) that takes full advantages of data, models (dynamic and steady-state), and prior knowledge on industrial DSF. The HFCM is then extended for use in deep deterministic policy gradient (DDPG) algorithm to learn the DSF control task. This is done so by solving a multi-objective optimization problem, which is formulated as a constrained Markov decision process (CMDP) with continuous state and actions. Some practical and relevant findings were made with this HFCM. Firstly, it allowed the DDPG agent to learn independently a temperature state and a disturbance state, using only temperature measurement and single realizations of state vector. It also permitted the testing of an operational state violation strategy on a DDPG actions, and at the same time, the correction of the effects at a safety layer, if needed. Furthermore, the proposed control strategy simultaneously reduced mean temperature tracking error by 24%–51% and energy gain by 13.9%–17.35% when compared respectively to model-free DDPG and MPC baselines.

MFRLMO: Model-free reinforcement learning for multi-objective optimization of apache spark

Hyperparameter optimization (HO) is a must to figure out to what extent can a specific configuration of hyperparameters contribute to the performance of a machine learning task. The hardware and MLlib library of Apache Spark have the potential to improve big data processing performance when a tuning operation is combined with the exploitation of hyperparameters. To the best of our knowledge, the most of existing studies employ a black-box approach that results in misleading results due to ignoring the interior dynamics of big data processing. They suffer from one or more drawbacks including high computational cost, large search space, and sensitivity to the dimension of multi-objective functions. To address the issues above, this work proposes a new model-free reinforcement learning for multi-objective optimization of Apache Spark, thereby leveraging reinforcement learning (RL) agents to uncover the internal dynamics of Apache Spark in HO. To bridge the gap between multi-objective optimization and interior constraints of Apache Spark, our method runs a lot of iterations to update each cell of the RL grid. The proposed model-free learning mechanism achieves a tradeoff between three objective functions comprising time, memory, and accuracy. To this end, optimal values of the hyperparameters are obtained via an ensemble technique that analyzes the individual results yielded by each objective function. The results of the experiments show that the number of cores has not a direct effect on $speedup$. Further, although grid size has an impact on the time passed between two adjoining iterations, it is negligible in the computational burden. Dispersion and risk values of model-free RL differ when the size of the data is small. On average, MFRLMO produced $speedup$ that is 37% better than those of the competitors. Last, our approach is very competitive in terms of converging to a high accuracy when optimizing Convolutional Neural networks (CNN).

Non-myopic Bayesian optimization using model-free reinforcement learning and its application to optimization in electrochemistry

Bayesian Optimization (BO) is a robust tool for tackling black-box optimization problems, yet traditional acquisition functions often suffer from a short-sighted approach, leading to suboptimal sampling. In this study, we introduce a novel methodology that defines a Markov Decision Process (MDP) tailored for BO, enabling the incorporation of Reinforcement Learning (RL) into the BO framework. Our RL-based BO method strategically evaluates the utility of current data acquisition in the context of future decision making, this forward-looking perspective ensures that data collected at present is optimally leveraged in subsequent steps, resulting in a more data-efficient and effective sampling strategy compared to traditional BO methods. Our method has been tested across a range of benchmark functions, it consistently demonstrates superior data efficiency over existing BO algorithms, irrespective of the nature of the underlying function. We also explore how adjusting the lookahead horizon influences the performance of our RL-BO approach. Our RL-BO method is further applied to an Ag/C catalyst optimization problem to compare its data efficiency with other BO strategies, further showcasing the method's adaptability to complex real-world problems.

Electric vehicle charging design: The factored action based reinforcement learning approach

Charging optimization design for Electric Vehicles (EV) is challenging because it must account for various uncertainties and design aspects such as random EVs’ arrivals and departures, battery degradation, and transformer Loss of Life (LoL). Model-free reinforcement learning (RL) can be employed to tackle such the EV charging design where it does not require to explicitly model the environment dynamics and accurately predict relevant system parameters. However, the high complexity involved in conventional RL-based approaches usually limits its application to only small-scale EV charging settings, which is impractical. To overcome this limitation, we employ the factored action based RL method to transform the formulated Markov Decision Process (MDP). Then, we propose novel reward shaping and hybrid learning methods combining the Convolutional Neural Network (CNN) and Proximal Policy Optimization (PPO) algorithm to extract relevant features from high-dimension state space and efficiently solve the transformed MDP problem. Extensive numerical studies demonstrate that the proposed design can be used to control a charging station (CS) supporting a large number of EVs. Moreover, we show that the proposed framework greatly outperforms other baselines including single-agent and multi-agent RL based strategies and a heuristic power scheduling algorithm in terms of the achieved reward.