Reward shaping is a technique in reinforcement learning that addresses the sparse-reward problem by providing more frequent and informative rewards. We introduce a self-adaptive and highly efficient reward shaping mechanism that incorporates success rates derived from historical experiences as shaped rewards. The success rates are sampled from Beta distributions, which dynamically evolve from uncertain to reliable values as data accumulates. Initially, the shaped rewards exhibit more randomness to encourage exploration, while over time, the increasing certainty enhances exploitation, naturally balancing exploration and exploitation. Our approach employs Kernel Density Estimation (KDE) combined with Random Fourier Features (RFF) to derive the Beta distributions, providing a computationally efficient, non-parametric, and learning-free solution for high-dimensional continuous state spaces. Our method is validated on various tasks with extremely sparse rewards, demonstrating notable improvements in sample efficiency and convergence stability over relevant baselines.
Conventional uncertainty-aware temporal difference (TD) learning often assumes a zero-mean Gaussian distribution for TD errors, leading to inaccurate error representations and compromised uncertainty estimation. We introduce a novel framework for generalized Gaussian error modeling in deep reinforcement learning to enhance the flexibility of error distribution modeling by incorporating additional higher-order moment, particularly kurtosis, thereby improving the estimation and mitigation of data-dependent aleatoric uncertainty. We examine the influence of the shape parameter of the generalized Gaussian distribution (GGD) on aleatoric uncertainty and provide a closed-form expression that demonstrates an inverse relationship between uncertainty and the shape parameter. Additionally, we propose a theoretically grounded weighting scheme to address epistemic uncertainty by fully leveraging the GGD. We refine batch inverse variance weighting with bias reduction and kurtosis considerations, enhancing robustness. Experiments with policy gradient algorithms demonstrate significant performance gains.
For on-policy reinforcement learning, discretizing action space for continuous control can easily express multiple modes and is straightforward to optimize. However, without considering the inherent ordering between the discrete atomic actions, the explosion in the number of discrete actions can possess undesired properties and induce a higher variance for the policy gradient estimator. In this paper, we introduce a straightforward architecture that addresses this issue by constraining the discrete policy to be unimodal using Poisson probability distributions. This unimodal architecture can better leverage the continuity in the underlying continuous action space using explicit unimodal probability distributions. We conduct extensive experiments to show that the discrete policy with the unimodal probability distribution provides significantly faster convergence and higher performance for on-policy reinforcement learning algorithms in challenging control tasks, especially in highly complex tasks such as Humanoid. We provide theoretical analysis on the variance of the policy gradient estimator, which suggests that our attentively designed unimodal discrete policy can retain a lower variance and yield a stable learning process.
Reinforcement Learning (RL) has been widely used to solve tasks where the environment consistently provides a dense reward value. However, in real-world scenarios, rewards can often be poorly defined or sparse. Auxiliary signals are indispensable for discovering efficient exploration strategies and aiding the learning process. In this work, inspired by intrinsic motivation theory, we postulate that the intrinsic stimuli of novelty and surprise can assist in improving exploration in complex, sparsely rewarded environments. We introduce a novel sample-efficient method able to learn directly from pixels, an image-based extension of TD3 with an autoencoder called \textit{NaSA-TD3}. The experiments demonstrate that NaSA-TD3 is easy to train and an efficient method for tackling complex continuous-control robotic tasks, both in simulated environments and real-world settings. NaSA-TD3 outperforms existing state-of-the-art RL image-based methods in terms of final performance without requiring pre-trained models or human demonstrations.
Despite the successes and progress of deep reinforcement learning over the last decade, several challenges remain that hinder its broader application. Some fundamental aspects to improve include data efficiency, generalization capability, and ability to learn in sparse-reward environments, which often require human-designed dense rewards. Meta-learning has emerged as a promising approach to address these issues by optimizing components of the learning algorithm to meet desired characteristics. Additionally, a different line of work has extensively studied the use of intrinsic rewards to enhance the exploration capabilities of algorithms. This work investigates how meta-learning can improve the training signal received by RL agents. The focus is on meta-learning intrinsic rewards under a framework that doesn't rely on the use of meta-gradients. We analyze and compare this approach to the use of extrinsic rewards and a meta-learned advantage function. The developed algorithms are evaluated on distributions of continuous control tasks with both parametric and non-parametric variations, and with only sparse rewards accessible for the evaluation tasks.
Inspired by the success of the Transformer architecture in natural language processing and computer vision, we investigate the use of Transformers in Reinforcement Learning (RL), specifically in modeling the environment's dynamics using Transformer Dynamics Models (TDMs). We evaluate the capabilities of TDMs for continuous control in real-time planning scenarios with Model Predictive Control (MPC). While Transformers excel in long-horizon prediction, their tokenization mechanism and autoregressive nature lead to costly planning over long horizons, especially as the environment's dimensionality increases. To alleviate this issue, we use a TDM for short-term planning, and learn an autoregressive discrete Q-function using a separate Q-Transformer (QT) model to estimate a long-term return beyond the short-horizon planning. Our proposed method, QT-TDM, integrates the robust predictive capabilities of Transformers as dynamics models with the efficacy of a model-free Q-Transformer to mitigate the computational burden associated with real-time planning. Experiments in diverse state-based continuous control tasks show that QT-TDM is superior in performance and sample efficiency compared to existing Transformer-based RL models while achieving fast and computationally efficient inference.
This paper introduces a new empirical methodology, the Cross-environment Hyperparameter Setting Benchmark, that compares RL algorithms across environments using a single hyperparameter setting, encouraging algorithmic development which is insensitive to hyperparameters. We demonstrate that this benchmark is robust to statistical noise and obtains qualitatively similar results across repeated applications, even when using few samples. This robustness makes the benchmark computationally cheap to apply, allowing statistically sound insights at low cost. We demonstrate two example instantiations of the CHS, on a set of six small control environments (SC-CHS) and on the entire DM Control suite of 28 environments (DMC-CHS). Finally, to illustrate the applicability of the CHS to modern RL algorithms on challenging environments, we conduct a novel empirical study of an open question in the continuous control literature. We show, with high confidence, that there is no meaningful difference in performance between Ornstein-Uhlenbeck noise and uncorrelated Gaussian noise for exploration with the DDPG algorithm on the DMC-CHS.
Reinforcement learning (RL) has progressed substantially over the past decade, with much of this progress being driven by benchmarks. Many benchmarks are focused on video or board games, and a large number of robotics benchmarks lack diversity and real-world applicability. In this paper, we aim to simplify the process of applying reinforcement learning in the 3D simulation league of Robocup, a robotic football competition. To this end, we introduce a Robocup-based RL environment based on the open source rcssserver3d soccer server, simple pre-defined tasks, and integration with a popular RL library, Stable Baselines 3. Our environment enables the creation of high-dimensional continuous control tasks within a robotics football simulation. In each task, an RL agent controls a simulated Nao robot, and can interact with the ball or other agents. We open-source our environment and training code at https://github.com/Michael-Beukman/RobocupGym.
We introduce Proximal Policy Distillation (PPD), a novel policy distillation method that integrates student-driven distillation and Proximal Policy Optimization (PPO) to increase sample efficiency and to leverage the additional rewards that the student policy collects during distillation. To assess the efficacy of our method, we compare PPD with two common alternatives, student-distill and teacher-distill, over a wide range of reinforcement learning environments that include discrete actions and continuous control (ATARI, Mujoco, and Procgen). For each environment and method, we perform distillation to a set of target student neural networks that are smaller, identical (self-distillation), or larger than the teacher network. Our findings indicate that PPD improves sample efficiency and produces better student policies compared to typical policy distillation approaches. Moreover, PPD demonstrates greater robustness than alternative methods when distilling policies from imperfect demonstrations. The code for the paper is released as part of a new Python library built on top of stable-baselines3 to facilitate policy distillation: `sb3-distill'.
Constrained Reinforcement Learning (CRL) tackles sequential decision-making problems where agents are required to achieve goals by maximizing the expected return while meeting domain-specific constraints, which are often formulated as expected costs. In this setting, policy-based methods are widely used since they come with several advantages when dealing with continuous-control problems. These methods search in the policy space with an action-based or parameter-based exploration strategy, depending on whether they learn directly the parameters of a stochastic policy or those of a stochastic hyperpolicy. In this paper, we propose a general framework for addressing CRL problems via gradient-based primal-dual algorithms, relying on an alternate ascent/descent scheme with dual-variable regularization. We introduce an exploration-agnostic algorithm, called C-PG, which exhibits global last-iterate convergence guarantees under (weak) gradient domination assumptions, improving and generalizing existing results. Then, we design C-PGAE and C-PGPE, the action-based and the parameter-based versions of C-PG, respectively, and we illustrate how they naturally extend to constraints defined in terms of risk measures over the costs, as it is often requested in safety-critical scenarios. Finally, we numerically validate our algorithms on constrained control problems, and compare them with state-of-the-art baselines, demonstrating their effectiveness.
Drawing upon recent advances in language model alignment, we formulate offline Reinforcement Learning as a two-stage optimization problem: First pretraining expressive generative policies on reward-free behavior datasets, then fine-tuning these policies to align with task-specific annotations like Q-values. This strategy allows us to leverage abundant and diverse behavior data to enhance generalization and enable rapid adaptation to downstream tasks using minimal annotations. In particular, we introduce Efficient Diffusion Alignment (EDA) for solving continuous control problems. EDA utilizes diffusion models for behavior modeling. However, unlike previous approaches, we represent diffusion policies as the derivative of a scalar neural network with respect to action inputs. This representation is critical because it enables direct density calculation for diffusion models, making them compatible with existing LLM alignment theories. During policy fine-tuning, we extend preference-based alignment methods like Direct Preference Optimization (DPO) to align diffusion behaviors with continuous Q-functions. Our evaluation on the D4RL benchmark shows that EDA exceeds all baseline methods in overall performance. Notably, EDA maintains about 95\% of performance and still outperforms several baselines given only 1\% of Q-labelled data during fine-tuning.
Despite recent advances in improving the sample-efficiency of reinforcement learning (RL) algorithms, designing an RL algorithm that can be practically deployed in real-world environments remains a challenge. In this paper, we present Coarse-to-fine Reinforcement Learning (CRL), a framework that trains RL agents to zoom-into a continuous action space in a coarse-to-fine manner, enabling the use of stable, sample-efficient value-based RL algorithms for fine-grained continuous control tasks. Our key idea is to train agents that output actions by iterating the procedure of (i) discretizing the continuous action space into multiple intervals and (ii) selecting the interval with the highest Q-value to further discretize at the next level. We then introduce a concrete, value-based algorithm within the CRL framework called Coarse-to-fine Q-Network (CQN). Our experiments demonstrate that CQN significantly outperforms RL and behavior cloning baselines on 20 sparsely-rewarded RLBench manipulation tasks with a modest number of environment interactions and expert demonstrations. We also show that CQN robustly learns to solve real-world manipulation tasks within a few minutes of online training.
Components of cyber physical systems, which affect real-world processes, are often exposed to the internet. Replacing conventional control methods with Deep Reinforcement Learning (DRL) in energy systems is an active area of research, as these systems become increasingly complex with the advent of renewable energy sources and the desire to improve their efficiency. Artificial Neural Networks (ANN) are vulnerable to specific perturbations of their inputs or features, called adversarial examples. These perturbations are difficult to detect when properly regularized, but have significant effects on the ANN's output. Because DRL uses ANN to map optimal actions to observations, they are similarly vulnerable to adversarial examples. This work proposes a novel attack technique for continuous control using Group Difference Logits loss with a bifurcation layer. By combining aspects of targeted and untargeted attacks, the attack significantly increases the impact compared to an untargeted attack, with drastically smaller distortions than an optimally targeted attack. We demonstrate the impacts of powerful gradient-based attacks in a realistic smart energy environment, show how the impacts change with different DRL agents and training procedures, and use statistical and time-series analysis to evaluate attacks' stealth. The results show that adversarial attacks can have significant impacts on DRL controllers, and constraining an attack's perturbations makes it difficult to detect. However, certain DRL architectures are far more robust, and robust training methods can further reduce the impact.
Experience replay serves as a key component in the success of online reinforcement learning (RL). Prioritized experience replay (PER) reweights experiences by the temporal difference (TD) error empirically enhancing the performance. However, few works have explored the motivation of using TD error. In this work, we provide an alternative perspective on TD-error-based reweighting. We show the connections between the experience prioritization and occupancy optimization. By using a regularized RL objective with $f-$divergence regularizer and employing its dual form, we show that an optimal solution to the objective is obtained by shifting the distribution of off-policy data in the replay buffer towards the on-policy optimal distribution using TD-error-based occupancy ratios. Our derivation results in a new pipeline of TD error prioritization. We specifically explore the KL divergence as the regularizer and obtain a new form of prioritization scheme, the regularized optimal experience replay (ROER). We evaluate the proposed prioritization scheme with the Soft Actor-Critic (SAC) algorithm in continuous control MuJoCo and DM Control benchmark tasks where our proposed scheme outperforms baselines in 6 out of 11 tasks while the results of the rest match with or do not deviate far from the baselines. Further, using pretraining, ROER achieves noticeable improvement on difficult Antmaze environment where baselines fail, showing applicability to offline-to-online fine-tuning. Code is available at \url{https://github.com/XavierChanglingLi/Regularized-Optimal-Experience-Replay}.
Reinforcement Learning (RL) has made significant strides in complex tasks but struggles in multi-task settings with different embodiments. World model methods offer scalability by learning a simulation of the environment but often rely on inefficient gradient-free optimization methods for policy extraction. In contrast, gradient-based methods exhibit lower variance but fail to handle discontinuities. Our work reveals that well-regularized world models can generate smoother optimization landscapes than the actual dynamics, facilitating more effective first-order optimization. We introduce Policy learning with multi-task World Models (PWM), a novel model-based RL algorithm for continuous control. Initially, the world model is pre-trained on offline data, and then policies are extracted from it using first-order optimization in less than 10 minutes per task. PWM effectively solves tasks with up to 152 action dimensions and outperforms methods that use ground-truth dynamics. Additionally, PWM scales to an 80-task setting, achieving up to 27% higher rewards than existing baselines without relying on costly online planning. Visualizations and code are available at https://www.imgeorgiev.com/pwm/.
Trajectory planning in automated driving typically focuses on satisfying safety and comfort requirements within the vehicle's onboard sensor range. This paper introduces a method that leverages anticipatory road data, such as speed limits, road slopes, and traffic lights, beyond the local perception range to optimize energy-efficient braking trajectories. For that, coasting, which reduces energy consumption, and active braking are combined to transition from the current vehicle velocity to a lower target velocity at a given distance ahead. Finding the switching instants between the coasting phases and the continuous control for the braking phase is addressed as an optimal trade-off between maximizing coasting periods and minimizing braking effort. The resulting switched optimal control problem is solved by deriving necessary optimality conditions. To facilitate the incorporation of additional feasibility constraints for multi-phase trajectories, a sub-optimal alternative solution based on parametric optimization is proposed. Both methods are compared in simulation.
Dataset distillation aims to condense large datasets into a small number of synthetic examples that can be used as drop-in replacements when training new models. It has applications to interpretability, neural architecture search, privacy, and continual learning. Despite strong successes in supervised domains, such methods have not yet been extended to reinforcement learning, where the lack of a fixed dataset renders most distillation methods unusable. Filling the gap, we formalize behaviour distillation, a setting that aims to discover and then condense the information required for training an expert policy into a synthetic dataset of state-action pairs, without access to expert data. We then introduce Hallucinating Datasets with Evolution Strategies (HaDES), a method for behaviour distillation that can discover datasets of just four state-action pairs which, under supervised learning, train agents to competitive performance levels in continuous control tasks. We show that these datasets generalize out of distribution to training policies with a wide range of architectures and hyperparameters. We also demonstrate application to a downstream task, namely training multi-task agents in a zero-shot fashion. Beyond behaviour distillation, HaDES provides significant improvements in neuroevolution for RL over previous approaches and achieves SoTA results on one standard supervised dataset distillation task. Finally, we show that visualizing the synthetic datasets can provide human-interpretable task insights.
Floating wind turbines (FWTs) hold significant potential for the exploitation of offshore renewable energy resources. Nevertheless, prior to the construction of FWTs, it is imperative to tackle several critical challenges, especially the issue of performance degradation under combined wind and wave loads. This study initiates with the development of a simplified nonlinear dynamical model for a semi-submersible FWT. In particular, both the rotor dynamics and the finite rotations of the platform are considered in presented modeling approach, thereby effectively capturing the complex interplay between the platform, tower, nacelle, and rotor under combined wind and wave loads. Subsequently, based on the developed FWT model, a novel adaptive nonlinear pitch controller is formulated with the goal of striking a trade-off between regulating power generation and reducing platform motion. Notably, the proposed control strategy adopts a continuous control approach, strategically beneficial in circumventing the chattering phenomenon commonly associated with sliding mode control. Furthermore, the controller integrates an online approximator and a robust integral of the sign of the tracking error, facilitating real-time learning of system unknown dynamics while compensating for bounded disturbances. Finally, both the accuracy of the established nonlinear FWT model in predicting key dynamics and the superiority of the presented pitch controller are validated through comprehensive comparative studies.
Often times in imitation learning (IL), the environment we collect expert demonstrations in and the environment we want to deploy our learned policy in aren't exactly the same (e.g. demonstrations collected in simulation but deployment in the real world). Compared to policy-centric approaches to IL like behavioural cloning, reward-centric approaches like inverse reinforcement learning (IRL) often better replicate expert behaviour in new environments. This transfer is usually performed by optimising the recovered reward under the dynamics of the target environment. However, (a) we find that modern deep IL algorithms frequently recover rewards which induce policies far weaker than the expert, even in the same environment the demonstrations were collected in. Furthermore, (b) these rewards are often quite poorly shaped, necessitating extensive environment interaction to optimise effectively. We provide simple and scalable fixes to both of these concerns. For (a), we find that reward model ensembles combined with a slightly different training objective significantly improves re-training and transfer performance. For (b), we propose a novel evolution-strategies based method EvIL to optimise for a reward-shaping term that speeds up re-training in the target environment, closing a gap left open by the classical theory of IRL. On a suite of continuous control tasks, we are able to re-train policies in target (and source) environments more interaction-efficiently than prior work.
Reinforcement learning (RL) agents are commonly trained and evaluated in the same environment. In contrast, humans often train in a specialized environment before being evaluated, such as studying a book before taking an exam. The potential of such specialized training environments is still vastly underexplored, despite their capacity to dramatically speed up training. The framework of synthetic environments takes a first step in this direction by meta-learning neural network-based Markov decision processes (MDPs). The initial approach was limited to toy problems and produced environments that did not transfer to unseen RL algorithms. We extend this approach in three ways: Firstly, we modify the meta-learning algorithm to discover environments invariant towards hyperparameter configurations and learning algorithms. Secondly, by leveraging hardware parallelism and introducing a curriculum on an agent's evaluation episode horizon, we can achieve competitive results on several challenging continuous control problems. Thirdly, we surprisingly find that contextual bandits enable training RL agents that transfer well to their evaluation environment, even if it is a complex MDP. Hence, we set up our experiments to train synthetic contextual bandits, which perform on par with synthetic MDPs, yield additional insights into the evaluation environment, and can speed up downstream applications.