Imitation learning (IL) is a framework that learns to imitate expert behavior from demonstrations. Recently, IL shows promising results on high dimensional and control tasks. However, IL typically suffers from sample inefficiency in terms of environment interaction, which severely limits their application to simulated domains. In industrial applications, learner usually have a high interaction cost, the more interactions with environment, the more damage it causes to the environment and the learner itself. In this article, we make an effort to improve sample efficiency by introducing a novel scheme of inverse reinforcement learning. Our method, which we call \textit{Model Reward Function Based Imitation Learning} (MRFIL), uses an ensemble dynamic model as a reward function, what is trained with expert demonstrations. The key idea is to provide the agent with an incentive to match the demonstrations over a long horizon, by providing a positive reward upon encountering states in line with the expert demonstration distribution. In addition, we demonstrate the convergence guarantee for new objective function. Experimental results show that our algorithm reaches the competitive performance and significantly reducing the environment interactions compared to IL methods.
相關內容
Most existing imitation learning approaches assume the demonstrations are drawn from experts who are optimal, but relaxing this assumption enables us to use a wider range of data. Standard imitation learning may learn a suboptimal policy from demonstrations with varying optimality. Prior works use confidence scores or rankings to capture beneficial information from demonstrations with varying optimality, but they suffer from many limitations, e.g., manually annotated confidence scores or high average optimality of demonstrations. In this paper, we propose a general framework to learn from demonstrations with varying optimality that jointly learns the confidence score and a well-performing policy. Our approach, Confidence-Aware Imitation Learning (CAIL) learns a well-performing policy from confidence-reweighted demonstrations, while using an outer loss to track the performance of our model and to learn the confidence. We provide theoretical guarantees on the convergence of CAIL and evaluate its performance in both simulated and real robot experiments. Our results show that CAIL significantly outperforms other imitation learning methods from demonstrations with varying optimality. We further show that even without access to any optimal demonstrations, CAIL can still learn a successful policy, and outperforms prior work.
Moving away from repetitive tasks, robots nowadays demand versatile skills that adapt to different situations. Task-parameterized learning improves the generalization of motion policies by encoding relevant contextual information in the task parameters, hence enabling flexible task executions. However, training such a policy often requires collecting multiple demonstrations in different situations. To comprehensively create different situations is non-trivial thus renders the method less applicable to real-world problems. Therefore, training with fewer demonstrations/situations is desirable. This paper presents a novel concept to augment the original training dataset with synthetic data for policy improvements, thus allows learning task-parameterized skills with few demonstrations.
This paper develops the Continuous Pontryagin Differentiable Programming (Continuous PDP) method that enables a robot to learn a control utility function from a few number of sparsely demonstrated keyframes. The keyframes are few desired sequential outputs that a robot is wanted to follow at certain time instances. The duration of the keyframes may be different from that of the robot actual execution. The method jointly searches for a robot control utility function and a time-warping function such that the robot motion sequentially follows the given keyframes with minimal discrepancy loss. Continuous PDP minimizes the discrepancy loss using projected gradient descent, by efficiently solving the gradient of robot motion with respect to the unknown parameters. The method is first evaluated on a simulated two-link robot arm, and then applied to a 6-DoF maneuvering quadrotor to learn a utility function from keyframes for its motion planning in un-modeled environments with obstacles. The results show the efficiency of the method, its ability to handle time misalignment between keyframes and robot execution, and the generalization of the learned utility function into unseen motion conditions.
We study constrained reinforcement learning (CRL) from a novel perspective by setting constraints directly on state density functions, rather than the value functions considered by previous works. State density has a clear physical and mathematical interpretation, and is able to express a wide variety of constraints such as resource limits and safety requirements. Density constraints can also avoid the time-consuming process of designing and tuning cost functions required by value function-based constraints to encode system specifications. We leverage the duality between density functions and Q functions to develop an effective algorithm to solve the density constrained RL problem optimally and the constrains are guaranteed to be satisfied. We prove that the proposed algorithm converges to a near-optimal solution with a bounded error even when the policy update is imperfect. We use a set of comprehensive experiments to demonstrate the advantages of our approach over state-of-the-art CRL methods, with a wide range of density constrained tasks as well as standard CRL benchmarks such as Safety-Gym.
We address the issue of tuning hyperparameters (HPs) for imitation learning algorithms in the context of continuous-control, when the underlying reward function of the demonstrating expert cannot be observed at any time. The vast literature in imitation learning mostly considers this reward function to be available for HP selection, but this is not a realistic setting. Indeed, would this reward function be available, it could then directly be used for policy training and imitation would not be necessary. To tackle this mostly ignored problem, we propose a number of possible proxies to the external reward. We evaluate them in an extensive empirical study (more than 10'000 agents across 9 environments) and make practical recommendations for selecting HPs. Our results show that while imitation learning algorithms are sensitive to HP choices, it is often possible to select good enough HPs through a proxy to the reward function.
Recent successes of value-based multi-agent deep reinforcement learning employ optimism in value function by carefully controlling learning rate(Omidshafiei et al., 2017) or reducing update prob-ability (Palmer et al., 2018). We introduce a de-centralized quantile estimator: Responsible Implicit Quantile Network (RIQN), while robust to teammate-environment interactions, able to reduce the amount of imposed optimism. Upon benchmarking against related Hysteretic-DQN(HDQN) and Lenient-DQN (LDQN), we findRIQN agents more stable, sample efficient and more likely to converge to the optimal policy.
This paper tackles a new problem setting: reinforcement learning with pixel-wise rewards (pixelRL) for image processing. After the introduction of the deep Q-network, deep RL has been achieving great success. However, the applications of deep RL for image processing are still limited. Therefore, we extend deep RL to pixelRL for various image processing applications. In pixelRL, each pixel has an agent, and the agent changes the pixel value by taking an action. We also propose an effective learning method for pixelRL that significantly improves the performance by considering not only the future states of the own pixel but also those of the neighbor pixels. The proposed method can be applied to some image processing tasks that require pixel-wise manipulations, where deep RL has never been applied. We apply the proposed method to three image processing tasks: image denoising, image restoration, and local color enhancement. Our experimental results demonstrate that the proposed method achieves comparable or better performance, compared with the state-of-the-art methods based on supervised learning.
Eligibility traces are an effective technique to accelerate reinforcement learning by smoothly assigning credit to recently visited states. However, their online implementation is incompatible with modern deep reinforcement learning algorithms, which rely heavily on i.i.d. training data and offline learning. We utilize an efficient, recursive method for computing {\lambda}-returns offline that can provide the benefits of eligibility traces to any value-estimation or actor-critic method. We demonstrate how our method can be combined with DQN, DRQN, and A3C to greatly enhance the learning speed of these algorithms when playing Atari 2600 games, even under partial observability. Our results indicate several-fold improvements to sample efficiency on Seaquest and Q*bert. We expect similar results for other algorithms and domains not considered here, including those with continuous actions.
Recent studies have shown the vulnerability of reinforcement learning (RL) models in noisy settings. The sources of noises differ across scenarios. For instance, in practice, the observed reward channel is often subject to noise (e.g., when observed rewards are collected through sensors), and thus observed rewards may not be credible as a result. Also, in applications such as robotics, a deep reinforcement learning (DRL) algorithm can be manipulated to produce arbitrary errors. In this paper, we consider noisy RL problems where observed rewards by RL agents are generated with a reward confusion matrix. We call such observed rewards as perturbed rewards. We develop an unbiased reward estimator aided robust RL framework that enables RL agents to learn in noisy environments while observing only perturbed rewards. Our framework draws upon approaches for supervised learning with noisy data. The core ideas of our solution include estimating a reward confusion matrix and defining a set of unbiased surrogate rewards. We prove the convergence and sample complexity of our approach. Extensive experiments on different DRL platforms show that policies based on our estimated surrogate reward can achieve higher expected rewards, and converge faster than existing baselines. For instance, the state-of-the-art PPO algorithm is able to obtain 67.5% and 46.7% improvements in average on five Atari games, when the error rates are 10% and 30% respectively.
This paper proposes a Reinforcement Learning (RL) algorithm to synthesize policies for a Markov Decision Process (MDP), such that a linear time property is satisfied. We convert the property into a Limit Deterministic Buchi Automaton (LDBA), then construct a product MDP between the automaton and the original MDP. A reward function is then assigned to the states of the product automaton, according to accepting conditions of the LDBA. With this reward function, our algorithm synthesizes a policy that satisfies the linear time property: as such, the policy synthesis procedure is "constrained" by the given specification. Additionally, we show that the RL procedure sets up an online value iteration method to calculate the maximum probability of satisfying the given property, at any given state of the MDP - a convergence proof for the procedure is provided. Finally, the performance of the algorithm is evaluated via a set of numerical examples. We observe an improvement of one order of magnitude in the number of iterations required for the synthesis compared to existing approaches.
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191