Learning from small data sets is critical in many practical applications where data collection is time consuming or expensive, e.g., robotics, animal experiments or drug design. Meta learning is one way to increase the data efficiency of learning algorithms by generalizing learned concepts from a set of training tasks to unseen, but related, tasks. Often, this relationship between tasks is hard coded or relies in some other way on human expertise. In this paper, we frame meta learning as a hierarchical latent variable model and infer the relationship between tasks automatically from data. We apply our framework in a model-based reinforcement learning setting and show that our meta-learning model effectively generalizes to novel tasks by identifying how new tasks relate to prior ones from minimal data. This results in up to a 60% reduction in the average interaction time needed to solve tasks compared to strong baselines.
Research on reinforcement learning has demonstrated promising results in manifold applications and domains. Still, efficiently learning effective robot behaviors is very difficult, due to unstructured scenarios, high uncertainties, and large state dimensionality (e.g. multi-agent systems or hyper-redundant robots). To alleviate this problem, we present DOP, a deep model-based reinforcement learning algorithm, which exploits action values to both (1) guide the exploration of the state space and (2) plan effective policies. Specifically, we exploit deep neural networks to learn Q-functions that are used to attack the curse of dimensionality during a Monte-Carlo tree search. Our algorithm, in fact, constructs upper confidence bounds on the learned value function to select actions optimistically. We implement and evaluate DOP on different scenarios: (1) a cooperative navigation problem, (2) a fetching task for a 7-DOF KUKA robot, and (3) a human-robot handover with a humanoid robot (both in simulation and real). The obtained results show the effectiveness of DOP in the chosen applications, where action values drive the exploration and reduce the computational demand of the planning process while achieving good performance.
Reinforcement learning has emerged as a promising methodology for training robot controllers. However, most results have been limited to simulation due to the need for a large number of samples and the lack of automated-yet-safe data collection methods. Model-based reinforcement learning methods provide an avenue to circumvent these challenges, but the traditional concern has been the mismatch between the simulator and the real world. Here, we show that control policies learned in simulation can successfully transfer to a physical system, composed of three Phantom robots pushing an object to various desired target positions. We use a modified form of the natural policy gradient algorithm for learning, applied to a carefully identified simulation model. The resulting policies, trained entirely in simulation, work well on the physical system without additional training. In addition, we show that training with an ensemble of models makes the learned policies more robust to modeling errors, thus compensating for difficulties in system identification.
Consider an assistive system that guides visually impaired users through speech and haptic feedback to their destination. Existing robotic and ubiquitous navigation technologies (e.g., portable, ground, or wearable systems) often operate in a generic, user-agnostic manner. However, to minimize confusion and navigation errors, our real-world analysis reveals a crucial need to adapt the instructional guidance across different end-users with diverse mobility skills. To address this practical issue in scalable system design, we propose a novel model-based reinforcement learning framework for personalizing the system-user interaction experience. When incrementally adapting the system to new users, we propose to use a weighted experts model for addressing data-efficiency limitations in transfer learning with deep models. A real-world dataset of navigation by blind users is used to show that the proposed approach allows for (1) more accurate long-term human behavior prediction (up to 20 seconds into the future) through improved reasoning over personal mobility characteristics, interaction with surrounding obstacles, and the current navigation goal, and (2) quick adaptation at the onset of learning, when data is limited.
In model-based reinforcement learning, generative and temporal models of environments can be leveraged to boost agent performance, either by tuning the agent's representations during training or via use as part of an explicit planning mechanism. However, their application in practice has been limited to simplistic environments, due to the difficulty of training such models in larger, potentially partially-observed and 3D environments. In this work we introduce a novel action-conditioned generative model of such challenging environments. The model features a non-parametric spatial memory system in which we store learned, disentangled representations of the environment. Low-dimensional spatial updates are computed using a state-space model that makes use of knowledge on the prior dynamics of the moving agent, and high-dimensional visual observations are modelled with a Variational Auto-Encoder. The result is a scalable architecture capable of performing coherent predictions over hundreds of time steps across a range of partially observed 2D and 3D environments.
Inspired by recent successes of Monte-Carlo tree search (MCTS) in a number of artificial intelligence (AI) application domains, we propose a model-based reinforcement learning (RL) technique that iteratively applies MCTS on batches of small, finite-horizon versions of the original infinite-horizon Markov decision process. The terminal condition of the finite-horizon problems, or the leaf-node evaluator of the decision tree generated by MCTS, is specified using a combination of an estimated value function and an estimated policy function. The recommendations generated by the MCTS procedure are then provided as feedback in order to refine, through classification and regression, the leaf-node evaluator for the next iteration. We provide the first sample complexity bounds for a tree search-based RL algorithm. In addition, we show that a deep neural network implementation of the technique can create a competitive AI agent for the popular multi-player online battle arena (MOBA) game King of Glory.
Model-based reinforcement learning (MBRL) has been proposed as a promising alternative solution to tackle the high sampling cost challenge in the canonical reinforcement learning (RL), by leveraging a learned model to generate synthesized data for policy training purpose. The MBRL framework, nevertheless, is inherently limited by the convoluted process of jointly learning control policy and configuring hyper-parameters (e.g., global/local models, real and synthesized data, etc). The training process could be tedious and prohibitively costly. In this research, we propose an "reinforcement on reinforcement" (RoR) architecture to decompose the convoluted tasks into two layers of reinforcement learning. The inner layer is the canonical model-based RL training process environment (TPE), which learns the control policy for the underlying system and exposes interfaces to access states, actions and rewards. The outer layer presents an RL agent, called as AI trainer, to learn an optimal hyper-parameter configuration for the inner TPE. This decomposition approach provides a desirable flexibility to implement different trainer designs, called as "train the trainer". In our research, we propose and optimize two alternative trainer designs: 1) a uni-head trainer and 2) a multi-head trainer. Our proposed RoR framework is evaluated for five tasks in the OpenAI gym (i.e., Pendulum, Mountain Car, Reacher, Half Cheetah and Swimmer). Compared to three other baseline algorithms, our proposed Train-the-Trainer algorithm has a competitive performance in auto-tuning capability, with upto 56% expected sampling cost saving without knowing the best parameter setting in advance. The proposed trainer framework can be easily extended to other cases in which the hyper-parameter tuning is costly.
Model-based reinforcement learning (RL) algorithms can attain excellent sample efficiency, but often lag behind the best model-free algorithms in terms of asymptotic performance. This is especially true with high-capacity parametric function approximators, such as deep networks. In this paper, we study how to bridge this gap, by employing uncertainty-aware dynamics models. We propose a new algorithm called probabilistic ensembles with trajectory sampling (PETS) that combines uncertainty-aware deep network dynamics models with sampling-based uncertainty propagation. Our comparison to state-of-the-art model-based and model-free deep RL algorithms shows that our approach matches the asymptotic performance of model-free algorithms on several challenging benchmark tasks, while requiring significantly fewer samples (e.g., 8 and 125 times fewer samples than Soft Actor Critic and Proximal Policy Optimization respectively on the half-cheetah task).
Although reinforcement learning methods can achieve impressive results in simulation, the real world presents two major challenges: generating samples is exceedingly expensive, and unexpected perturbations or unseen situations cause proficient but specialized policies to fail at test time. Given that it is impractical to train separate policies to accommodate all situations the agent may see in the real world, this work proposes to learn how to quickly and effectively adapt online to new tasks. To enable sample-efficient learning, we consider learning online adaptation in the context of model-based reinforcement learning. Our approach uses meta-learning to train a dynamics model prior such that, when combined with recent data, this prior can be rapidly adapted to the local context. Our experiments demonstrate online adaptation for continuous control tasks on both simulated and real-world agents. We first show simulated agents adapting their behavior online to novel terrains, crippled body parts, and highly-dynamic environments. We also illustrate the importance of incorporating online adaptation into autonomous agents that operate in the real world by applying our method to a real dynamic legged millirobot. We demonstrate the agent's learned ability to quickly adapt online to a missing leg, adjust to novel terrains and slopes, account for miscalibration or errors in pose estimation, and compensate for pulling payloads.
Humans learn to play video games significantly faster than the state-of-the-art reinforcement learning (RL) algorithms. People seem to build simple models that are easy to learn to support planning and strategic exploration. Inspired by this, we investigate two issues in leveraging model-based RL for sample efficiency. First we investigate how to perform strategic exploration when exact planning is not feasible and empirically show that optimistic Monte Carlo Tree Search outperforms posterior sampling methods. Second we show how to learn simple deterministic models to support fast learning using object representation. We illustrate the benefit of these ideas by introducing a novel algorithm, Strategic Object Oriented Reinforcement Learning (SOORL), that outperforms state-of-the-art algorithms in the game of Pitfall! in less than 50 episodes.
Learning a generative model is a key component of model-based reinforcement learning. Though learning a good model in the tabular setting is a simple task, learning a useful model in the approximate setting is challenging. In this context, an important question is the loss function used for model learning as varying the loss function can have a remarkable impact on effectiveness of planning. Recently Farahmand et al. (2017) proposed a value-aware model learning (VAML) objective that captures the structure of value function during model learning. Using tools from Asadi et al. (2018), we show that minimizing the VAML objective is in fact equivalent to minimizing the Wasserstein metric. This equivalence improves our understanding of value-aware models, and also creates a theoretical foundation for applications of Wasserstein in model-based reinforcement~learning.
We examine the impact of learning Lipschitz continuous models in the context of model-based reinforcement learning. We provide a novel bound on multi-step prediction error of Lipschitz models where we quantify the error using the Wasserstein metric. We go on to prove an error bound for the value-function estimate arising from Lipschitz models and show that the estimated value function is itself Lipschitz. We conclude with empirical results that show the benefits of controlling the Lipschitz constant of neural-network models.
Dyna is a fundamental approach to model-based reinforcement learning (MBRL) that interleaves planning, acting, and learning in an online setting. In the most typical application of Dyna, the dynamics model is used to generate one-step transitions from selected start states from the agent's history, which are used to update the agent's value function or policy as if they were real experiences. In this work, one-step Dyna was applied to several games from the Arcade Learning Environment (ALE). We found that the model-based updates offered surprisingly little benefit over simply performing more updates with the agent's existing experience, even when using a perfect model. We hypothesize that to get the most from planning, the model must be used to generate unfamiliar experience. To test this, we experimented with the "shape" of planning in multiple different concrete instantiations of Dyna, performing fewer, longer rollouts, rather than many short rollouts. We found that planning shape has a profound impact on the efficacy of Dyna for both perfect and learned models. In addition to these findings regarding Dyna in general, our results represent, to our knowledge, the first time that a learned dynamics model has been successfully used for planning in the ALE, suggesting that Dyna may be a viable approach to MBRL in the ALE and other high-dimensional problems.
In model-based reinforcement learning it is typical to decouple the problems of learning the dynamics model and learning the reward function. However, when the dynamics model is flawed, it may generate erroneous states that would never occur in the true environment. It is not clear a priori what value the reward function should assign to such states. This paper presents a novel error bound that accounts for the reward model's behavior in states sampled from the model. This bound is used to extend the existing Hallucinated DAgger-MC algorithm, which offers theoretical performance guarantees in deterministic MDPs that do not assume a perfect model can be learned. Empirically, this approach to reward learning can yield dramatic improvements in control performance when the dynamics model is flawed.
This paper presents a method for identifying mechanical parameters of robots or objects, such as their mass and friction coefficients. Key features are the use of off-the-shelf physics engines and the adaptation of a Bayesian optimization technique towards minimizing the number of real-world experiments needed for model-based reinforcement learning. The proposed framework reproduces in a physics engine experiments performed on a real robot and optimizes the model's mechanical parameters so as to match real-world trajectories. The optimized model is then used for learning a policy in simulation, before real-world deployment. It is well understood, however, that it is hard to exactly reproduce real trajectories in simulation. Moreover, a near-optimal policy can be frequently found with an imperfect model. Therefore, this work proposes a strategy for identifying a model that is just good enough to approximate the value of a locally optimal policy with a certain confidence, instead of wasting effort on identifying the most accurate model. Evaluations, performed both in simulation and on a real robotic manipulation task, indicate that the proposed strategy results in an overall time-efficient, integrated model identification and learning solution, which significantly improves the data-efficiency of existing policy search algorithms.