Robotic skills can be learned via imitation learning (IL) using user-provided demonstrations, or via reinforcement learning (RL) using large amountsof autonomously collected experience.Both methods have complementarystrengths and weaknesses: RL can reach a high level of performance, but requiresexploration, which can be very time consuming and unsafe; IL does not requireexploration, but only learns skills that are as good as the provided demonstrations.Can a single method combine the strengths of both approaches? A number ofprior methods have aimed to address this question, proposing a variety of tech-niques that integrate elements of IL and RL. However, scaling up such methodsto complex robotic skills that integrate diverse offline data and generalize mean-ingfully to real-world scenarios still presents a major challenge. In this paper, ouraim is to test the scalability of prior IL + RL algorithms and devise a system basedon detailed empirical experimentation that combines existing components in themost effective and scalable way. To that end, we present a series of experimentsaimed at understanding the implications of each design decision, so as to develop acombined approach that can utilize demonstrations and heterogeneous prior datato attain the best performance on a range of real-world and realistic simulatedrobotic problems. Our complete method, which we call AW-Opt, combines ele-ments of advantage-weighted regression [1, 2] and QT-Opt [3], providing a unifiedapproach for integrating demonstrations and offline data for robotic manipulation.Please see https://awopt.github.io for more details.
Reinforcement learning can train policies that effectively perform complex tasks. However for long-horizon tasks, the performance of these methods degrades with horizon, often necessitating reasoning over and composing lower-level skills. Hierarchical reinforcement learning aims to enable this by providing a bank of low-level skills as action abstractions. Hierarchies can further improve on this by abstracting the space states as well. We posit that a suitable state abstraction should depend on the capabilities of the available lower-level policies. We propose Value Function Spaces: a simple approach that produces such a representation by using the value functions corresponding to each lower-level skill. These value functions capture the affordances of the scene, thus forming a representation that compactly abstracts task relevant information and robustly ignores distractors. Empirical evaluations for maze-solving and robotic manipulation tasks demonstrate that our approach improves long-horizon performance and enables better zero-shot generalization than alternative model-free and model-based methods.
The aim in imitation learning is to learn effective policies by utilizing near-optimal expert demonstrations. However, high-quality demonstrations from human experts can be expensive to obtain in large numbers. On the other hand, it is often much easier to obtain large quantities of suboptimal or task-agnostic trajectories, which are not useful for direct imitation, but can nevertheless provide insight into the dynamical structure of the environment, showing what could be done in the environment even if not what should be done. We ask the question, is it possible to utilize such suboptimal offline datasets to facilitate provably improved downstream imitation learning? In this work, we answer this question affirmatively and present training objectives that use offline datasets to learn a factored transition model whose structure enables the extraction of a latent action space. Our theoretical analysis shows that the learned latent action space can boost the sample-efficiency of downstream imitation learning, effectively reducing the need for large near-optimal expert datasets through the use of auxiliary non-expert data. To learn the latent action space in practice, we propose TRAIL (Transition-Reparametrized Actions for Imitation Learning), an algorithm that learns an energy-based transition model contrastively, and uses the transition model to reparametrize the action space for sample-efficient imitation learning. We evaluate the practicality of our objective through experiments on a set of navigation and locomotion tasks. Our results verify the benefits suggested by our theory and show that TRAIL is able to improve baseline imitation learning by up to 4x in performance.
The recent history of machine learning research has taught us that machine learning methods can be most effective when they are provided with very large, high-capacity models, and trained on very large and diverse datasets. This has spurred the community to search for ways to remove any bottlenecks to scale. Often the foremost among such bottlenecks is the need for human effort, including the effort of curating and labeling datasets. As a result, considerable attention in recent years has been devoted to utilizing unlabeled data, which can be collected in vast quantities. However, some of the most widely used methods for training on such unlabeled data themselves require human-designed objective functions that must correlate in some meaningful way to downstream tasks. I will argue that a general, principled, and powerful framework for utilizing unlabeled data can be derived from reinforcement learning, using general purpose unsupervised or self-supervised reinforcement learning objectives in concert with offline reinforcement learning methods that can leverage large datasets. I will discuss how such a procedure is more closely aligned with potential downstream tasks, and how it could build on existing techniques that have been developed in recent years.
Goal-conditioned reinforcement learning (RL) can solve tasks in a wide range of domains, including navigation and manipulation, but learning to reach distant goals remains a central challenge to the field. Learning to reach such goals is particularly hard without any offline data, expert demonstrations, and reward shaping. In this paper, we propose an algorithm to solve the distant goal-reaching task by using search at training time to automatically generate a curriculum of intermediate states. Our algorithm, Classifier-Planning (C-Planning), frames the learning of the goal-conditioned policies as expectation maximization: the E-step corresponds to planning an optimal sequence of waypoints using graph search, while the M-step aims to learn a goal-conditioned policy to reach those waypoints. Unlike prior methods that combine goal-conditioned RL with graph search, ours performs search only during training and not testing, significantly decreasing the compute costs of deploying the learned policy. Empirically, we demonstrate that our method is more sample efficient than prior methods. Moreover, it is able to solve very long horizons manipulation and navigation tasks, tasks that prior goal-conditioned methods and methods based on graph search fail to solve.
Industry has gradually moved towards application-specific hardware accelerators in order to attain higher efficiency. While such a paradigm shift is already starting to show promising results, designers need to spend considerable manual effort and perform a large number of time-consuming simulations to find accelerators that can accelerate multiple target applications while obeying design constraints. Moreover, such a "simulation-driven" approach must be re-run from scratch every time the set of target applications or design constraints change. An alternative paradigm is to use a "data-driven", offline approach that utilizes logged simulation data, to architect hardware accelerators, without needing any form of simulations. Such an approach not only alleviates the need to run time-consuming simulation, but also enables data reuse and applies even when set of target applications changes. In this paper, we develop such a data-driven offline optimization method for designing hardware accelerators, dubbed PRIME, that enjoys all of these properties. Our approach learns a conservative, robust estimate of the desired cost function, utilizes infeasible points, and optimizes the design against this estimate without any additional simulator queries during optimization. PRIME architects accelerators -- tailored towards both single and multiple applications -- improving performance upon state-of-the-art simulation-driven methods by about 1.54x and 1.20x, while considerably reducing the required total simulation time by 93% and 99%, respectively. In addition, PRIME also architects effective accelerators for unseen applications in a zero-shot setting, outperforming simulation-based methods by 1.26x.
While deep neural networks can attain good accuracy on in-distribution test points, many applications require robustness even in the face of unexpected perturbations in the input, changes in the domain, or other sources of distribution shift. We study the problem of test time robustification, i.e., using the test input to improve model robustness. Recent prior works have proposed methods for test time adaptation, however, they each introduce additional assumptions, such as access to multiple test points, that prevent widespread adoption. In this work, we aim to study and devise methods that make no assumptions about the model training process and are broadly applicable at test time. We propose a simple approach that can be used in any test setting where the model is probabilistic and adaptable: when presented with a test example, perform different data augmentations on the data point, and then adapt (all of) the model parameters by minimizing the entropy of the model's average, or marginal, output distribution across the augmentations. Intuitively, this objective encourages the model to make the same prediction across different augmentations, thus enforcing the invariances encoded in these augmentations, while also maintaining confidence in its predictions. In our experiments, we demonstrate that this approach consistently improves robust ResNet and vision transformer models, achieving accuracy gains of 1-8% over standard model evaluation and also generally outperforming prior augmentation and adaptation strategies. We achieve state-of-the-art results for test shifts caused by image corruptions (ImageNet-C), renditions of common objects (ImageNet-R), and, among ResNet-50 models, adversarially chosen natural examples (ImageNet-A).
Reinforcement learning (RL) can in principle make it possible for robots to automatically adapt to new tasks, but in practice current RL methods require a very large number of trials to accomplish this. In this paper, we tackle rapid adaptation to new tasks through the framework of meta-learning, which utilizes past tasks to learn to adapt, with a specific focus on industrial insertion tasks. We address two specific challenges by applying meta-learning in this setting. First, conventional meta-RL algorithms require lengthy online meta-training phases. We show that this can be replaced with appropriately chosen offline data, resulting in an offline meta-RL method that only requires demonstrations and trials from each of the prior tasks, without the need to run costly meta-RL procedures online. Second, meta-RL methods can fail to generalize to new tasks that are too different from those seen at meta-training time, which poses a particular challenge in industrial applications, where high success rates are critical. We address this by combining contextual meta-learning with direct online finetuning: if the new task is similar to those seen in the prior data, then the contextual meta-learner adapts immediately, and if it is too different, it gradually adapts through finetuning. We show that our approach is able to quickly adapt to a variety of different insertion tasks, learning how to perform them with a success rate of 100% using only a fraction of the samples needed for learning the tasks from scratch. Experiment videos and details are available at https://sites.google.com/view/offline-metarl-insertion.
Offline reinforcement learning requires reconciling two conflicting aims: learning a policy that improves over the behavior policy that collected the dataset, while at the same time minimizing the deviation from the behavior policy so as to avoid errors due to distributional shift. This trade-off is critical, because most current offline reinforcement learning methods need to query the value of unseen actions during training to improve the policy, and therefore need to either constrain these actions to be in-distribution, or else regularize their values. We propose an offline RL method that never needs to evaluate actions outside of the dataset, but still enables the learned policy to improve substantially over the best behavior in the data through generalization. The main insight in our work is that, instead of evaluating unseen actions from the latest policy, we can approximate the policy improvement step implicitly by treating the state value function as a random variable, with randomness determined by the action (while still integrating over the dynamics to avoid excessive optimism), and then taking a state conditional upper expectile of this random variable to estimate the value of the best actions in that state. This leverages the generalization capacity of the function approximator to estimate the value of the best available action at a given state without ever directly querying a Q-function with this unseen action. Our algorithm alternates between fitting this upper expectile value function and backing it up into a Q-function. Then, we extract the policy via advantage-weighted behavioral cloning. We dub our method implicit Q-learning (IQL). IQL demonstrates the state-of-the-art performance on D4RL, a standard benchmark for offline reinforcement learning. We also demonstrate that IQL achieves strong performance fine-tuning using online interaction after offline initialization.
Legged robots are physically capable of traversing a wide range of challenging environments, but designing controllers that are sufficiently robust to handle this diversity has been a long-standing challenge in robotics. Reinforcement learning presents an appealing approach for automating the controller design process and has been able to produce remarkably robust controllers when trained in a suitable range of environments. However, it is difficult to predict all likely conditions the robot will encounter during deployment and enumerate them at training-time. What if instead of training controllers that are robust enough to handle any eventuality, we enable the robot to continually learn in any setting it finds itself in? This kind of real-world reinforcement learning poses a number of challenges, including efficiency, safety, and autonomy. To address these challenges, we propose a practical robot reinforcement learning system for fine-tuning locomotion policies in the real world. We demonstrate that a modest amount of real-world training can substantially improve performance during deployment, and this enables a real A1 quadrupedal robot to autonomously fine-tune multiple locomotion skills in a range of environments, including an outdoor lawn and a variety of indoor terrains.