Large language models can encode a wealth of semantic knowledge about the world. Such knowledge could be extremely useful to robots aiming to act upon high-level, temporally extended instructions expressed in natural language. However, a significant weakness of language models is that they lack real-world experience, which makes it difficult to leverage them for decision making within a given embodiment. For example, asking a language model to describe how to clean a spill might result in a reasonable narrative, but it may not be applicable to a particular agent, such as a robot, that needs to perform this task in a particular environment. We propose to provide real-world grounding by means of pretrained skills, which are used to constrain the model to propose natural language actions that are both feasible and contextually appropriate. The robot can act as the language model's "hands and eyes," while the language model supplies high-level semantic knowledge about the task. We show how low-level skills can be combined with large language models so that the language model provides high-level knowledge about the procedures for performing complex and temporally-extended instructions, while value functions associated with these skills provide the grounding necessary to connect this knowledge to a particular physical environment. We evaluate our method on a number of real-world robotic tasks, where we show the need for real-world grounding and that this approach is capable of completing long-horizon, abstract, natural language instructions on a mobile manipulator. The project's website and the video can be found at https://say-can.github.io/
Large foundation models can exhibit unique capabilities depending on the domain of data they are trained on. While these domains are generic, they may only barely overlap. For example, visual-language models (VLMs) are trained on Internet-scale image captions, but large language models (LMs) are further trained on Internet-scale text with no images (e.g. from spreadsheets, to SAT questions). As a result, these models store different forms of commonsense knowledge across different domains. In this work, we show that this model diversity is symbiotic, and can be leveraged to build AI systems with structured Socratic dialogue -- in which new multimodal tasks are formulated as a guided language-based exchange between different pre-existing foundation models, without additional finetuning. In the context of egocentric perception, we present a case study of Socratic Models (SMs) that can provide meaningful results for complex tasks such as generating free-form answers to contextual questions about egocentric video, by formulating video Q&A as short story Q&A, i.e. summarizing the video into a short story, then answering questions about it. Additionally, SMs can generate captions for Internet images, and are competitive with state-of-the-art on zero-shot video-to-text retrieval with 42.8 R@1 on MSR-VTT 1k-A. SMs demonstrate how to compose foundation models zero-shot to capture new multimodal functionalities, without domain-specific data collection. Prototypes are available at socraticmodels.github.io.
Efficiently finding an occluded object with lateral access arises in many contexts such as warehouses, retail, healthcare, shipping, and homes. We introduce LAX-RAY (Lateral Access maXimal Reduction of occupancY support Area), a system to automate the mechanical search for occluded objects on shelves. For such lateral access environments, LAX-RAY couples a perception pipeline predicting a target object occupancy support distribution with a mechanical search policy that sequentially selects occluding objects to push to the side to reveal the target as efficiently as possible. Within the context of extruded polygonal objects and a stationary target with a known aspect ratio, we explore three lateral access search policies: Distribution Area Reduction (DAR), Distribution Entropy Reduction (DER), and Distribution Entropy Reduction over Multiple Time Steps (DER-MT) utilizing the support distribution and prior information. We evaluate these policies using the First-Order Shelf Simulator (FOSS) in which we simulate 800 random shelf environments of varying difficulty, and in a physical shelf environment with a Fetch robot and an embedded PrimeSense RGBD Camera. Average simulation results of 87.3% success rate demonstrate better performance of DER-MT with 2 prediction steps. When deployed on the robot, results show a success rate of at least 80% for all policies, suggesting that LAX-RAY can efficiently reveal the target object in reality. Both results show significantly better performance of the three proposed policies compared to a baseline policy with uniform probability distribution assumption in non-trivial cases, showing the importance of distribution prediction. Code, videos, and supplementary material can be found at https://sites.google.com/berkeley.edu/lax-ray.
Mapping and localization, preferably from a small number of observations, are fundamental tasks in robotics. We address these tasks by combining spatial structure (differentiable mapping) and end-to-end learning in a novel neural network architecture: the Differentiable Mapping Network (DMN). The DMN constructs a spatially structured view-embedding map and uses it for subsequent visual localization with a particle filter. Since the DMN architecture is end-to-end differentiable, we can jointly learn the map representation and localization using gradient descent. We apply the DMN to sparse visual localization, where a robot needs to localize in a new environment with respect to a small number of images from known viewpoints. We evaluate the DMN using simulated environments and a challenging real-world Street View dataset. We find that the DMN learns effective map representations for visual localization. The benefit of spatial structure increases with larger environments, more viewpoints for mapping, and when training data is scarce. Project website: http://sites.google.com/view/differentiable-mapping
For applications in e-commerce, warehouses, healthcare, and home service, robots are often required to search through heaps of objects to grasp a specific target object. For mechanical search, we introduce X-Ray, an algorithm based on learned occupancy distributions. We train a neural network using a synthetic dataset of RGBD heap images labeled for a set of standard bounding box targets with varying aspect ratios. X-Ray minimizes support of the learned distribution as part of a mechanical search policy in both simulated and real environments. We benchmark these policies against two baseline policies on 1,000 heaps of 15 objects in simulation where the target object is partially or fully occluded. Results suggest that X-Ray is significantly more efficient, as it succeeds in extracting the target object 82% of the time, 15% more often than the best-performing baseline. Experiments on an ABB YuMi robot with 20 heaps of 25 household objects suggest that the learned policy transfers easily to a physical system, where it outperforms baseline policies by 15% in success rate with 17% fewer actions. Datasets, videos, and experiments are available at http://sites.google.com/berkeley.edu/x-ray .
We propose an architecture for learning complex controllable behaviors by having simple Policies Modulate Trajectory Generators (PMTG), a powerful combination that can provide both memory and prior knowledge to the controller. The result is a flexible architecture that is applicable to a class of problems with periodic motion for which one has an insight into the class of trajectories that might lead to a desired behavior. We illustrate the basics of our architecture using a synthetic control problem, then go on to learn speed-controlled locomotion for a quadrupedal robot by using Deep Reinforcement Learning and Evolutionary Strategies. We demonstrate that a simple linear policy, when paired with a parametric Trajectory Generator for quadrupedal gaits, can induce walking behaviors with controllable speed from 4-dimensional IMU observations alone, and can be learned in under 1000 rollouts. We also transfer these policies to a real robot and show locomotion with controllable forward velocity.
In this paper, we study the problem of learning vision-based dynamic manipulation skills using a scalable reinforcement learning approach. We study this problem in the context of grasping, a longstanding challenge in robotic manipulation. In contrast to static learning behaviors that choose a grasp point and then execute the desired grasp, our method enables closed-loop vision-based control, whereby the robot continuously updates its grasp strategy based on the most recent observations to optimize long-horizon grasp success. To that end, we introduce QT-Opt, a scalable self-supervised vision-based reinforcement learning framework that can leverage over 580k real-world grasp attempts to train a deep neural network Q-function with over 1.2M parameters to perform closed-loop, real-world grasping that generalizes to 96% grasp success on unseen objects. Aside from attaining a very high success rate, our method exhibits behaviors that are quite distinct from more standard grasping systems: using only RGB vision-based perception from an over-the-shoulder camera, our method automatically learns regrasping strategies, probes objects to find the most effective grasps, learns to reposition objects and perform other non-prehensile pre-grasp manipulations, and responds dynamically to disturbances and perturbations.
Well structured visual representations can make robot learning faster and can improve generalization. In this paper, we study how we can acquire effective object-centric representations for robotic manipulation tasks without human labeling by using autonomous robot interaction with the environment. Such representation learning methods can benefit from continuous refinement of the representation as the robot collects more experience, allowing them to scale effectively without human intervention. Our representation learning approach is based on object persistence: when a robot removes an object from a scene, the representation of that scene should change according to the features of the object that was removed. We formulate an arithmetic relationship between feature vectors from this observation, and use it to learn a representation of scenes and objects that can then be used to identify object instances, localize them in the scene, and perform goal-directed grasping tasks where the robot must retrieve commanded objects from a bin. The same grasping procedure can also be used to automatically collect training data for our method, by recording images of scenes, grasping and removing an object, and recording the outcome. Our experiments demonstrate that this self-supervised approach for tasked grasping substantially outperforms direct reinforcement learning from images and prior representation learning methods.
We introduce TensorFlow Agents, an efficient infrastructure paradigm for building parallel reinforcement learning algorithms in TensorFlow. We simulate multiple environments in parallel, and group them to perform the neural network computation on a batch rather than individual observations. This allows the TensorFlow execution engine to parallelize computation, without the need for manual synchronization. Environments are stepped in separate Python processes to progress them in parallel without interference of the global interpreter lock. As part of this project, we introduce BatchPPO, an efficient implementation of the proximal policy optimization algorithm. By open sourcing TensorFlow Agents, we hope to provide a flexible starting point for future projects that accelerates future research in the field.