Learning to Learn Faster from Human Feedback with Language Model Predictive Control

Large language models (LLMs) have been shown to exhibit a wide range of capabilities, such as writing robot code from language commands -- enabling non-experts to direct robot behaviors, modify them based on feedback, or compose them to perform new tasks. However, these capabilities (driven by in-context learning) are limited to short-term interactions, where users' feedback remains relevant for only as long as it fits within the context size of the LLM, and can be forgotten over longer interactions. In this work, we investigate fine-tuning the robot code-writing LLMs, to remember their in-context interactions and improve their teachability i.e., how efficiently they adapt to human inputs (measured by average number of corrections before the user considers the task successful). Our key observation is that when human-robot interactions are formulated as a partially observable Markov decision process (in which human language inputs are observations, and robot code outputs are actions), then training an LLM to complete previous interactions can be viewed as training a transition dynamics model -- that can be combined with classic robotics techniques such as model predictive control (MPC) to discover shorter paths to success. This gives rise to Language Model Predictive Control (LMPC), a framework that fine-tunes PaLM 2 to improve its teachability on 78 tasks across 5 robot embodiments -- improving non-expert teaching success rates of unseen tasks by 26.9% while reducing the average number of human corrections from 2.4 to 1.9. Experiments show that LMPC also produces strong meta-learners, improving the success rate of in-context learning new tasks on unseen robot embodiments and APIs by 31.5%. See videos, code, and demos at: https://robot-teaching.github.io/.

PIVOT: Iterative Visual Prompting Elicits Actionable Knowledge for VLMs

Vision language models (VLMs) have shown impressive capabilities across a variety of tasks, from logical reasoning to visual understanding. This opens the door to richer interaction with the world, for example robotic control. However, VLMs produce only textual outputs, while robotic control and other spatial tasks require outputting continuous coordinates, actions, or trajectories. How can we enable VLMs to handle such settings without fine-tuning on task-specific data? In this paper, we propose a novel visual prompting approach for VLMs that we call Prompting with Iterative Visual Optimization (PIVOT), which casts tasks as iterative visual question answering. In each iteration, the image is annotated with a visual representation of proposals that the VLM can refer to (e.g., candidate robot actions, localizations, or trajectories). The VLM then selects the best ones for the task. These proposals are iteratively refined, allowing the VLM to eventually zero in on the best available answer. We investigate PIVOT on real-world robotic navigation, real-world manipulation from images, instruction following in simulation, and additional spatial inference tasks such as localization. We find, perhaps surprisingly, that our approach enables zero-shot control of robotic systems without any robot training data, navigation in a variety of environments, and other capabilities. Although current performance is far from perfect, our work highlights potentials and limitations of this new regime and shows a promising approach for Internet-Scale VLMs in robotic and spatial reasoning domains. Website: pivot-prompt.github.io and HuggingFace: https://huggingface.co/spaces/pivot-prompt/pivot-prompt-demo.

Chain of Code: Reasoning with a Language Model-Augmented Code Emulator

Code provides a general syntactic structure to build complex programs and perform precise computations when paired with a code interpreter - we hypothesize that language models (LMs) can leverage code-writing to improve Chain of Thought reasoning not only for logic and arithmetic tasks, but also for semantic ones (and in particular, those that are a mix of both). For example, consider prompting an LM to write code that counts the number of times it detects sarcasm in an essay: the LM may struggle to write an implementation for "detect_sarcasm(string)" that can be executed by the interpreter (handling the edge cases would be insurmountable). However, LMs may still produce a valid solution if they not only write code, but also selectively "emulate" the interpreter by generating the expected output of "detect_sarcasm(string)" and other lines of code that cannot be executed. In this work, we propose Chain of Code (CoC), a simple yet surprisingly effective extension that improves LM code-driven reasoning. The key idea is to encourage LMs to format semantic sub-tasks in a program as flexible pseudocode that the interpreter can explicitly catch undefined behaviors and hand off to simulate with an LM (as an "LMulator"). Experiments demonstrate that Chain of Code outperforms Chain of Thought and other baselines across a variety of benchmarks; on BIG-Bench Hard, Chain of Code achieves 84%, a gain of 12% over Chain of Thought. CoC scales well with large and small models alike, and broadens the scope of reasoning questions that LMs can correctly answer by "thinking in code". Project webpage: https://chain-of-code.github.io.

Open X-Embodiment: Robotic Learning Datasets and RT-X Models

Large, high-capacity models trained on diverse datasets have shown remarkable successes on efficiently tackling downstream applications. In domains from NLP to Computer Vision, this has led to a consolidation of pretrained models, with general pretrained backbones serving as a starting point for many applications. Can such a consolidation happen in robotics? Conventionally, robotic learning methods train a separate model for every application, every robot, and even every environment. Can we instead train generalist X-robot policy that can be adapted efficiently to new robots, tasks, and environments? In this paper, we provide datasets in standardized data formats and models to make it possible to explore this possibility in the context of robotic manipulation, alongside experimental results that provide an example of effective X-robot policies. We assemble a dataset from 22 different robots collected through a collaboration between 21 institutions, demonstrating 527 skills (160266 tasks). We show that a high-capacity model trained on this data, which we call RT-X, exhibits positive transfer and improves the capabilities of multiple robots by leveraging experience from other platforms. More details can be found on the project website $\href{https://robotics-transformer-x.github.io}{\text{robotics-transformer-x.github.io}}$.

Code as Policies: Language Model Programs for Embodied Control

Large language models (LLMs) trained on code completion have been shown to be capable of synthesizing simple Python programs from docstrings [1]. We find that these code-writing LLMs can be re-purposed to write robot policy code, given natural language commands. Specifically, policy code can express functions or feedback loops that process perception outputs (e.g.,from object detectors [2], [3]) and parameterize control primitive APIs. When provided as input several example language commands (formatted as comments) followed by corresponding policy code (via few-shot prompting), LLMs can take in new commands and autonomously re-compose API calls to generate new policy code respectively. By chaining classic logic structures and referencing third-party libraries (e.g., NumPy, Shapely) to perform arithmetic, LLMs used in this way can write robot policies that (i) exhibit spatial-geometric reasoning, (ii) generalize to new instructions, and (iii) prescribe precise values (e.g., velocities) to ambiguous descriptions ("faster") depending on context (i.e., behavioral commonsense). This paper presents code as policies: a robot-centric formalization of language model generated programs (LMPs) that can represent reactive policies (e.g., impedance controllers), as well as waypoint-based policies (vision-based pick and place, trajectory-based control), demonstrated across multiple real robot platforms. Central to our approach is prompting hierarchical code-gen (recursively defining undefined functions), which can write more complex code and also improves state-of-the-art to solve 39.8% of problems on the HumanEval [1] benchmark. Code and videos are available at https://code-as-policies.github.io

Inner Monologue: Embodied Reasoning through Planning with Language Models

Recent works have shown how the reasoning capabilities of Large Language Models (LLMs) can be applied to domains beyond natural language processing, such as planning and interaction for robots. These embodied problems require an agent to understand many semantic aspects of the world: the repertoire of skills available, how these skills influence the world, and how changes to the world map back to the language. LLMs planning in embodied environments need to consider not just what skills to do, but also how and when to do them - answers that change over time in response to the agent's own choices. In this work, we investigate to what extent LLMs used in such embodied contexts can reason over sources of feedback provided through natural language, without any additional training. We propose that by leveraging environment feedback, LLMs are able to form an inner monologue that allows them to more richly process and plan in robotic control scenarios. We investigate a variety of sources of feedback, such as success detection, scene description, and human interaction. We find that closed-loop language feedback significantly improves high-level instruction completion on three domains, including simulated and real table top rearrangement tasks and long-horizon mobile manipulation tasks in a kitchen environment in the real world.

Learning Preconditions of Hybrid Force-Velocity Controllers for Contact-Rich Manipulation

Robots need to manipulate objects in constrained environments like shelves and cabinets when assisting humans in everyday settings like homes and offices. These constraints make manipulation difficult by reducing grasp accessibility, so robots need to use non-prehensile strategies that leverage object-environment contacts to perform manipulation tasks. To tackle the challenge of planning and controlling contact-rich behaviors in such settings, this work uses Hybrid Force-Velocity Controllers (HFVCs) as the skill representation and plans skill sequences with learned preconditions. While HFVCs naturally enable robust and compliant contact-rich behaviors, solvers that synthesize them have traditionally relied on precise object models and closed-loop feedback on object pose, which are difficult to obtain in constrained environments due to occlusions. We first relax HFVCs' need for precise models and feedback with our HFVC synthesis framework, then learn a point-cloud-based precondition function to classify where HFVC executions will still be successful despite modeling inaccuracies. Finally, we use the learned precondition in a search-based task planner to complete contact-rich manipulation tasks in a shelf domain. Our method achieves a task success rate of $73.2\%$, outperforming the $51.5\%$ achieved by a baseline without the learned precondition. While the precondition function is trained in simulation, it can also transfer to a real-world setup without further fine-tuning.

Search-Based Task Planning with Learned Skill Effect Models for Lifelong Robotic Manipulation

Lifelong-learning robots need to be able to acquire new skills and plan for new tasks over time. Prior works on planning with skills often make assumptions on the structure of skills and tasks, like subgoal skills, shared skill implementations, or learning task-specific plan skeletons, that limit their application to new and different skills and tasks. By contrast, we propose doing task planning by jointly searching in the space of skills and their parameters with skill effect models learned in simulation. Our approach is flexible about skill parameterizations and task specifications, and we use an iterative training procedure to efficiently generate relevant data to train such models. Experiments demonstrate the ability of our planner to integrate new skills in a lifelong manner, finding new task strategies with lower costs in both train and test tasks. We additionally show that our method can transfer to the real world without further fine-tuning.

Visual Identification of Articulated Object Parts

As autonomous robots interact and navigate around real-world environments such as homes, it is useful to reliably identify and manipulate articulated objects, such as doors and cabinets. Many prior works in object articulation identification require manipulation of the object, either by the robot or a human. While recent works have addressed predicting articulation types from visual observations alone, they often assume prior knowledge of category-level kinematic motion models or sequence of observations where the articulated parts are moving according to their kinematic constraints. In this work, we propose training a neural network through large-scale domain randomization to identify the articulation type of object parts from a single image observation. Training data is generated via photorealistic rendering in simulation. Our proposed model predicts motion residual flows of object parts, and these residuals are used to determine the articulation type and parameters. We train the network on six object categories with 149 objects and 100K rendered images, achieving an accuracy of 82.5%. Experiments show our method generalizes to novel object categories in simulation and can be applied to real-world images without fine-tuning.

Learning to Compose Hierarchical Object-Centric Controllers for Robotic Manipulation

Manipulation tasks can often be decomposed into multiple subtasks performed in parallel, e.g., sliding an object to a goal pose while maintaining contact with a table. Individual subtasks can be achieved by task-axis controllers defined relative to the objects being manipulated, and a set of object-centric controllers can be combined in an hierarchy. In prior works, such combinations are defined manually or learned from demonstrations. By contrast, we propose using reinforcement learning to dynamically compose hierarchical object-centric controllers for manipulation tasks. Experiments in both simulation and real world show how the proposed approach leads to improved sample efficiency, zero-shot generalization to novel test environments, and simulation-to-reality transfer without fine-tuning.