We introduce Action-GPT, a plug-and-play framework for incorporating Large Language Models (LLMs) into text-based action generation models. Action phrases in current motion capture datasets contain minimal and to-the-point information. By carefully crafting prompts for LLMs, we generate richer and fine-grained descriptions of the action. We show that utilizing these detailed descriptions instead of the original action phrases leads to better alignment of text and motion spaces. We introduce a generic approach compatible with stochastic (e.g. VAE-based) and deterministic (e.g. MotionCLIP) text-to-motion models. In addition, the approach enables multiple text descriptions to be utilized. Our experiments show (i) noticeable qualitative and quantitative improvement in the quality of synthesized motions, (ii) benefits of utilizing multiple LLM-generated descriptions, (iii) suitability of the prompt function, and (iv) zero-shot generation capabilities of the proposed approach. Project page: https://actiongpt.github.io
Text editing, such as grammatical error correction, arises naturally from imperfect textual data. Recent works frame text editing as a multi-round sequence tagging task, where operations -- such as insertion and substitution -- are represented as a sequence of tags. While achieving good results, this encoding is limited in flexibility as all actions are bound to token-level tags. In this work, we reformulate text editing as an imitation game using behavioral cloning. Specifically, we convert conventional sequence-to-sequence data into state-to-action demonstrations, where the action space can be as flexible as needed. Instead of generating the actions one at a time, we introduce a dual decoders structure to parallel the decoding while retaining the dependencies between action tokens, coupled with trajectory augmentation to alleviate the distribution shift that imitation learning often suffers. In experiments on a suite of Arithmetic Equation benchmarks, our model consistently outperforms the autoregressive baselines in terms of performance, efficiency, and robustness. We hope our findings will shed light on future studies in reinforcement learning applying sequence-level action generation to natural language processing.
Motivations, emotions, and actions are inter-related essential factors in human activities. While motivations and emotions have long been considered at the core of exploring how people take actions in human activities, there has been relatively little research supporting analyzing the relationship between human mental states and actions. We present the first study that investigates the viability of modeling motivations, emotions, and actions in language-based human activities, named COMMA (Cognitive Framework of Human Activities). Guided by COMMA, we define three natural language processing tasks (emotion understanding, motivation understanding and conditioned action generation), and build a challenging dataset Hail through automatically extracting samples from Story Commonsense. Experimental results on NLP applications prove the effectiveness of modeling the relationship. Furthermore, our models inspired by COMMA can better reveal the essential relationship among motivations, emotions and actions than existing methods.
Building a conversational embodied agent to execute real-life tasks has been a long-standing yet quite challenging research goal, as it requires effective human-agent communication, multi-modal understanding, long-range sequential decision making, etc. Traditional symbolic methods have scaling and generalization issues, while end-to-end deep learning models suffer from data scarcity and high task complexity, and are often hard to explain. To benefit from both worlds, we propose JARVIS, a neuro-symbolic commonsense reasoning framework for modular, generalizable, and interpretable conversational embodied agents. First, it acquires symbolic representations by prompting large language models (LLMs) for language understanding and sub-goal planning, and by constructing semantic maps from visual observations. Then the symbolic module reasons for sub-goal planning and action generation based on task- and action-level common sense. Extensive experiments on the TEACh dataset validate the efficacy and efficiency of our JARVIS framework, which achieves state-of-the-art (SOTA) results on all three dialog-based embodied tasks, including Execution from Dialog History (EDH), Trajectory from Dialog (TfD), and Two-Agent Task Completion (TATC) (e.g., our method boosts the unseen Success Rate on EDH from 6.1\% to 15.8\%). Moreover, we systematically analyze the essential factors that affect the task performance and also demonstrate the superiority of our method in few-shot settings. Our JARVIS model ranks first in the Alexa Prize SimBot Public Benchmark Challenge.
Image-goal navigation is a challenging task, as it requires the agent to navigate to a target indicated by an image in a previously unseen scene. Current methods introduce diverse memory mechanisms which save navigation history to solve this task. However, these methods use all observations in the memory for generating navigation actions without considering which fraction of this memory is informative. To address this limitation, we present the MemoNav, a novel memory mechanism for image-goal navigation, which retains the agent's informative short-term memory and long-term memory to improve the navigation performance on a multi-goal task. The node features on the agent's topological map are stored in the short-term memory, as these features are dynamically updated. To aid the short-term memory, we also generate long-term memory by continuously aggregating the short-term memory via a graph attention module. The MemoNav retains the informative fraction of the short-term memory via a forgetting module based on a Transformer decoder and then incorporates this retained short-term memory and the long-term memory into working memory. Lastly, the agent uses the working memory for action generation. We evaluate our model on a new multi-goal navigation dataset. The experimental results show that the MemoNav outperforms the SoTA methods by a large margin with a smaller fraction of navigation history. The results also empirically show that our model is less likely to be trapped in a deadlock, which further validates that the MemoNav improves the agent's navigation efficiency by reducing redundant steps.
Intention, emotion and action are important psychological factors in human activities, which play an important role in the interaction between individuals. How to model the interaction process between individuals by analyzing the relationship of their intentions, emotions, and actions at the cognitive level is challenging. In this paper, we propose a novel cognitive framework of individual interaction. The core of the framework is that individuals achieve interaction through external action driven by their inner intention. Based on this idea, the interactions between individuals can be constructed by establishing relationships between the intention, emotion and action. Furthermore, we conduct analysis on the interaction between individuals and give a reasonable explanation for the predicting results. To verify the effectiveness of the framework, we reconstruct a dataset and propose three tasks as well as the corresponding baseline models, including action abduction, emotion prediction and action generation. The novel framework shows an interesting perspective on mimicking the mental state of human beings in cognitive science.
Abstract— Proactive taxi dispatching is of great importance to balance taxi demand-supply gaps among different locations in a city. Recent advances primarily rely on deep reinforcement learning (DRL) to directly learn the optimal dispatching policy. These works, however, are still not sufficiently efficient because they overlook several pieces of valuable context information. As a result, they may generate quite a few improper actions and introduce unnecessary coordination costs. To improve existing works, we present COX – a context-aware taxi dispatching approach that incorporates rich contexts into DRL modeling for more efficient taxi reallocations. Specifically, rather than simply dividing the service area into grids, COX proposes a road connectivity aware clustering algorithm to divide the road network graph into zones for practical taxi dispatching. In addition, COX comprehensively analyzes zone-level taxi demands and supplies through accurate taxi demand prediction and timely updates of taxi statuses. COX improves the DRL modeling by integrating these derived contexts, e.g., state representation with complete demand/supply data and sequential action generation with full coordination among idle taxis. In particular, we implement an environment simulator to train and evaluate COX using a large real-world taxi dataset. Extensive experiments show that COX outperforms state-of-the-art approaches on various performance metrics, e.g., on average improving the total order values by 6.74%, while reducing the number of unserved taxi orders and passengers’ waiting time by 4.92% and 44.84%, respectively.
We propose novel methods to develop action controllable agent that behaves like a human and has the ability to align with human players in Multiplayer Online Battle Arena (MOBA) games. By modeling the control problem as an action generation process, we devise a deep latent alignment neural network model for training agent, and a corresponding sampling algorithm for controlling an agent's action. Particularly, we propose deterministic and stochastic attention implementations of the core latent alignment model. Both simulated and online experiments in the game Honor of Kings demonstrate the efficacy of the proposed methods.
The idea that the brain is a probabilistic (Bayesian) inference machine, continuously trying to figure out the hidden causes of its inputs, has become very influential in cognitive (neuro)science over recent decades. Here I present a relatively straightforward generalization of this idea: the primary computational task that the brain is faced with is to track the probabilistic structure of observations themselves, without recourse to hidden states. Taking this starting point seriously turns out to have considerable explanatory power, and several key ideas are developed from it: (1) past experience, encoded in prior expectations, has an influence over the future that is analogous to regularization as known from machine learning; (2) action generation (interpreted as constraint satisfaction) is a special case of such regularization; (3) the concept of attractors in dynamical systems provides a useful lens through which prior expectations, regularization, and action induction can be viewed; these thus appear as different perspectives on the same phenomenon; (4) the phylogenetically ancient imperative of acting to ensure and thereby observe conditions beneficial for survival is likely the same as that which underlies perceptual inference. The Bayesian brain hypothesis has been touted as promising to deliver a "unified science of mind and action". In this paper, I sketch an informal step towards fulfilling that promise, while avoiding some pitfalls that other such attempts have fallen prey to.
We introduce MUGL, a novel deep neural model for large-scale, diverse generation of single and multi-person pose-based action sequences with locomotion. Our controllable approach enables variable-length generations customizable by action category, across more than 100 categories. To enable intra/inter-category diversity, we model the latent generative space using a Conditional Gaussian Mixture Variational Autoencoder. To enable realistic generation of actions involving locomotion, we decouple local pose and global trajectory components of the action sequence. We incorporate duration-aware feature representations to enable variable-length sequence generation. We use a hybrid pose sequence representation with 3D pose sequences sourced from videos and 3D Kinect-based sequences of NTU-RGBD-120. To enable principled comparison of generation quality, we employ suitably modified strong baselines during evaluation. Although smaller and simpler compared to baselines, MUGL provides better quality generations, paving the way for practical and controllable large-scale human action generation.
Synthesising the spatial and temporal dynamics of the human body skeleton remains a challenging task, not only in terms of the quality of the generated shapes, but also of their diversity, particularly to synthesise realistic body movements of a specific action (action conditioning). In this paper, we propose Kinetic-GAN, a novel architecture that leverages the benefits of Generative Adversarial Networks and Graph Convolutional Networks to synthesise the kinetics of the human body. The proposed adversarial architecture can condition up to 120 different actions over local and global body movements while improving sample quality and diversity through latent space disentanglement and stochastic variations. Our experiments were carried out in three well-known datasets, where Kinetic-GAN notably surpasses the state-of-the-art methods in terms of distribution quality metrics while having the ability to synthesise more than one order of magnitude regarding the number of different actions. Our code and models are publicly available at https://github.com/DegardinBruno/Kinetic-GAN.
To accurately pour drinks into various containers is an essential skill for service robots. However, drink pouring is a dynamic process and difficult to model. Traditional deep imitation learning techniques for implementing autonomous robotic pouring have an inherent black-box effect and require a large amount of demonstration data for model training. To address these issues, an Explainable Hierarchical Imitation Learning (EHIL) method is proposed in this paper such that a robot can learn high-level general knowledge and execute low-level actions across multiple drink pouring scenarios. Moreover, with EHIL, a logical graph can be constructed for task execution, through which the decision-making process for action generation can be made explainable to users and the causes of failure can be traced out. Based on the logical graph, the framework is manipulable to achieve different targets while the adaptability to unseen scenarios can be achieved in an explainable manner. A series of experiments have been conducted to verify the effectiveness of the proposed method. Results indicate that EHIL outperforms the traditional behavior cloning method in terms of success rate, adaptability, manipulability and explainability.
Imitation learning is a data-driven approach to acquiring skills that relies on expert demonstrations to learn a policy that maps observations to actions. When performing demonstrations, experts are not always consistent and might accomplish the same task in slightly different ways. In this paper, we demonstrate inherent stochasticity in demonstrations collected for tasks including line following with a remote-controlled car and manipulation tasks including reaching, pushing, and picking and placing an object. We model stochasticity in the data distribution using autoregressive action generation, generative adversarial nets, and variational prediction and compare the performance of these approaches. We find that accounting for stochasticity in the expert data leads to substantial improvement in the success rate of task completion.
Thompson sampling (TS) has emerged as a robust technique for contextual bandit problems. However, TS requires posterior inference and optimization for action generation, prohibiting its use in many online platforms where latency and ease of deployment are of concern. We operationalize TS by proposing a novel imitation-learning-based algorithm that distills a TS policy into an explicit policy representation, allowing fast decision-making and easy deployment in mobile and server-based environments. Using batched data collected under the imitation policy, our algorithm iteratively performs offline updates to the TS policy, and learns a new explicit policy representation to imitate it. Empirically, our imitation policy achieves performance comparable to batch TS while allowing more than an order of magnitude reduction in decision-time latency. Buoyed by low latency and simplicity of implementation, our algorithm has been successfully deployed in multiple video upload systems for Meta. Using a randomized controlled trial, we show our algorithm resulted in significant improvements in video quality and watch time.
Text-based games present a unique challenge for autonomous agents to operate in natural language and handle enormous action spaces. In this paper, we propose the Contextual Action Language Model (CALM) to generate a compact set of action candidates at each game state. Our key insight is to train language models on human gameplay, where people demonstrate linguistic priors and a general game sense for promising actions conditioned on game history. We combine CALM with a reinforcement learning agent which re-ranks the generated action candidates to maximize in-game rewards. We evaluate our approach using the Jericho benchmark, on games unseen by CALM during training. Our method obtains a 69% relative improvement in average game score over the previous state-of-the-art model. Surprisingly, on half of these games, CALM is competitive with or better than other models that have access to ground truth admissible actions. Code and data are available at https://github.com/princeton-nlp/calm-textgame.
Action recognition is a relatively established task, where givenan input sequence of human motion, the goal is to predict its ac-tion category. This paper, on the other hand, considers a relativelynew problem, which could be thought of as an inverse of actionrecognition: given a prescribed action type, we aim to generateplausible human motion sequences in 3D. Importantly, the set ofgenerated motions are expected to maintain itsdiversityto be ableto explore the entire action-conditioned motion space; meanwhile,each sampled sequence faithfully resembles anaturalhuman bodyarticulation dynamics. Motivated by these objectives, we followthe physics law of human kinematics by adopting the Lie Algebratheory to represent thenaturalhuman motions; we also propose atemporal Variational Auto-Encoder (VAE) that encourages adiversesampling of the motion space. A new 3D human motion dataset, HumanAct12, is also constructed. Empirical experiments overthree distinct human motion datasets (including ours) demonstratethe effectiveness of our approach.
Generating long-range skeleton-based human actions has been a challenging problem since small deviations of one frame can cause a malformed action sequence. Most existing methods borrow ideas from video generation, which naively treat skeleton nodes/joints as pixels of images without considering the rich inter-frame and intra-frame structure information, leading to potential distorted actions. Graph convolutional networks (GCNs) is a promising way to leverage structure information to learn structure representations. However, directly adopting GCNs to tackle such continuous action sequences both in spatial and temporal spaces is challenging as the action graph could be huge. To overcome this issue, we propose a variant of GCNs to leverage the powerful self-attention mechanism to adaptively sparsify a complete action graph in the temporal space. Our method could dynamically attend to important past frames and construct a sparse graph to apply in the GCN framework, well-capturing the structure information in action sequences. Extensive experimental results demonstrate the superiority of our method on two standard human action datasets compared with existing methods.
Sample-based planning is a powerful family of algorithms for generating intelligent behavior from a model of the environment. Generating good candidate actions is critical to the success of sample-based planners, particularly in continuous or large action spaces. Typically, candidate action generation exhausts the action space, uses domain knowledge, or more recently, involves learning a stochastic policy to provide such search guidance. In this paper we explore explicitly learning a candidate action generator by optimizing a novel objective, marginal utility. The marginal utility of an action generator measures the increase in value of an action over previously generated actions. We validate our approach in both curling, a challenging stochastic domain with continuous state and action spaces, and a location game with a discrete but large action space. We show that a generator trained with the marginal utility objective outperforms hand-coded schemes built on substantial domain knowledge, trained stochastic policies, and other natural objectives for generating actions for sampled-based planners.
Japanese character figurines are popular and have pivot position in Otaku culture. Although numerous robots have been developed, less have focused on otaku-culture or on embodying the anime character figurine. Therefore, we take the first steps to bridge this gap by developing Hatsuki, which is a humanoid robot platform with anime based design. Hatsuki's novelty lies in aesthetic design, 2D facial expressions, and anime-style behaviors that allows it to deliver rich interaction experiences resembling anime-characters. We explain our design implementation process of Hatsuki, followed by our evaluations. In order to explore user impressions and opinions towards Hatsuki, we conducted a questionnaire in the world's largest anime-figurine event. The results indicate that participants were generally very satisfied with Hatsuki's design, and proposed various use case scenarios and deployment contexts for Hatsuki. The second evaluation focused on imitation learning, as such method can provide better interaction ability in the real world and generate rich, context-adaptive behavior in different situations. We made Hatsuki learn 11 actions, combining voice, facial expressions and motions, through neuron network based policy model with our proposed interface. Results show our approach was successfully able to generate the actions through self-organized contexts, which shows the potential for generalizing our approach in further actions under different contexts. Lastly, we present our future research direction for Hatsuki, and provide our conclusion.
Interactive Fiction games are text-based simulations in which an agent interacts with the world purely through natural language. They are ideal environments for studying how to extend reinforcement learning agents to meet the challenges of natural language understanding, partial observability, and action generation in combinatorially-large text-based action spaces. We present KG-A2C, an agent that builds a dynamic knowledge graph while exploring and generates actions using a template-based action space. We contend that the dual uses of the knowledge graph to reason about game state and to constrain natural language generation are the keys to scalable exploration of combinatorially large natural language actions. Results across a wide variety of IF games show that KG-A2C outperforms current IF agents despite the exponential increase in action space size.