The autoregressive world model exhibits robust generalization capabilities in vectorized scene understanding but encounters difficulties in deriving actions due to insufficient uncertainty modeling and self-delusion. In this paper, we explore the feasibility of deriving decisions from an autoregressive world model by addressing these challenges through the formulation of multiple probabilistic hypotheses. We propose LatentDriver, a framework models the environment's next states and the ego vehicle's possible actions as a mixture distribution, from which a deterministic control signal is then derived. By incorporating mixture modeling, the stochastic nature of decisionmaking is captured. Additionally, the self-delusion problem is mitigated by providing intermediate actions sampled from a distribution to the world model. Experimental results on the recently released close-loop benchmark Waymax demonstrate that LatentDriver surpasses state-of-the-art reinforcement learning and imitation learning methods, achieving expert-level performance. The code and models will be made available at https://github.com/Sephirex-X/LatentDriver.
Robot models, particularly those trained with large amounts of data, have recently shown a plethora of real-world manipulation and navigation capabilities. Several independent efforts have shown that given sufficient training data in an environment, robot policies can generalize to demonstrated variations in that environment. However, needing to finetune robot models to every new environment stands in stark contrast to models in language or vision that can be deployed zero-shot for open-world problems. In this work, we present Robot Utility Models (RUMs), a framework for training and deploying zero-shot robot policies that can directly generalize to new environments without any finetuning. To create RUMs efficiently, we develop new tools to quickly collect data for mobile manipulation tasks, integrate such data into a policy with multi-modal imitation learning, and deploy policies on-device on Hello Robot Stretch, a cheap commodity robot, with an external mLLM verifier for retrying. We train five such utility models for opening cabinet doors, opening drawers, picking up napkins, picking up paper bags, and reorienting fallen objects. Our system, on average, achieves 90% success rate in unseen, novel environments interacting with unseen objects. Moreover, the utility models can also succeed in different robot and camera set-ups with no further data, training, or fine-tuning. Primary among our lessons are the importance of training data over training algorithm and policy class, guidance about data scaling, necessity for diverse yet high-quality demonstrations, and a recipe for robot introspection and retrying to improve performance on individual environments. Our code, data, models, hardware designs, as well as our experiment and deployment videos are open sourced and can be found on our project website: https://robotutilitymodels.com
The problem of generalization in learning from demonstration (LfD) has received considerable attention over the years, particularly within the context of movement primitives, where a number of approaches have emerged. Recently, two important approaches have gained recognition. While one leverages via-points to adapt skills locally by modulating demonstrated trajectories, another relies on so-called task-parameterized models that encode movements with respect to different coordinate systems, using a product of probabilities for generalization. While the former are well-suited to precise, local modulations, the latter aim at generalizing over large regions of the workspace and often involve multiple objects. Addressing the quality of generalization by leveraging both approaches simultaneously has received little attention. In this work, we propose an interactive imitation learning framework that simultaneously leverages local and global modulations of trajectory distributions. Building on the kernelized movement primitives (KMP) framework, we introduce novel mechanisms for skill modulation from direct human corrective feedback. Our approach particularly exploits the concept of via-points to incrementally and interactively 1) improve the model accuracy locally, 2) add new objects to the task during execution and 3) extend the skill into regions where demonstrations were not provided. We evaluate our method on a bearing ring-loading task using a torque-controlled, 7-DoF, DLR SARA robot.
Multi-agent pathfinding (MAPF) is a problem that generally requires finding collision-free paths for multiple agents in a shared environment. Solving MAPF optimally, even under restrictive assumptions, is NP-hard, yet efficient solutions for this problem are critical for numerous applications, such as automated warehouses and transportation systems. Recently, learning-based approaches to MAPF have gained attention, particularly those leveraging deep reinforcement learning. Typically, such learning-based MAPF solvers are augmented with additional components like single-agent planning or communication. Orthogonally, in this work we rely solely on imitation learning that leverages a large dataset of expert MAPF solutions and transformer-based neural network to create a foundation model for MAPF called MAPF-GPT. The latter is capable of generating actions without additional heuristics or communication. MAPF-GPT demonstrates zero-shot learning abilities when solving the MAPF problems that are not present in the training dataset. We show that MAPF-GPT notably outperforms the current best-performing learnable MAPF solvers on a diverse range of problem instances and is computationally efficient during inference.
We explore how to enhance next-token prediction models to perform in-context imitation learning on a real robot, where the robot executes new tasks by interpreting contextual information provided during the input phase, without updating its underlying policy parameters. We propose In-Context Robot Transformer (ICRT), a causal transformer that performs autoregressive prediction on sensorimotor trajectories without relying on any linguistic data or reward function. This formulation enables flexible and training-free execution of new tasks at test time, achieved by prompting the model with sensorimotor trajectories of the new task composing of image observations, actions and states tuples, collected through human teleoperation. Experiments with a Franka Emika robot demonstrate that the ICRT can adapt to new tasks specified by prompts, even in environment configurations that differ from both the prompt and the training data. In a multitask environment setup, ICRT significantly outperforms current state-of-the-art next-token prediction models in robotics on generalizing to unseen tasks. Code, checkpoints and data are available on https://icrt.dev/
Increasingly large imitation learning datasets are being collected with the goal of training foundation models for robotics. However, despite the fact that data selection has been of utmost importance in vision and natural language processing, little work in robotics has questioned what data such models should actually be trained on. In this work we investigate how to weigh different subsets or ``domains'' of robotics datasets for robot foundation model pre-training. Concrete, we use distributionally robust optimization (DRO) to maximize worst-case performance across all possible downstream domains. Our method, Re-Mix, addresses the wide range of challenges that arise when applying DRO to robotics datasets including variability in action spaces and dynamics across different datasets. Re-Mix employs early stopping, action normalization, and discretization to counteract these issues. Through extensive experimentation on the largest open-source robot manipulation dataset, the Open X-Embodiment dataset, we demonstrate that data curation can have an outsized impact on downstream performance. Specifically, domain weights learned by Re-Mix outperform uniform weights by 38\% on average and outperform human-selected weights by 32\% on datasets used to train existing generalist robot policies, specifically the RT-X models.
Reinforcement learning techniques leveraging deep learning have made tremendous progress in recent years. However, the complexity of neural networks prevents practitioners from understanding their behavior. Decision trees have gained increased attention in supervised learning for their inherent interpretability, enabling modelers to understand the exact prediction process after learning. This paper considers the problem of optimizing interpretable decision tree policies to replace neural networks in reinforcement learning settings. Previous works have relaxed the tree structure, restricted to optimizing only tree leaves, or applied imitation learning techniques to approximately copy the behavior of a neural network policy with a decision tree. We propose the Decision Tree Policy Optimization (DTPO) algorithm that directly optimizes the complete decision tree using policy gradients. Our technique uses established decision tree heuristics for regression to perform policy optimization. We empirically show that DTPO is a competitive algorithm compared to imitation learning algorithms for optimizing decision tree policies in reinforcement learning.
Embodied learning for object-centric robotic manipulation is a rapidly developing and challenging area in embodied AI. It is crucial for advancing next-generation intelligent robots and has garnered significant interest recently. Unlike data-driven machine learning methods, embodied learning focuses on robot learning through physical interaction with the environment and perceptual feedback, making it especially suitable for robotic manipulation. In this paper, we provide a comprehensive survey of the latest advancements in this field and categorize the existing work into three main branches: 1) Embodied perceptual learning, which aims to predict object pose and affordance through various data representations; 2) Embodied policy learning, which focuses on generating optimal robotic decisions using methods such as reinforcement learning and imitation learning; 3) Embodied task-oriented learning, designed to optimize the robot's performance based on the characteristics of different tasks in object grasping and manipulation. In addition, we offer an overview and discussion of public datasets, evaluation metrics, representative applications, current challenges, and potential future research directions. A project associated with this survey has been established at https://github.com/RayYoh/OCRM_survey.
Balancing game difficulty in video games is a key task to create interesting gaming experiences for players. Mismatching the game difficulty and a player's skill or commitment results in frustration or boredom on the player's side, and hence reduces time spent playing the game. In this work, we explore balancing game difficulty using machine learning-based agents to challenge players based on their current behavior. This is achieved by a combination of two agents, in which one learns to imitate the player, while the second is trained to beat the first. In our demo, we investigate the proposed framework for personalized dynamic difficulty adjustment of AI agents in the context of the fighting game AI competition.
This paper addresses navigation in crowded environments by integrating goal-conditioned generative models with Sampling-based Model Predictive Control (SMPC). We introduce goal-conditioned autoregressive models to generate crowd behaviors, capturing intricate interactions among individuals. The model processes potential robot trajectory samples and predicts the reactions of surrounding individuals, enabling proactive robotic navigation in complex scenarios. Extensive experiments show that this algorithm enables real-time navigation, significantly reducing collision rates and path lengths, and outperforming selected baseline methods. The practical effectiveness of this algorithm is validated on an actual robotic platform, demonstrating its capability in dynamic settings.
Autonomous parking is a crucial task in the intelligent driving field. Traditional parking algorithms are usually implemented using rule-based schemes. However, these methods are less effective in complex parking scenarios due to the intricate design of the algorithms. In contrast, neural-network-based methods tend to be more intuitive and versatile than the rule-based methods. By collecting a large number of expert parking trajectory data and emulating human strategy via learning-based methods, the parking task can be effectively addressed. In this paper, we employ imitation learning to perform end-to-end planning from RGB images to path planning by imitating human driving trajectories. The proposed end-to-end approach utilizes a target query encoder to fuse images and target features, and a transformer-based decoder to autoregressively predict future waypoints. We conducted extensive experiments in real-world scenarios, and the results demonstrate that the proposed method achieved an average parking success rate of 87.8% across four different real-world garages. Real-vehicle experiments further validate the feasibility and effectiveness of the method proposed in this paper.
Power grid operation is becoming increasingly complex due to the increase in generation of renewable energy. The recent series of Learning To Run a Power Network (L2RPN) competitions have encouraged the use of artificial agents to assist human dispatchers in operating power grids. In this paper we study the performance of imitation learning for day-ahead power grid operation through topology actions. In particular, we consider two rule-based expert agents: a greedy agent and a N-1 agent. While the latter is more computationally expensive since it takes N-1 safety considerations into account, it exhibits a much higher operational performance. We train a fully-connected neural network (FCNN) on expert state-action pairs and evaluate it in two ways. First, we find that classification accuracy is limited despite extensive hyperparameter tuning, due to class imbalance and class overlap. Second, as a power system agent, the FCNN performs only slightly worse than expert agents. Furthermore, hybrid agents, which incorporate minimal additional simulations, match expert agents' performance with significantly lower computational cost. Consequently, imitation learning shows promise for developing fast, high-performing power grid agents, motivating its further exploration in future L2RPN studies.
Personalized motion planning holds significant importance within urban automated driving, catering to the unique requirements of individual users. Nevertheless, prior endeavors have frequently encountered difficulties in simultaneously addressing two crucial aspects: personalized planning within intricate urban settings and enhancing planning performance through data utilization. The challenge arises from the expensive and limited nature of user data, coupled with the scene state space tending towards infinity. These factors contribute to overfitting and poor generalization problems during model training. Henceforth, we propose an instance-based transfer imitation learning approach. This method facilitates knowledge transfer from extensive expert domain data to the user domain, presenting a fundamental resolution to these issues. We initially train a pre-trained model using large-scale expert data. Subsequently, during the fine-tuning phase, we feed the batch data, which comprises expert and user data. Employing the inverse reinforcement learning technique, we extract the style feature distribution from user demonstrations, constructing the regularization term for the approximation of user style. In our experiments, we conducted extensive evaluations of the proposed method. Compared to the baseline methods, our approach mitigates the overfitting issue caused by sparse user data. Furthermore, we discovered that integrating the driving model with a differentiable nonlinear optimizer as a safety protection layer for end-to-end personalized fine-tuning results in superior planning performance.
Generating human mobility trajectories is of great importance to solve the lack of large-scale trajectory data in numerous applications, which is caused by privacy concerns. However, existing mobility trajectory generation methods still require real-world human trajectories centrally collected as the training data, where there exists an inescapable risk of privacy leakage. To overcome this limitation, in this paper, we propose PateGail, a privacy-preserving imitation learning model to generate mobility trajectories, which utilizes the powerful generative adversary imitation learning model to simulate the decision-making process of humans. Further, in order to protect user privacy, we train this model collectively based on decentralized mobility data stored in user devices, where personal discriminators are trained locally to distinguish and reward the real and generated human trajectories. In the training process, only the generated trajectories and their rewards obtained based on personal discriminators are shared between the server and devices, whose privacy is further preserved by our proposed perturbation mechanisms with theoretical proof to satisfy differential privacy. Further, to better model the human decision-making process, we propose a novel aggregation mechanism of the rewards obtained from personal discriminators. We theoretically prove that under the reward obtained based on the aggregation mechanism, our proposed model maximizes the lower bound of the discounted total rewards of users. Extensive experiments show that the trajectories generated by our model are able to resemble real-world trajectories in terms of five key statistical metrics, outperforming state-of-the-art algorithms by over 48.03%. Furthermore, we demonstrate that the synthetic trajectories are able to efficiently support practical applications, including mobility prediction and location recommendation.
We propose C-LAIfO, a computationally efficient algorithm designed for imitation learning from videos in the presence of visual mismatch between agent and expert domains. We analyze the problem of imitation from expert videos with visual discrepancies, and introduce a solution for robust latent space estimation using contrastive learning and data augmentation. Provided a visually robust latent space, our algorithm performs imitation entirely within this space using off-policy adversarial imitation learning. We conduct a thorough ablation study to justify our design and test C-LAIfO on high-dimensional continuous robotic tasks. Additionally, we demonstrate how C-LAIfO can be combined with other reward signals to facilitate learning on a set of challenging hand manipulation tasks with sparse rewards. Our experiments show improved performance compared to baseline methods, highlighting the effectiveness of C-LAIfO. To ensure reproducibility, we open source our code.
The flexibility of electrical heating devices can help address the issues arising from the growing presence of unpredictable renewable energy sources in the energy system. In particular, heat pumps offer an effective solution by employing smart control methods that adjust the heat pump's power output in reaction to demand response signals. This paper combines imitation learning based on an artificial neural network with an intelligent control approach for heat pumps. We train the model using the output data of an optimization problem to determine the optimal operation schedule of a heat pump. The objective is to minimize the electricity cost with a time-variable electricity tariff while keeping the building temperature within acceptable boundaries. We evaluate our developed novel method, PSC-ANN, on various multi-family buildings with differing insulation levels that utilize an underfloor heating system as thermal storage. The results show that PSC-ANN outperforms a positively evaluated intelligent control approach from the literature and a conventional control approach. Further, our experiments reveal that a trained imitation learning model for a specific building is also applicable to other similar buildings without the need to train it again with new data. Our developed approach also reduces the execution time compared to optimally solving the corresponding optimization problem. PSC-ANN can be integrated into multiple buildings, enabling them to better utilize renewable energy sources by adjusting their electricity consumption in response to volatile external signals.
Task Parametrized Gaussian Mixture Models (TP-GMM) are a sample-efficient method for learning object-centric robot manipulation tasks. However, there are several open challenges to applying TP-GMMs in the wild. In this work, we tackle three crucial challenges synergistically. First, end-effector velocities are non-Euclidean and thus hard to model using standard GMMs. We thus propose to factorize the robot's end-effector velocity into its direction and magnitude, and model them using Riemannian GMMs. Second, we leverage the factorized velocities to segment and sequence skills from complex demonstration trajectories. Through the segmentation, we further align skill trajectories and hence leverage time as a powerful inductive bias. Third, we present a method to automatically detect relevant task parameters per skill from visual observations. Our approach enables learning complex manipulation tasks from just five demonstrations while using only RGB-D observations. Extensive experimental evaluations on RLBench demonstrate that our approach achieves state-of-the-art performance with 20-fold improved sample efficiency. Our policies generalize across different environments, object instances, and object positions, while the learned skills are reusable.
Learning a locomotion controller for a musculoskeletal system is challenging due to over-actuation and high-dimensional action space. While many reinforcement learning methods attempt to address this issue, they often struggle to learn human-like gaits because of the complexity involved in engineering an effective reward function. In this paper, we demonstrate that adversarial imitation learning can address this issue by analyzing key problems and providing solutions using both current literature and novel techniques. We validate our methodology by learning walking and running gaits on a simulated humanoid model with 16 degrees of freedom and 92 Muscle-Tendon Units, achieving natural-looking gaits with only a few demonstrations.
Generalising vision-based manipulation policies to novel environments remains a challenging area with limited exploration. Current practices involve collecting data in one location, training imitation learning or reinforcement learning policies with this data, and deploying the policy in the same location. However, this approach lacks scalability as it necessitates data collection in multiple locations for each task. This paper proposes a novel approach where data is collected in a location predominantly featuring green screens. We introduce Green-screen Augmentation (GreenAug), employing a chroma key algorithm to overlay background textures onto a green screen. Through extensive real-world empirical studies with over 850 training demonstrations and 8.2k evaluation episodes, we demonstrate that GreenAug surpasses no augmentation, standard computer vision augmentation, and prior generative augmentation methods in performance. While no algorithmic novelties are claimed, our paper advocates for a fundamental shift in data collection practices. We propose that real-world demonstrations in future research should utilise green screens, followed by the application of GreenAug. We believe GreenAug unlocks policy generalisation to visually distinct novel locations, addressing the current scene generalisation limitations in robot learning.
Some imitation learning methods combine behavioural cloning with self-supervision to infer actions from state pairs. However, most rely on a large number of expert trajectories to increase generalisation and human intervention to capture key aspects of the problem, such as domain constraints. In this paper, we propose Continuous Imitation Learning from Observation (CILO), a new method augmenting imitation learning with two important features: (i) exploration, allowing for more diverse state transitions, requiring less expert trajectories and resulting in fewer training iterations; and (ii) path signatures, allowing for automatic encoding of constraints, through the creation of non-parametric representations of agents and expert trajectories. We compared CILO with a baseline and two leading imitation learning methods in five environments. It had the best overall performance of all methods in all environments, outperforming the expert in two of them.