Mastering basketball skills such as diverse layups and dribbling involves complex interactions with the ball and requires real-time adjustments. Traditional reinforcement learning methods for interaction skills rely on labor-intensive, manually designed rewards that do not generalize well across different skills. Inspired by how humans learn from demonstrations, we propose SkillMimic, a data-driven approach that mimics both human and ball motions to learn a wide variety of basketball skills. SkillMimic employs a unified configuration to learn diverse skills from human-ball motion datasets, with skill diversity and generalization improving as the dataset grows. This approach allows training a single policy to learn multiple skills, enabling smooth skill switching even if these switches are not present in the reference dataset. The skills acquired by SkillMimic can be easily reused by a high-level controller to accomplish complex basketball tasks. To evaluate our approach, we introduce two basketball datasets: one estimated through monocular RGB videos and the other using advanced motion capture equipment, collectively containing about 35 minutes of diverse basketball skills. Experiments show that our method can effectively learn various basketball skills included in the dataset with a unified configuration, including various styles of dribbling, layups, and shooting. Furthermore, by training a high-level controller to reuse the acquired skills, we can achieve complex basketball tasks such as layup scoring, which involves dribbling toward the basket, timing the dribble and layup to score, retrieving the rebound, and repeating the process. The project page and video demonstrations are available at https://ingrid789.github.io/SkillMimic/

Federated Learning (FL) is a distributed machine learning approach that enables devices to collaboratively train models without sharing their local data, ensuring user privacy and scalability. However, applying FL to real-world data presents challenges, particularly as most existing FL research focuses on unimodal data. Multimodal Federated Learning (MFL) has emerged to address these challenges, leveraging modality-specific encoder models to process diverse datasets. Current MFL methods often uniformly allocate computational frequencies across all modalities, which is inefficient for IoT devices with limited resources. In this paper, we propose FlexMod, a novel approach to enhance computational efficiency in MFL by adaptively allocating training resources for each modality encoder based on their importance and training requirements. We employ prototype learning to assess the quality of modality encoders, use Shapley values to quantify the importance of each modality, and adopt the Deep Deterministic Policy Gradient (DDPG) method from deep reinforcement learning to optimize the allocation of training resources. Our method prioritizes critical modalities, optimizing model performance and resource utilization. Experimental results on three real-world datasets demonstrate that our proposed method significantly improves the performance of MFL models.

For both humans and robots, the sense of touch, known as tactile sensing, is critical for performing contact-rich manipulation tasks. Three key challenges in robotic tactile sensing are 1) interpreting sensor signals, 2) generating sensor signals in novel scenarios, and 3) learning sensor-based policies. For visuotactile sensors, interpretation has been facilitated by their close relationship with vision sensors (e.g., RGB cameras). However, generation is still difficult, as visuotactile sensors typically involve contact, deformation, illumination, and imaging, all of which are expensive to simulate; in turn, policy learning has been challenging, as simulation cannot be leveraged for large-scale data collection. We present \textbf{TacSL} (\textit{taxel}), a library for GPU-based visuotactile sensor simulation and learning. \textbf{TacSL} can be used to simulate visuotactile images and extract contact-force distributions over $200\times$ faster than the prior state-of-the-art, all within the widely-used Isaac Gym simulator. Furthermore, \textbf{TacSL} provides a learning toolkit containing multiple sensor models, contact-intensive training environments, and online/offline algorithms that can facilitate policy learning for sim-to-real applications. On the algorithmic side, we introduce a novel online reinforcement-learning algorithm called asymmetric actor-critic distillation (\sysName), designed to effectively and efficiently learn tactile-based policies in simulation that can transfer to the real world. Finally, we demonstrate the utility of our library and algorithms by evaluating the benefits of distillation and multimodal sensing for contact-rich manip ulation tasks, and most critically, performing sim-to-real transfer. Supplementary videos and results are at \url{https://iakinola23.github.io/tacsl/}.

Current visual navigation systems often treat the environment as static, lacking the ability to adaptively interact with obstacles. This limitation leads to navigation failure when encountering unavoidable obstructions. In response, we introduce IN-Sight, a novel approach to self-supervised path planning, enabling more effective navigation strategies through interaction with obstacles. Utilizing RGB-D observations, IN-Sight calculates traversability scores and incorporates them into a semantic map, facilitating long-range path planning in complex, maze-like environments. To precisely navigate around obstacles, IN-Sight employs a local planner, trained imperatively on a differentiable costmap using representation learning techniques. The entire framework undergoes end-to-end training within the state-of-the-art photorealistic Intel SPEAR Simulator. We validate the effectiveness of IN-Sight through extensive benchmarking in a variety of simulated scenarios and ablation studies. Moreover, we demonstrate the system's real-world applicability with zero-shot sim-to-real transfer, deploying our planner on the legged robot platform ANYmal, showcasing its practical potential for interactive navigation in real environments.

This paper establishes a new and comprehensive theoretical analysis for the application of reinforcement learning (RL) in high-frequency market making. We bridge the modern RL theory and the continuous-time statistical models in high-frequency financial economics. Different with most existing literature on methodological research about developing various RL methods for market making problem, our work is a pilot to provide the theoretical analysis. We target the effects of sampling frequency, and find an interesting tradeoff between error and complexity of RL algorithm when tweaking the values of the time increment $\Delta$ $-$ as $\Delta$ becomes smaller, the error will be smaller but the complexity will be larger. We also study the two-player case under the general-sum game framework and establish the convergence of Nash equilibrium to the continuous-time game equilibrium as $\Delta\rightarrow0$. The Nash Q-learning algorithm, which is an online multi-agent RL method, is applied to solve the equilibrium. Our theories are not only useful for practitioners to choose the sampling frequency, but also very general and applicable to other high-frequency financial decision making problems, e.g., optimal executions, as long as the time-discretization of a continuous-time markov decision process is adopted. Monte Carlo simulation evidence support all of our theories.

We propose RHEA CL, which combines Curriculum Learning (CL) with Rolling Horizon Evolutionary Algorithms (RHEA) to automatically produce effective curricula during the training of a reinforcement learning agent. RHEA CL optimizes a population of curricula, using an evolutionary algorithm, and selects the best-performing curriculum as the starting point for the next training epoch. Performance evaluations are conducted after every curriculum step in all environments. We evaluate the algorithm on the \textit{DoorKey} and \textit{DynamicObstacles} environments within the Minigrid framework. It demonstrates adaptability and consistent improvement, particularly in the early stages, while reaching a stable performance later that is capable of outperforming other curriculum learners. In comparison to other curriculum schedules, RHEA CL has been shown to yield performance improvements for the final Reinforcement learning (RL) agent at the cost of additional evaluation during training.

We study the problem of estimating the optimal Q-function of $\gamma$-discounted Markov decision processes (MDPs) under the synchronous setting, where independent samples for all state-action pairs are drawn from a generative model at each iteration. We introduce and analyze a novel model-free algorithm called Variance-Reduced Cascade Q-learning (VRCQ). VRCQ comprises two key building blocks: (i) the established direct variance reduction technique and (ii) our proposed variance reduction scheme, Cascade Q-learning. By leveraging these techniques, VRCQ provides superior guarantees in the $\ell_\infty$-norm compared with the existing model-free stochastic approximation-type algorithms. Specifically, we demonstrate that VRCQ is minimax optimal. Additionally, when the action set is a singleton (so that the Q-learning problem reduces to policy evaluation), it achieves non-asymptotic instance optimality while requiring the minimum number of samples theoretically possible. Our theoretical results and their practical implications are supported by numerical experiments.

Expected free energy (EFE) is a central quantity in active inference which has recently gained popularity due to its intuitive decomposition of the expected value of control into a pragmatic and an epistemic component. While numerous conjectures have been made to justify EFE as a decision making objective function, the most widely accepted is still its intuitiveness and resemblance to variational free energy in approximate Bayesian inference. In this work, we take a bottom up approach and ask: taking EFE as given, what's the resulting agent's optimality gap compared with a reward-driven reinforcement learning (RL) agent, which is well understood? By casting EFE under a particular class of belief MDP and using analysis tools from RL theory, we show that EFE approximates the Bayes optimal RL policy via information value. We discuss the implications for objective specification of active inference agents.

Large Language Models (LLMs) have demonstrated remarkable abilities in various language tasks, making them promising candidates for decision-making in robotics. Inspired by Hierarchical Reinforcement Learning (HRL), we propose Hierarchical in-Context Reinforcement Learning (HCRL), a novel framework that decomposes complex tasks into sub-tasks using an LLM-based high-level policy, in which a complex task is decomposed into sub-tasks by a high-level policy on-the-fly. The sub-tasks, defined by goals, are assigned to the low-level policy to complete. Once the LLM agent determines that the goal is finished, a new goal will be proposed. To improve the agent's performance in multi-episode execution, we propose Hindsight Modular Reflection (HMR), where, instead of reflecting on the full trajectory, we replace the task objective with intermediate goals and let the agent reflect on shorter trajectories to improve reflection efficiency. We evaluate the decision-making ability of the proposed HCRL in three benchmark environments--ALFWorld, Webshop, and HotpotQA. Results show that HCRL can achieve 9%, 42%, and 10% performance improvement in 5 episodes of execution over strong in-context learning baselines.

In model-based reinforcement learning (RL), an agent can leverage a learned model to improve its way of behaving in different ways. Two of the prevalent ways to do this are through decision-time and background planning methods. In this study, we are interested in understanding how the value-based versions of these two planning methods will compare against each other across different settings. Towards this goal, we first consider the simplest instantiations of value-based decision-time and background planning methods and provide theoretical results on which one will perform better in the regular RL and transfer learning settings. Then, we consider the modern instantiations of them and provide hypotheses on which one will perform better in the same settings. Finally, we perform illustrative experiments to validate these theoretical results and hypotheses. Overall, our findings suggest that even though value-based versions of the two planning methods perform on par in their simplest instantiations, the modern instantiations of value-based decision-time planning methods can perform on par or better than the modern instantiations of value-based background planning methods in both the regular RL and transfer learning settings.