Foundation models (FMs), large deep learning models pre-trained on vast, unlabeled datasets, exhibit powerful capabilities in understanding complex patterns and generating sophisticated outputs. However, they often struggle to adapt to specific tasks. Reinforcement learning (RL), which allows agents to learn through interaction and feedback, offers a compelling solution. Integrating RL with FMs enables these models to achieve desired outcomes and excel at particular tasks. Additionally, RL can be enhanced by leveraging the reasoning and generalization capabilities of FMs. This synergy is revolutionizing various fields, including robotics. FMs, rich in knowledge and generalization, provide robots with valuable information, while RL facilitates learning and adaptation through real-world interactions. This survey paper comprehensively explores this exciting intersection, examining how these paradigms can be integrated to advance robotic intelligence. We analyze the use of foundation models as action planners, the development of robotics-specific foundation models, and the mutual benefits of combining FMs with RL. Furthermore, we present a taxonomy of integration approaches, including large language models, vision-language models, diffusion models, and transformer-based RL models. We also explore how RL can utilize world representations learned from FMs to enhance robotic task execution. Our survey aims to synthesize current research and highlight key challenges in robotic reasoning and control, particularly in the context of integrating FMs and RL--two rapidly evolving technologies. By doing so, we seek to spark future research and emphasize critical areas that require further investigation to enhance robotics. We provide an updated collection of papers based on our taxonomy, accessible on our open-source project website at: https://github.com/clmoro/Robotics-RL-FMs-Integration.

Recent works have demonstrated the vulnerability of Deep Reinforcement Learning (DRL) algorithms against training-time, backdoor poisoning attacks. These attacks induce pre-determined, adversarial behavior in the agent upon observing a fixed trigger during deployment while allowing the agent to solve its intended task during training. Prior attacks rely on arbitrarily large perturbations to the agent's rewards to achieve both of these objectives - leaving them open to detection. Thus, in this work, we propose a new class of backdoor attacks against DRL which achieve state of the art performance while minimally altering the agent's rewards. These "inception" attacks train the agent to associate the targeted adversarial behavior with high returns by inducing a disjunction between the agent's chosen action and the true action executed in the environment during training. We formally define these attacks and prove they can achieve both adversarial objectives. We then devise an online inception attack which significantly out-performs prior attacks under bounded reward constraints.

Statistical inference tasks, such as constructing confidence regions or simultaneous confidence intervals, are often addressed by deriving distributional theory such as central limit theorems (CLTs) for the estimator in use. Due to the high dimensionality of modern science and engineering applications, there has been a surge of interests in deriving convergence results that are valid in a finite-sample manner. In Reinforcement Learning (RL), a discipline that underpins many recent machine learning breakthroughs (Agarwal et al. (2019); Sutton and Barto (2018)) one central question is to evaluate with confidence the quality of a given policy, measured by its value function. As RL is often modeled as decision making in Markov decision processes (MDPs), the task of statistical inference needs to accommodate the online nature of the sampling mechanism. Temporal Difference (TD) learning (Sutton (1988)) is arguably the most widely used algorithm designed for this purpose. TD learning, which is an instance of stochastic approximation (SA) method (Robbins and Monro (1951)), approximates the value function of a given policy in an iterative manner. Towards understanding the non-asymptotic convergence rate of TD to the target value function, extensive recent efforts have been made (see, e.g.

Reinforcement learning (RL) policies are prone to high-frequency oscillations, especially undesirable when deploying to hardware in the real-world. In this paper, we identify, categorize, and compare methods from the literature that aim to mitigate high-frequency oscillations in deep RL. We define two broad classes: loss regularization and architectural methods. At their core, these methods incentivize learning a smooth mapping, such that nearby states in the input space produce nearby actions in the output space. We present benchmarks in terms of policy performance and control smoothness on traditional RL environments from the Gymnasium and a complex manipulation task, as well as three robotics locomotion tasks that include deployment and evaluation with real-world hardware. Finally, we also propose hybrid methods that combine elements from both loss regularization and architectural methods. We find that the best-performing hybrid outperforms other methods, and improves control smoothness by 26.8% over the baseline, with a worst-case performance degradation of just 2.8%.

StarCraft Multi-Agent Challenge (SMAC) is one of the most commonly used experimental environments in multi-agent reinforcement learning (MARL), where the specific task is to control a set number of allied units to defeat enemy forces. Traditional MARL algorithms often require interacting with the environment for up to 1 million steps to train a model, and the resulting policies are typically non-interpretable with weak transferability. In this paper, we propose a novel approach to solving SMAC tasks called LLM-SMAC. In our framework, agents leverage large language models (LLMs) to generate decision tree code by providing task descriptions. The model is further self-reflection using feedback from the rewards provided by the environment. We conduct experiments in the SMAC and demonstrate that our method can produce high-quality, interpretable decision trees with minimal environmental exploration. Moreover, these models exhibit strong transferability, successfully applying to similar SMAC environments without modification. We believe this approach offers a new direction for solving decision-making tasks in the future.

In recent years, reinforcement learning (RL) has gained popularity and has been applied to a wide range of tasks. One such popular domain where RL has been effective is resource management problems in systems. We look to extend work on RL for resource management problems by considering the novel domain of dynamic memory allocation management. We consider dynamic memory allocation to be a suitable domain for RL since current algorithms like first-fit, best-fit, and worst-fit can fail to adapt to changing conditions and can lead to fragmentation and suboptimal efficiency. In this paper, we present a framework in which an RL agent continuously learns from interactions with the system to improve memory management tactics. We evaluate our approach through various experiments using high-level and low-level action spaces and examine different memory allocation patterns. Our results show that RL can successfully train agents that can match and surpass traditional allocation strategies, particularly in environments characterized by adversarial request patterns. We also explore the potential of history-aware policies that leverage previous allocation requests to enhance the allocator's ability to handle complex request patterns. Overall, we find that RL offers a promising avenue for developing more adaptive and efficient memory allocation strategies, potentially overcoming limitations of hardcoded allocation algorithms.

A critical goal in robotics and autonomy is to teach robots to adapt to real-world collaborative tasks, particularly in automatic assembly. The ability of a robot to understand the original intent of an incomplete assembly and complete missing features without human instruction is valuable but challenging. This paper introduces 3D combinatorial assembly completion, which is demonstrated using combinatorial unit primitives (i.e., Lego bricks). Combinatorial assembly is challenging due to the possible assembly combinations and complex physical constraints (e.g., no brick collisions, structure stability, inventory constraints, etc.). To address these challenges, we propose a two-part deep reinforcement learning (DRL) framework that tackles teaching the robot to understand the objective of an incomplete assembly and learning a construction policy to complete the assembly. The robot queries a stable object library to facilitate assembly inference and guide learning. In addition to the robot policy, an action mask is developed to rule out invalid actions that violate physical constraints for object-oriented construction. We demonstrate the proposed framework's feasibility and robustness in a variety of assembly scenarios in which the robot satisfies real-life assembly with respect to both solution and runtime quality. Furthermore, results demonstrate that the proposed framework effectively infers and assembles incomplete structures for unseen and unique object types.

Reinforcement learning (RL) faces challenges in trajectory planning for urban automated driving due to the poor convergence of RL and the difficulty in designing reward functions. The convergence problem is alleviated by combining RL with supervised learning. However, most existing approaches only reason one step ahead and lack the capability to plan for multiple future steps. Besides, although inverse reinforcement learning holds promise for solving the reward function design issue, existing methods for automated driving impose a linear structure assumption on reward functions, making them difficult to apply to urban automated driving. In light of these challenges, this paper proposes a novel RL-based trajectory planning method that integrates RL with imitation learning to enable multi-step planning. Furthermore, a transformer-based Bayesian reward function is developed, providing effective reward signals for RL in urban scenarios. Moreover, a hybrid-driven trajectory planning framework is proposed to enhance safety and interpretability. The proposed methods were validated on the large-scale real-world urban automated driving nuPlan dataset. The results demonstrated the significant superiority of the proposed methods over the baselines in terms of the closed-loop metrics. The code is available at https://github.com/Zigned/nuplan_zigned.

Spatial Crowdsourcing (SC) is gaining traction in both academia and industry, with tasks on SC platforms becoming increasingly complex and requiring collaboration among workers with diverse skills. Recent research works address complex tasks by dividing them into subtasks with dependencies and assigning them to suitable workers. However, the dependencies among subtasks and their heterogeneous skill requirements, as well as the need for efficient utilization of workers' limited work time in the multi-task allocation mode, pose challenges in achieving an optimal task allocation scheme. Therefore, this paper formally investigates the problem of Dependency-aware Multi-task Allocation (DMA) and presents a well-designed framework to solve it, known as Heterogeneous Graph Reinforcement Learning-based Task Allocation (HGRL-TA). To address the challenges associated with representing and embedding diverse problem instances to ensure robust generalization, we propose a multi-relation graph model and a Compound-path-based Heterogeneous Graph Attention Network (CHANet) for effectively representing and capturing intricate relations among tasks and workers, as well as providing embedding of problem state. The task allocation decision is determined sequentially by a policy network, which undergoes simultaneous training with CHANet using the proximal policy optimization algorithm. Extensive experiment results demonstrate the effectiveness and generality of the proposed HGRL-TA in solving the DMA problem, leading to average profits that is 21.78% higher than those achieved using the metaheuristic methods.

We study the problem of Non-Stationary Reinforcement Learning (NS-RL) without prior knowledge about the system's non-stationarity. A state-of-the-art, black-box algorithm, known as MASTER, is considered, with a focus on identifying the conditions under which it can achieve its stated goals. Specifically, we prove that MASTER's non-stationarity detection mechanism is not triggered for practical choices of horizon, leading to performance akin to a random restarting algorithm. Moreover, we show that the regret bound for MASTER, while being order optimal, stays above the worst-case linear regret until unreasonably large values of the horizon. To validate these observations, MASTER is tested for the special case of piecewise stationary multi-armed bandits, along with methods that employ random restarting, and others that use quickest change detection to restart. A simple, order optimal random restarting algorithm, that has prior knowledge of the non-stationarity is proposed as a baseline. The behavior of the MASTER algorithm is validated in simulations, and it is shown that methods employing quickest change detection are more robust and consistently outperform MASTER and other random restarting approaches.