Julia Kiseleva, Ziming Li, Mohammad Aliannejadi et al. · meta-ai, mit
Human intelligence has the remarkable ability to quickly adapt to new tasks and environments. Starting from a very young age, humans acquire new skills and learn how to solve new tasks either by imitating the behavior of others or by following provided natural language instructions. To facilitate research in this direction, we propose \emph{IGLU: Interactive Grounded Language Understanding in a Collaborative Environment}. The primary goal of the competition is to approach the problem of how to build interactive agents that learn to solve a task while provided with grounded natural language instructions in a collaborative environment. Understanding the complexity of the challenge, we split it into sub-tasks to make it feasible for participants.
Julia Kiseleva, Alexey Skrynnik, Artem Zholus et al. · meta-ai, microsoft-research
Human intelligence has the remarkable ability to adapt to new tasks and environments quickly. Starting from a very young age, humans acquire new skills and learn how to solve new tasks either by imitating the behavior of others or by following provided natural language instructions. To facilitate research in this direction, we propose IGLU: Interactive Grounded Language Understanding in a Collaborative Environment. The primary goal of the competition is to approach the problem of how to develop interactive embodied agents that learn to solve a task while provided with grounded natural language instructions in a collaborative environment. Understanding the complexity of the challenge, we split it into sub-tasks to make it feasible for participants. This research challenge is naturally related, but not limited, to two fields of study that are highly relevant to the NeurIPS community: Natural Language Understanding and Generation (NLU/G) and Reinforcement Learning (RL). Therefore, the suggested challenge can bring two communities together to approach one of the crucial challenges in AI. Another critical aspect of the challenge is the dedication to perform a human-in-the-loop evaluation as a final evaluation for the agents developed by contestants.
Human intelligence can remarkably adapt quickly to new tasks and environments. Starting from a very young age, humans acquire new skills and learn how to solve new tasks either by imitating the behavior of others or by following provided natural language instructions. To facilitate research which can enable similar capabilities in machines, we made the following contributions (1) formalized the collaborative embodied agent using natural language task; (2) developed a tool for extensive and scalable data collection; and (3) collected the first dataset for interactive grounded language understanding.
Artem Zholus, Alexey Skrynnik, Shrestha Mohanty et al. · microsoft-research, mit
We present the IGLU Gridworld: a reinforcement learning environment for building and evaluating language conditioned embodied agents in a scalable way. The environment features visual agent embodiment, interactive learning through collaboration, language conditioned RL, and combinatorically hard task (3d blocks building) space.
Alexey Skrynnik, Zoya Volovikova, Marc-Alexandre Côté et al. · microsoft-research, mit
The adoption of pre-trained language models to generate action plans for embodied agents is a promising research strategy. However, execution of instructions in real or simulated environments requires verification of the feasibility of actions as well as their relevance to the completion of a goal. We propose a new method that combines a language model and reinforcement learning for the task of building objects in a Minecraft-like environment according to the natural language instructions. Our method first generates a set of consistently achievable sub-goals from the instructions and then completes associated sub-tasks with a pre-trained RL policy. The proposed method formed the RL baseline at the IGLU 2022 competition.
Alexey Skrynnik, Anton Andreychuk, Konstantin Yakovlev et al.
We introduce POGEMA (https://github.com/AIRI-Institute/pogema) a sandbox for challenging partially observable multi-agent pathfinding (PO-MAPF) problems . This is a grid-based environment that was specifically designed to be a flexible, tunable and scalable benchmark. It can be tailored to a variety of PO-MAPF, which can serve as an excellent testing ground for planning and learning methods, and their combination, which will allow us to move towards filling the gap between AI planning and learning.
Stepan Dergachev, Artem Pshenitsyn, Aleksandr Panov et al.
Decentralized collision avoidance is a core challenge for scalable multi-robot systems. A promising approach to this problem is Model Predictive Path Integral (MPPI) control - a framework that naturally handles arbitrary motion models and provides strong theoretical guarantees. Still, in practice an MPPI-based controller may produce suboptimal trajectories because its performance relies heavily on uninformed random sampling. We introduce CoRL-MPPI, a fusion of Cooperative Reinforcement Learning and MPPI that addresses this limitation. We train an action policy, approximated by a deep neural network, in simulation to learn local cooperative collision-avoidance behaviors. This learned policy is then embedded into the MPPI framework to guide its sampling distribution, biasing it toward more intelligent and cooperative actions in scenarios that may differ substantially from those used during training. Moreover, CoRL-MPPI preserves the theoretical guarantees of regular MPPI. We evaluate our approach in dense, dynamic setups against classical and learning-based state-of-the-art baselines. Our results demonstrate that CoRL-MPPI outperforms competing methods and significantly improves navigation efficiency, measured by success rate and delay, as well as safety, enabling agile and robust multi-robot navigation.
Dmitry Yudin, Yaroslav Solomentsev, Ruslan Musaev et al.
We present a novel dataset named as HPointLoc, specially designed for exploring capabilities of visual place recognition in indoor environment and loop detection in simultaneous localization and mapping. The loop detection sub-task is especially relevant when a robot with an on-board RGB-D camera can drive past the same place (``Point") at different angles. The dataset is based on the popular Habitat simulator, in which it is possible to generate photorealistic indoor scenes using both own sensor data and open datasets, such as Matterport3D. To study the main stages of solving the place recognition problem on the HPointLoc dataset, we proposed a new modular approach named as PNTR. It first performs an image retrieval with the Patch-NetVLAD method, then extracts keypoints and matches them using R2D2, LoFTR or SuperPoint with SuperGlue, and finally performs a camera pose optimization step with TEASER++. Such a solution to the place recognition problem has not been previously studied in existing publications. The PNTR approach has shown the best quality metrics on the HPointLoc dataset and has a high potential for real use in localization systems for unmanned vehicles. The proposed dataset and framework are publicly available: https://github.com/metra4ok/HPointLoc.
Aleksei Staroverov, Muhammad Alhaddad, Aditya Narendra et al.
Generalist robot policies must operate safely and reliably in everyday human environments such as homes, offices, and warehouses, where people and objects move unpredictably. We present Dynamic Neural Potential Field (NPField-GPT), a learning-enhanced model predictive control (MPC) framework that couples classical optimization with a Transformer-based predictor of footprint-aware repulsive potentials. Given an occupancy sub-map, robot footprint, and optional dynamic-obstacle cues, our NPField-GPT model forecasts a horizon of differentiable potentials that are injected into a sequential quadratic MPC program via L4CasADi, yielding real-time, constraint-aware trajectory optimization. We additionally study two baselines: NPField-StaticMLP, where a dynamic scene is treated as a sequence of static maps; and NPField-DynamicMLP, which predicts the future potential sequence in parallel with an MLP. In dynamic indoor scenarios from BenchMR and on a Husky UGV in office corridors, NPField-GPT produces more efficient and safer trajectories under motion changes, while StaticMLP/DynamicMLP offer lower latency. We also compare with the CIAO* and MPPI baselines. Across methods, the Transformer+MPC synergy preserves the transparency and stability of model-based planning while learning only the part that benefits from data: spatiotemporal collision risk. Code and trained models are available at https://github.com/CognitiveAISystems/Dynamic-Neural-Potential-Field
Tatiana Zemskova, Aleksei Staroverov, Kirill Muravyev et al.
Visual object navigation using learning methods is one of the key tasks in mobile robotics. This paper introduces a new representation of a scene semantic map formed during the embodied agent interaction with the indoor environment. It is based on a neural network method that adjusts the weights of the segmentation model with backpropagation of the predicted fusion loss values during inference on a regular (backward) or delayed (forward) image sequence. We have implemented this representation into a full-fledged navigation approach called SkillTron, which can select robot skills from end-to-end policies based on reinforcement learning and classic map-based planning methods. The proposed approach makes it possible to form both intermediate goals for robot exploration and the final goal for object navigation. We conducted intensive experiments with the proposed approach in the Habitat environment, which showed a significant superiority in navigation quality metrics compared to state-of-the-art approaches. The developed code and used custom datasets are publicly available at github.com/AIRI-Institute/skill-fusion.
Tatiana Zemskova, Margarita Kichik, Dmitry Yudin et al.
This paper presents an adaptive transformer model named SegmATRon for embodied image semantic segmentation. Its distinctive feature is the adaptation of model weights during inference on several images using a hybrid multicomponent loss function. We studied this model on datasets collected in the photorealistic Habitat and the synthetic AI2-THOR Simulators. We showed that obtaining additional images using the agent's actions in an indoor environment can improve the quality of semantic segmentation. The code of the proposed approach and datasets are publicly available at https://github.com/wingrune/SegmATRon.
Alexey Skrynnik, Anton Andreychuk, Maria Nesterova et al.
Multi-agent Pathfinding (MAPF) problem generally asks to find a set of conflict-free paths for a set of agents confined to a graph and is typically solved in a centralized fashion. Conversely, in this work, we investigate the decentralized MAPF setting, when the central controller that posses all the information on the agents' locations and goals is absent and the agents have to sequientially decide the actions on their own without having access to a full state of the environment. We focus on the practically important lifelong variant of MAPF, which involves continuously assigning new goals to the agents upon arrival to the previous ones. To address this complex problem, we propose a method that integrates two complementary approaches: planning with heuristic search and reinforcement learning through policy optimization. Planning is utilized to construct and re-plan individual paths. We enhance our planning algorithm with a dedicated technique tailored to avoid congestion and increase the throughput of the system. We employ reinforcement learning to discover the collision avoidance policies that effectively guide the agents along the paths. The policy is implemented as a neural network and is effectively trained without any reward-shaping or external guidance. We evaluate our method on a wide range of setups comparing it to the state-of-the-art solvers. The results show that our method consistently outperforms the learnable competitors, showing higher throughput and better ability to generalize to the maps that were unseen at the training stage. Moreover our solver outperforms a rule-based one in terms of throughput and is an order of magnitude faster than a state-of-the-art search-based solver.
Muhammad Alhaddad, Konstantin Mironov, Aleksey Staroverov et al.
Model predictive control (MPC) may provide local motion planning for mobile robotic platforms. The challenging aspect is the analytic representation of collision cost for the case when both the obstacle map and robot footprint are arbitrary. We propose a Neural Potential Field: a neural network model that returns a differentiable collision cost based on robot pose, obstacle map, and robot footprint. The differentiability of our model allows its usage within the MPC solver. It is computationally hard to solve problems with a very high number of parameters. Therefore, our architecture includes neural image encoders, which transform obstacle maps and robot footprints into embeddings, which reduce problem dimensionality by two orders of magnitude. The reference data for network training are generated based on algorithmic calculation of a signed distance function. Comparative experiments showed that the proposed approach is comparable with existing local planners: it provides trajectories with outperforming smoothness, comparable path length, and safe distance from obstacles. Experiment on Husky UGV mobile robot showed that our approach allows real-time and safe local planning. The code for our approach is presented at https://github.com/cog-isa/NPField together with demo video.
Daniil Kirilenko, Anton Andreychuk, Aleksandr Panov et al.
Heuristic search algorithms, e.g. A*, are the commonly used tools for pathfinding on grids, i.e. graphs of regular structure that are widely employed to represent environments in robotics, video games etc. Instance-independent heuristics for grid graphs, e.g. Manhattan distance, do not take the obstacles into account and, thus, the search led by such heuristics performs poorly in the obstacle-rich environments. To this end, we suggest learning the instance-dependent heuristic proxies that are supposed to notably increase the efficiency of the search. The first heuristic proxy we suggest to learn is the correction factor, i.e. the ratio between the instance independent cost-to-go estimate and the perfect one (computed offline at the training phase). Unlike learning the absolute values of the cost-to-go heuristic function, which was known before, when learning the correction factor the knowledge of the instance-independent heuristic is utilized. The second heuristic proxy is the path probability, which indicates how likely the grid cell is lying on the shortest path. This heuristic can be utilized in the Focal Search framework as the secondary heuristic, allowing us to preserve the guarantees on the bounded sub-optimality of the solution. We learn both suggested heuristics in a supervised fashion with the state-of-the-art neural networks containing attention blocks (transformers). We conduct a thorough empirical evaluation on a comprehensive dataset of planning tasks, showing that the suggested techniques i) reduce the computational effort of the A* up to a factor of $4$x while producing the solutions, which costs exceed the costs of the optimal solutions by less than $0.3$% on average; ii) outperform the competitors, which include the conventional techniques from the heuristic search, i.e. weighted A*, as well as the state-of-the-art learnable planners.
Alexandr V. Korchemnyi, Anatoly O. Onishchenko, Eva A. Bakaeva et al.
Embodied agents tasked with complex scenarios, whether in real or simulated environments, rely heavily on robust planning capabilities. When instructions are formulated in natural language, large language models (LLMs) equipped with extensive linguistic knowledge can play this role. However, to effectively exploit the ability of such models to handle linguistic ambiguity, to retrieve information from the environment, and to be based on the available skills of an agent, an appropriate architecture must be designed. We propose a Hierarchical Embodied Language Planner, called HELP, consisting of a set of LLM-based agents, each dedicated to solving a different subtask. We evaluate the proposed approach on a household task and perform real-world experiments with an embodied agent. We also focus on the use of open source LLMs with a relatively small number of parameters, to enable autonomous deployment.
Yelisey Pitanov, Alexey Skrynnik, Anton Andreychuk et al.
In this work we study a well-known and challenging problem of Multi-agent Pathfinding, when a set of agents is confined to a graph, each agent is assigned a unique start and goal vertices and the task is to find a set of collision-free paths (one for each agent) such that each agent reaches its respective goal. We investigate how to utilize Monte-Carlo Tree Search (MCTS) to solve the problem. Although MCTS was shown to demonstrate superior performance in a wide range of problems like playing antagonistic games (e.g. Go, Chess etc.), discovering faster matrix multiplication algorithms etc., its application to the problem at hand was not well studied before. To this end we introduce an original variant of MCTS, tailored to multi-agent pathfinding. The crux of our approach is how the reward, that guides MCTS, is computed. Specifically, we use individual paths to assist the agents with the the goal-reaching behavior, while leaving them freedom to get off the track if it is needed to avoid collisions. We also use a dedicated decomposition technique to reduce the branching factor of the tree search procedure. Empirically we show that the suggested method outperforms the baseline planning algorithm that invokes heuristic search, e.g. A*, at each re-planning step.
Anton Andreychuk, Konstantin Yakovlev, Aleksandr Panov et al.
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.
Alexey Skrynnik, Anton Andreychuk, Anatolii Borzilov et al.
Multi-agent reinforcement learning (MARL) has recently excelled in solving challenging cooperative and competitive multi-agent problems in various environments, typically involving a small number of agents and full observability. Moreover, a range of crucial robotics-related tasks, such as multi-robot pathfinding, which have traditionally been approached with classical non-learnable methods (e.g., heuristic search), are now being suggested for solution using learning-based or hybrid methods. However, in this domain, it remains difficult, if not impossible, to conduct a fair comparison between classical, learning-based, and hybrid approaches due to the lack of a unified framework that supports both learning and evaluation. To address this, we introduce POGEMA, a comprehensive set of tools that includes a fast environment for learning, a problem instance generator, a collection of predefined problem instances, a visualization toolkit, and a benchmarking tool for automated evaluation. We also introduce and define an evaluation protocol that specifies a range of domain-related metrics, computed based on primary evaluation indicators (such as success rate and path length), enabling a fair multi-fold comparison. The results of this comparison, which involves a variety of state-of-the-art MARL, search-based, and hybrid methods, are presented.
Many challenging reinforcement learning (RL) problems require designing a distribution of tasks that can be applied to train effective policies. This distribution of tasks can be specified by the curriculum. A curriculum is meant to improve the results of learning and accelerate it. We introduce Success Induced Task Prioritization (SITP), a framework for automatic curriculum learning, where a task sequence is created based on the success rate of each task. In this setting, each task is an algorithmically created environment instance with a unique configuration. The algorithm selects the order of tasks that provide the fastest learning for agents. The probability of selecting any of the tasks for the next stage of learning is determined by evaluating its performance score in previous stages. Experiments were carried out in the Partially Observable Grid Environment for Multiple Agents (POGEMA) and Procgen benchmark. We demonstrate that SITP matches or surpasses the results of other curriculum design methods. Our method can be implemented with handful of minor modifications to any standard RL framework and provides useful prioritization with minimal computational overhead.
The reinforcement learning research area contains a wide range of methods for solving the problems of intelligent agent control. Despite the progress that has been made, the task of creating a highly autonomous agent is still a significant challenge. One potential solution to this problem is intrinsic motivation, a concept derived from developmental psychology. This review considers the existing methods for determining intrinsic motivation based on the world model obtained by the agent. We propose a systematic approach to current research in this field, which consists of three categories of methods, distinguished by the way they utilize a world model in the agent's components: complementary intrinsic reward, exploration policy, and intrinsically motivated goals. The proposed unified framework describes the architecture of agents using a world model and intrinsic motivation to improve learning. The potential for developing new techniques in this area of research is also examined.
Artem Tsypin, Leonid Ugadiarov, Kuzma Khrabrov et al.
Molecular conformation optimization is crucial to computer-aided drug discovery and materials design. Traditional energy minimization techniques rely on iterative optimization methods that use molecular forces calculated by a physical simulator (oracle) as anti-gradients. However, this is a computationally expensive approach that requires many interactions with a physical simulator. One way to accelerate this procedure is to replace the physical simulator with a neural network. Despite recent progress in neural networks for molecular conformation energy prediction, such models are prone to distribution shift, leading to inaccurate energy minimization. We find that the quality of energy minimization with neural networks can be improved by providing optimization trajectories as additional training data. Still, it takes around $5 \times 10^5$ additional conformations to match the physical simulator's optimization quality. In this work, we present the Gradual Optimization Learning Framework (GOLF) for energy minimization with neural networks that significantly reduces the required additional data. The framework consists of an efficient data-collecting scheme and an external optimizer. The external optimizer utilizes gradients from the energy prediction model to generate optimization trajectories, and the data-collecting scheme selects additional training data to be processed by the physical simulator. Our results demonstrate that the neural network trained with GOLF performs on par with the oracle on a benchmark of diverse drug-like molecules using $50$x less additional data.
Zoya Volovikova, Alexey Skrynnik, Petr Kuderov et al.
In this study, we address the issue of enabling an artificial intelligence agent to execute complex language instructions within virtual environments. In our framework, we assume that these instructions involve intricate linguistic structures and multiple interdependent tasks that must be navigated successfully to achieve the desired outcomes. To effectively manage these complexities, we propose a hierarchical framework that combines the deep language comprehension of large language models with the adaptive action-execution capabilities of reinforcement learning agents. The language module (based on LLM) translates the language instruction into a high-level action plan, which is then executed by a pre-trained reinforcement learning agent. We have demonstrated the effectiveness of our approach in two different environments: in IGLU, where agents are instructed to build structures, and in Crafter, where agents perform tasks and interact with objects in the surrounding environment according to language commands.
Daniil Kirilenko, Vitaliy Vorobyov, Alexey K. Kovalev et al.
Object-centric architectures usually apply a differentiable module to the entire feature map to decompose it into sets of entity representations called slots. Some of these methods structurally resemble clustering algorithms, where the cluster's center in latent space serves as a slot representation. Slot Attention is an example of such a method, acting as a learnable analog of the soft k-means algorithm. Our work employs a learnable clustering method based on the Gaussian Mixture Model. Unlike other approaches, we represent slots not only as centers of clusters but also incorporate information about the distance between clusters and assigned vectors, leading to more expressive slot representations. Our experiments demonstrate that using this approach instead of Slot Attention improves performance in object-centric scenarios, achieving state-of-the-art results in the set property prediction task.
Evgenii Dzhivelikian, Petr Kuderov, Aleksandr I. Panov
This paper presents a novel approach to address the challenge of online sequence learning for decision making under uncertainty in non-stationary, partially observable environments. The proposed algorithm, Distributed Hebbian Temporal Memory (DHTM), is based on the factor graph formalism and a multi-component neuron model. DHTM aims to capture sequential data relationships and make cumulative predictions about future observations, forming Successor Features (SFs). Inspired by neurophysiological models of the neocortex, the algorithm uses distributed representations, sparse transition matrices, and local Hebbian-like learning rules to overcome the instability and slow learning of traditional temporal memory algorithms such as RNN and HMM. Experimental results show that DHTM outperforms LSTM, RWKV and a biologically inspired HMM-like algorithm, CSCG, on non-stationary data sets. Our results suggest that DHTM is a promising approach to address the challenges of online sequence learning and planning in dynamic environments.
Alexey Skrynnik, Anton Andreychuk, Konstantin Yakovlev et al.
The Multi-Agent Pathfinding (MAPF) problem involves finding a set of conflict-free paths for a group of agents confined to a graph. In typical MAPF scenarios, the graph and the agents' starting and ending vertices are known beforehand, allowing the use of centralized planning algorithms. However, in this study, we focus on the decentralized MAPF setting, where the agents may observe the other agents only locally and are restricted in communications with each other. Specifically, we investigate the lifelong variant of MAPF, where new goals are continually assigned to the agents upon completion of previous ones. Drawing inspiration from the successful AlphaZero approach, we propose a decentralized multi-agent Monte Carlo Tree Search (MCTS) method for MAPF tasks. Our approach utilizes the agent's observations to recreate the intrinsic Markov decision process, which is then used for planning with a tailored for multi-agent tasks version of neural MCTS. The experimental results show that our approach outperforms state-of-the-art learnable MAPF solvers. The source code is available at https://github.com/AIRI-Institute/mats-lp.
Brian Angulo, Gregory Gorbov, Aleksandr Panov et al.
While reinforcement learning algorithms have had great success in the field of autonomous navigation, they cannot be straightforwardly applied to the real autonomous systems without considering the safety constraints. The later are crucial to avoid unsafe behaviors of the autonomous vehicle on the road. To highlight the importance of these constraints, in this study, we compare two learnable navigation policies: safe and unsafe. The safe policy takes the constraints into account, while the other does not. We show that the safe policy is able to generate trajectories with more clearance (distance to the obstacles) and makes less collisions while training without sacrificing the overall performance.
Anatoly O. Onishchenko, Alexey K. Kovalev, Aleksandr I. Panov
Methods that use Large Language Models (LLM) as planners for embodied instruction following tasks have become widespread. To successfully complete tasks, the LLM must be grounded in the environment in which the robot operates. One solution is to use a scene graph that contains all the necessary information. Modern methods rely on prebuilt scene graphs and assume that all task-relevant information is available at the start of planning. However, these approaches do not account for changes in the environment that may occur between the graph construction and the task execution. We propose LookPlanGraph - a method that leverages a scene graph composed of static assets and object priors. During plan execution, LookPlanGraph continuously updates the graph with relevant objects, either by verifying existing priors or discovering new entities. This is achieved by processing the agents egocentric camera view using a Vision Language Model. We conducted experiments with changed object positions VirtualHome and OmniGibson simulated environments, demonstrating that LookPlanGraph outperforms methods based on predefined static scene graphs. To demonstrate the practical applicability of our approach, we also conducted experiments in a real-world setting. Additionally, we introduce the GraSIF (Graph Scenes for Instruction Following) dataset with automated validation framework, comprising 514 tasks drawn from SayPlan Office, BEHAVIOR-1K, and VirtualHome RobotHow. Project page available at https://lookplangraph.github.io .
Danil S. Grigorev, Alexey K. Kovalev, Aleksandr I. Panov
In the field of robotics, researchers face a critical challenge in ensuring reliable and efficient task planning. Verifying high-level task plans before execution significantly reduces errors and enhance the overall performance of these systems. In this paper, we propose an architecture for automatically verifying high-level task plans before their execution in simulator or real-world environments. Leveraging Large Language Models (LLMs), our approach consists of two key steps: first, the conversion of natural language instructions into Linear Temporal Logic (LTL), followed by a comprehensive analysis of action sequences. The module uses the reasoning capabilities of the LLM to evaluate logical coherence and identify potential gaps in the plan. Rigorous testing on datasets of varying complexity demonstrates the broad applicability of the module to household tasks. We contribute to improving the reliability and efficiency of task planning and addresses the critical need for robust pre-execution verification in autonomous systems. The code is available at https://verifyllm.github.io.
Nikita Oskolkov, Huzhenyu Zhang, Dmitry Makarov et al.
The 3D scene graph models spatial relationships between objects, enabling the agent to efficiently navigate in a partially observable environment and predict the location of the target object.This paper proposes an original framework named SGN-CIRL (3D Scene Graph-Based Reinforcement Learning Navigation) for mapless reinforcement learning-based robot navigation with learnable representation of open-vocabulary 3D scene graph. To accelerate and stabilize the training of reinforcement learning-based algorithms, the framework also employs imitation learning and curriculum learning. The first one enables the agent to learn from demonstrations, while the second one structures the training process by gradually increasing task complexity from simple to more advanced scenarios. Numerical experiments conducted in the Isaac Sim environment showed that using a 3D scene graph for reinforcement learning significantly increased the success rate in difficult navigation cases. The code is open-sourced and available at: https://github.com/Xisonik/Aloha\_graph.
Anastasiia Ivanova, Eva Bakaeva, Zoya Volovikova et al.
As a part of an embodied agent, Large Language Models (LLMs) are typically used for behavior planning given natural language instructions from the user. However, dealing with ambiguous instructions in real-world environments remains a challenge for LLMs. Various methods for task ambiguity detection have been proposed. However, it is difficult to compare them because they are tested on different datasets and there is no universal benchmark. For this reason, we propose AmbiK (Ambiguous Tasks in Kitchen Environment), the fully textual dataset of ambiguous instructions addressed to a robot in a kitchen environment. AmbiK was collected with the assistance of LLMs and is human-validated. It comprises 1000 pairs of ambiguous tasks and their unambiguous counterparts, categorized by ambiguity type (Human Preferences, Common Sense Knowledge, Safety), with environment descriptions, clarifying questions and answers, user intents, and task plans, for a total of 2000 tasks. We hope that AmbiK will enable researchers to perform a unified comparison of ambiguity detection methods. AmbiK is available at https://github.com/cog-model/AmbiK-dataset.
Brian Angulo, Gregory Gorbov, Aleksandr Panov et al.
Reinforcement learning is a widely used approach to autonomous navigation, showing potential in various tasks and robotic setups. Still, it often struggles to reach distant goals when safety constraints are imposed (e.g., the wheeled robot is prohibited from moving close to the obstacles). One of the main reasons for poor performance in such setups, which is common in practice, is that the need to respect the safety constraints degrades the exploration capabilities of an RL agent. To this end, we introduce a novel learnable algorithm that is based on decomposing the initial problem into smaller sub-problems via intermediate goals, on the one hand, and respects the limit of the cumulative safety constraints, on the other hand -- SPEIS(Safe Policy Exploration Improvement via Subgoals). It comprises the two coupled policies trained end-to-end: subgoal and safe. The subgoal policy is trained to generate the subgoal based on the transitions from the buffer of the safe (main) policy that helps the safe policy to reach distant goals. Simultaneously, the safe policy maximizes its rewards while attempting not to violate the limit of the cumulative safety constraints, thus providing a certain level of safety. We evaluate SPEIS in a wide range of challenging (simulated) environments that involve different types of robots in two different environments: autonomous vehicles from the POLAMP environment and car, point, doggo, and sweep from the safety-gym environment. We demonstrate that our method consistently outperforms state-of-the-art competitors and can significantly reduce the collision rate while maintaining high success rates (higher by 80% compared to the best-performing methods).
Leonid Ugadiarov, Alexey Skrynnik, Aleksandr I. Panov
Exploration is an essential part of reinforcement learning, which restricts the quality of learned policy. Hard-exploration environments are defined by huge state space and sparse rewards. In such conditions, an exhaustive exploration of the environment is often impossible, and the successful training of an agent requires a lot of interaction steps. In this paper, we propose an exploration method called Rollback-Explore (RbExplore), which utilizes the concept of the persistent Markov decision process, in which agents during training can roll back to visited states. We test our algorithm in the hard-exploration Prince of Persia game, without rewards and domain knowledge. At all used levels of the game, our agent outperforms or shows comparable results with state-of-the-art curiosity methods with knowledge-based intrinsic motivation: ICM and RND. An implementation of RbExplore can be found at https://github.com/cds-mipt/RbExplore.
Andrey Gorodetskiy, Konstantin Mironov, Aleksandr Panov
The application of learning-based control methods in robotics presents significant challenges. One is that model-free reinforcement learning algorithms use observation data with low sample efficiency. To address this challenge, a prevalent approach is model-based reinforcement learning, which involves employing an environment dynamics model. We suggest approximating transition dynamics with symbolic expressions, which are generated via symbolic regression. Approximation of a mechanical system with a symbolic model has fewer parameters than approximation with neural networks, which can potentially lead to higher accuracy and quality of extrapolation. We use a symbolic dynamics model to generate trajectories in model-based policy optimization to improve the sample efficiency of the learning algorithm. We evaluate our approach across various tasks within simulated environments. Our method demonstrates superior sample efficiency in these tasks compared to model-free and model-based baseline methods.
Egor Cherepanov, Nikita Kachaev, Alexey K. Kovalev et al.
Memory is crucial for enabling agents to tackle complex tasks with temporal and spatial dependencies. While many reinforcement learning (RL) algorithms incorporate memory, the field lacks a universal benchmark to assess an agent's memory capabilities across diverse scenarios. This gap is particularly evident in tabletop robotic manipulation, where memory is essential for solving tasks with partial observability and ensuring robust performance, yet no standardized benchmarks exist. To address this, we introduce MIKASA (Memory-Intensive Skills Assessment Suite for Agents), a comprehensive benchmark for memory RL, with three key contributions: (1) we propose a comprehensive classification framework for memory-intensive RL tasks, (2) we collect MIKASA-Base -- a unified benchmark that enables systematic evaluation of memory-enhanced agents across diverse scenarios, and (3) we develop MIKASA-Robo (pip install mikasa-robo-suite) -- a novel benchmark of 32 carefully designed memory-intensive tasks that assess memory capabilities in tabletop robotic manipulation. Our work introduces a unified framework to advance memory RL research, enabling more robust systems for real-world use. MIKASA is available at https://tinyurl.com/membenchrobots.
Maxim A. Patratskiy, Alexey K. Kovalev, Aleksandr I. Panov
Vision-Language-Action models have demonstrated remarkable capabilities in predicting agent movements within virtual environments and real-world scenarios based on visual observations and textual instructions. Although recent research has focused on enhancing spatial and temporal understanding independently, this paper presents a novel approach that integrates both aspects through visual prompting. We introduce a method that projects visual traces of key points from observations onto depth maps, enabling models to capture both spatial and temporal information simultaneously. The experiments in SimplerEnv show that the mean number of tasks successfully solved increased for 4% compared to SpatialVLA and 19% compared to TraceVLA. Furthermore, we show that this enhancement can be achieved with minimal training data, making it particularly valuable for real-world applications where data collection is challenging. The project page is available at https://ampiromax.github.io/ST-VLA.
Egor Cherepanov, Nikita Kachaev, Artem Zholus et al.
The incorporation of memory into agents is essential for numerous tasks within the domain of Reinforcement Learning (RL). In particular, memory is paramount for tasks that require the utilization of past information, adaptation to novel environments, and improved sample efficiency. However, the term ``memory'' encompasses a wide range of concepts, which, coupled with the lack of a unified methodology for validating an agent's memory, leads to erroneous judgments about agents' memory capabilities and prevents objective comparison with other memory-enhanced agents. This paper aims to streamline the concept of memory in RL by providing practical precise definitions of agent memory types, such as long-term versus short-term memory and declarative versus procedural memory, inspired by cognitive science. Using these definitions, we categorize different classes of agent memory, propose a robust experimental methodology for evaluating the memory capabilities of RL agents, and standardize evaluations. Furthermore, we empirically demonstrate the importance of adhering to the proposed methodology when evaluating different types of agent memory by conducting experiments with different RL agents and what its violation leads to.
Nikita Kachaev, Daniil Zelezetsky, Egor Cherepanov et al.
Despite their effectiveness and popularity in offline or model-based reinforcement learning (RL), transformers remain underexplored in online model-free RL due to their sensitivity to training setups and model design decisions such as how to structure the policy and value networks, share components, or handle temporal information. In this paper, we show that transformers can be strong baselines for continuous control in online model-free RL. We investigate key design questions: how to condition inputs, share components between actor and critic, and slice sequential data for training. Our experiments reveal stable architectural and training strategies enabling competitive performance across fully and partially observable tasks, and in both vector- and image-based settings. These findings offer practical guidance for applying transformers in online RL.
Anton Andreychuk, Konstantin Yakovlev, Aleksandr Panov et al.
Multi-agent pathfinding (MAPF) is a common abstraction of multi-robot trajectory planning problems, where multiple homogeneous robots simultaneously move in the shared environment. While solving MAPF optimally has been proven to be NP-hard, scalable, and efficient, solvers are vital for real-world applications like logistics, search-and-rescue, etc. To this end, decentralized suboptimal MAPF solvers that leverage machine learning have come on stage. Building on the success of the recently introduced MAPF-GPT, a pure imitation learning solver, we introduce MAPF-GPT-DDG. This novel approach effectively fine-tunes the pre-trained MAPF model using centralized expert data. Leveraging a novel delta-data generation mechanism, MAPF-GPT-DDG accelerates training while significantly improving performance at test time. Our experiments demonstrate that MAPF-GPT-DDG surpasses all existing learning-based MAPF solvers, including the original MAPF-GPT, regarding solution quality across many testing scenarios. Remarkably, it can work with MAPF instances involving up to 1 million agents in a single environment, setting a new milestone for scalability in MAPF domains.
Zoya Volovikova, Gregory Gorbov, Petr Kuderov et al.
Following instructions in real-world conditions requires the ability to adapt to the world's volatility and entanglement: the environment is dynamic and unpredictable, instructions can be linguistically complex with diverse vocabulary, and the number of possible goals an agent may encounter is vast. Despite extensive research in this area, most studies are conducted in static environments with simple instructions and a limited vocabulary, making it difficult to assess agent performance in more diverse and challenging settings. To address this gap, we introduce CrafText, a benchmark for evaluating instruction following in a multimodal environment with diverse instructions and dynamic interactions. CrafText includes 3,924 instructions with 3,423 unique words, covering Localization, Conditional, Building, and Achievement tasks. Additionally, we propose an evaluation protocol that measures an agent's ability to generalize to novel instruction formulations and dynamically evolving task configurations, providing a rigorous test of both linguistic understanding and adaptive decision-making.
Alexandr Korchemnyi, Alexey K. Kovalev, Aleksandr I. Panov
The idea of disentangled representations is to reduce the data to a set of generative factors that produce it. Typically, such representations are vectors in latent space, where each coordinate corresponds to one of the generative factors. The object can then be modified by changing the value of a particular coordinate, but it is necessary to determine which coordinate corresponds to the desired generative factor -- a difficult task if the vector representation has a high dimension. In this article, we propose ArSyD (Architecture for Symbolic Disentanglement), which represents each generative factor as a vector of the same dimension as the resulting representation. In ArSyD, the object representation is obtained as a superposition of the generative factor vector representations. We call such a representation a \textit{symbolic disentangled representation}. We use the principles of Hyperdimensional Computing (also known as Vector Symbolic Architectures), where symbols are represented as hypervectors, allowing vector operations on them. Disentanglement is achieved by construction, no additional assumptions about the underlying distributions are made during training, and the model is only trained to reconstruct images in a weakly supervised manner. We study ArSyD on the dSprites and CLEVR datasets and provide a comprehensive analysis of the learned symbolic disentangled representations. We also propose new disentanglement metrics that allow comparison of methods using latent representations of different dimensions. ArSyD allows to edit the object properties in a controlled and interpretable way, and the dimensionality of the object property representation coincides with the dimensionality of the object representation itself.
Egor Cherepanov, Alexey K. Kovalev, Aleksandr I. Panov
Real-world robotic agents must act under partial observability and long horizons, where key cues may appear long before they affect decision making. However, most modern approaches rely solely on instantaneous information, without incorporating insights from the past. Standard recurrent or transformer models struggle with retaining and leveraging long-term dependencies: context windows truncate history, while naive memory extensions fail under scale and sparsity. We propose ELMUR (External Layer Memory with Update/Rewrite), a transformer architecture with structured external memory. Each layer maintains memory embeddings, interacts with them via bidirectional cross-attention, and updates them through an Least Recently Used (LRU) memory module using replacement or convex blending. ELMUR extends effective horizons up to 100,000 times beyond the attention window and achieves a 100% success rate on a synthetic T-Maze task with corridors up to one million steps. In POPGym, it outperforms baselines on more than half of the tasks. On MIKASA-Robo sparse-reward manipulation tasks with visual observations, it nearly doubles the performance of strong baselines. These results demonstrate that structured, layer-local external memory offers a simple and scalable approach to decision making under partial observability.
Daniil Zelezetsky, Alexey K. Kovalev, Aleksandr I. Panov
The appearance of transformer-based models in Reinforcement Learning (RL) has expanded the horizons of possibilities in robotics tasks, but it has simultaneously brought a wide range of challenges during its implementation, especially in model-free online RL. Some of the existing learning algorithms cannot be easily implemented with transformer-based models due to the instability of the latter. In this paper, we propose a method that uses the Accelerator policy as a transformer's trainer. The Accelerator, a simpler and more stable model, interacts with the environment independently while simultaneously training the transformer through behavior cloning during the first stage of the proposed algorithm. In the second stage, the pretrained transformer starts to interact with the environment in a fully online setting. As a result, this model-free algorithm accelerates the transformer in terms of its performance and helps it to train online in a more stable and faster way. By conducting experiments on both state-based and image-based ManiSkill environments, as well as on MuJoCo tasks in MDP and POMDP settings, we show that applying our algorithm not only enables stable training of transformers but also reduces training time on image-based environments by up to a factor of two. Moreover, it decreases the required replay buffer size in off-policy methods to 10-20 thousand, which significantly lowers the overall computational demands.
Cognitive maps provide a powerful framework for understanding spatial and abstract reasoning in biological and artificial agents. While recent computational models link cognitive maps to hippocampal-entorhinal mechanisms, they often rely on global optimization rules (e.g., backpropagation) that lack biological plausibility. In this work, we propose a novel cognitive architecture for structuring episodic memories into cognitive maps using local, Hebbian-like learning rules, compatible with neural substrate constraints. Our model integrates the Successor Features framework with episodic memories, enabling incremental, online learning through agent-environment interaction. We demonstrate its efficacy in a partially observable grid-world, where the architecture autonomously organizes memories into structured representations without centralized optimization. This work bridges computational neuroscience and AI, offering a biologically grounded approach to cognitive map formation in artificial adaptive agents.
Nikita Kachaev, Andrei Spiridonov, Andrey Gorodetsky et al.
Benchmarks are crucial for evaluating progress in robotics and embodied AI. However, a significant gap exists between benchmarks designed for high-level language instruction following, which often assume perfect low-level execution, and those for low-level robot control, which rely on simple, one-step commands. This disconnect prevents a comprehensive evaluation of integrated systems where both task planning and physical execution are critical. To address this, we propose Kitchen-R, a novel benchmark that unifies the evaluation of task planning and low-level control within a simulated kitchen environment. Built as a digital twin using the Isaac Sim simulator and featuring more than 500 complex language instructions, Kitchen-R supports a mobile manipulator robot. We provide baseline methods for our benchmark, including a task-planning strategy based on a vision-language model and a low-level control policy based on diffusion policy. We also provide a trajectory collection system. Our benchmark offers a flexible framework for three evaluation modes: independent assessment of the planning module, independent assessment of the control policy, and, crucially, an integrated evaluation of the whole system. Kitchen-R bridges a key gap in embodied AI research, enabling more holistic and realistic benchmarking of language-guided robotic agents.
Artem Pshenitsyn, Aleksandr Panov, Alexey Skrynnik
Multi-agent reinforcement learning (MARL) is a powerful paradigm for solving cooperative and competitive decision-making problems. While many MARL benchmarks have been proposed, few combine continuous state and action spaces with challenging coordination and planning tasks. We introduce CAMAR, a new MARL benchmark designed explicitly for multi-agent pathfinding in environments with continuous actions. CAMAR supports cooperative and competitive interactions between agents and runs efficiently at up to 100,000 environment steps per second. We also propose a three-tier evaluation protocol to better track algorithmic progress and enable deeper analysis of performance. In addition, CAMAR allows the integration of classical planning methods such as RRT and RRT* into MARL pipelines. We use them as standalone baselines and combine RRT* with popular MARL algorithms to create hybrid approaches. We provide a suite of test scenarios and benchmarking tools to ensure reproducibility and fair comparison. Experiments show that CAMAR presents a challenging and realistic testbed for the MARL community.
We introduce Massively Multi-Task Model-Based Policy Optimization (M3PO), a scalable model-based reinforcement learning (MBRL) framework designed to address sample inefficiency in single-task settings and poor generalization in multi-task domains. Existing model-based approaches like DreamerV3 rely on pixel-level generative models that neglect control-centric representations, while model-free methods such as PPO suffer from high sample complexity and weak exploration. M3PO integrates an implicit world model, trained to predict task outcomes without observation reconstruction, with a hybrid exploration strategy that combines model-based planning and model-free uncertainty-driven bonuses. This eliminates the bias-variance trade-off in prior methods by using discrepancies between model-based and model-free value estimates to guide exploration, while maintaining stable policy updates through a trust-region optimizer. M3PO provides an efficient and robust alternative to existing model-based policy optimization approaches and achieves state-of-the-art performance across multiple benchmarks.
Artem Latyshev, Gregory Gorbov, Aleksandr I. Panov
Reinforcement Learning (RL) applications in real-world scenarios must prioritize safety and reliability, which impose strict constraints on agent behavior. Model-based RL leverages predictive world models for action planning and policy optimization, but inherent model inaccuracies can lead to catastrophic failures in safety-critical settings. We propose a novel model-based RL framework that jointly optimizes task performance and safety. To address world model errors, our method incorporates an adaptive mechanism that dynamically switches between model-based planning and direct policy execution. We resolve the objective mismatch problem of traditional model-based approaches using an implicit world model. Furthermore, our framework employs dynamic safety thresholds that adapt to the agent's evolving capabilities, consistently selecting actions that surpass safe policy suggestions in both performance and safety. Experiments demonstrate significant improvements over non-adaptive methods, showing that our approach optimizes safety and performance simultaneously rather than merely meeting minimum safety requirements. The proposed framework achieves robust performance on diverse safety-critical continuous control tasks, outperforming existing methods.
Model-based reinforcement learning (MBRL) allows solving complex tasks in a sample-efficient manner. However, no information is reused between the tasks. In this work, we propose a meta-learned addressing model called RAMa that provides training samples for the MBRL agent taken from continuously growing task-agnostic storage. The model is trained to maximize the expected agent's performance by selecting promising trajectories solving prior tasks from the storage. We show that such retrospective exploration can accelerate the learning process of the MBRL agent by better informing learned dynamics and prompting agent with exploratory trajectories. We test the performance of our approach on several domains from the DeepMind control suite, from Metaworld multitask benchmark, and from our bespoke environment implemented with a robotic NVIDIA Isaac simulator to test the ability of the model to act in a photorealistic, ray-traced environment.
Julia Kiseleva, Ziming Li, Mohammad Aliannejadi et al.
Human intelligence has the remarkable ability to adapt to new tasks and environments quickly. Starting from a very young age, humans acquire new skills and learn how to solve new tasks either by imitating the behavior of others or by following provided natural language instructions. To facilitate research in this direction, we propose IGLU: Interactive Grounded Language Understanding in a Collaborative Environment. The primary goal of the competition is to approach the problem of how to build interactive agents that learn to solve a task while provided with grounded natural language instructions in a collaborative environment. Understanding the complexity of the challenge, we split it into sub-tasks to make it feasible for participants. This research challenge is naturally related, but not limited, to two fields of study that are highly relevant to the NeurIPS community: Natural Language Understanding and Generation (NLU/G) and Reinforcement Learning (RL). Therefore, the suggested challenge can bring two communities together to approach one of the important challenges in AI. Another important aspect of the challenge is the dedication to perform a human-in-the-loop evaluation as a final evaluation for the agents developed by contestants.
This work studies object goal navigation task, which involves navigating to the closest object related to the given semantic category in unseen environments. Recent works have shown significant achievements both in the end-to-end Reinforcement Learning approach and modular systems, but need a big step forward to be robust and optimal. We propose a hierarchical method that incorporates standard task formulation and additional area knowledge as landmarks, with a way to extract these landmarks. In a hierarchy, a low level consists of separately trained algorithms to the most intuitive skills, and a high level decides which skill is needed at this moment. With all proposed solutions, we achieve a 0.75 success rate in a realistic Habitat simulator. After a small stage of additional model training in a reconstructed virtual area at a simulator, we successfully confirmed our results in a real-world case.