Author name cluster

Aleksandr Panov

Possible papers associated with this exact author name in Arrow. This page groups case-insensitive exact name matches and is not a full identity disambiguation profile.

10 papers

1 author row

AAAI Conference 2026 Conference Paper

CAMAR: Continuous Actions Multi-Agent Routing

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.

PDF Details DOI

AAMAS Conference 2026 Conference Paper

MARL-GPT: Foundation Model for Multi-Agent Reinforcement Learning

Maria Nesterova
Mikhail Kolosov
Anton Andreychuk
Egor Cherepanov
Oleg Bulichev
Alexey Kovalev
Konstantin Yakovlev
Aleksandr Panov

Recent advances in multi-agent reinforcement learning (MARL) have demonstrated success in numerous challenging domains and environments, but typically require specialized models for each task. In this work, we propose a coherent methodology that makes it possible for a single GPT-based model to learn and perform well across diverse MARL environments and tasks, including StarCraft Multi-Agent Challenge, Google Research Football and POGEMA. Our method, MARL-GPT, applies offline reinforcement learning to train at scale on the expert trajectories (400M for SMACv2, 100M for GRF, and 1B for POGEMA) combined with a single transformerbased observation encoder that requires no task-specific tuning. Experiments show that MARL-GPT1 achieves competitive performance compared to specialized baselines in all tested environments. Thus, ourfindingssuggestthatitis, indeed, possibletobuildamultitask transformer-based model for a wide variety of (significantly different) multi-agent problems paving the way to the fundamental MARL model (akin to ChatGPT, Llama, Mistral etc. in natural language modeling).

PDF

EAAI Journal 2026 Journal Article

Scene graph-driven reasoning for action planning of humanoid robot

Dmitry Yudin
Alexander Lazarev
Eva Bakaeva
Angelika Kochetkova
Alexey Kovalev
Aleksandr Panov

Recent progress in visual data analysis has significantly improved the ability of autonomous robots to understand their surroundings and perform complex tasks. This paper presents a modular method named Scene Graph-driven Reasoning for Action Planning (SG-RAPL) designed for high-level planning in dynamic environments, enabling adaptive control of humanoid robots. The method employs a three-dimensional (3D) scene graph to represent the environment and detect abnormal situations, while a large language model (LLM) translates natural-language commands into consecutive low-level actions. An original perceptual segmentation and tracking module constructs the scene graph in real time by providing instance segmentation, obstacle detection, and object pose estimation using data fusion with Augmented Reality University of Cordoba (ArUco) markers. The Planner module decomposes high-level tasks into subtasks such as navigation and object manipulation. Extensive experiments conducted on a manually collected and annotated dataset demonstrate that the proposed artificial intelligence-based approach efficiently plans complex actions in both virtual and real-world warehouse environments. The code and dataset of the proposed approach will be made publicly available.

Details DOI

AAAI Conference 2025 Conference Paper

MAPF-GPT: Imitation Learning for Multi-Agent Pathfinding at Scale

Anton Andreychuk
Konstantin Yakovlev
Aleksandr Panov
Alexey Skrynnik

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.

PDF Details DOI

EAAI Journal 2025 Journal Article

Polygon decomposition for obstacle representation in motion planning with Model Predictive Control

Aleksey Logunov
Muhammad Alhaddad
Konstantin Mironov
Konstantin Yakovlev
Aleksandr Panov

Model Predictive Control (MPC) is a powerful tool for planning the local trajectory of autonomous mobile robots. The paper considers a new algorithm for trajectory planning and obstacle avoidance based on the MPC technique known in Artificial Intelligence (AI) planning and robotics. We have proposed an original method for decomposing obstacles to form a potential field, which in turn is used as an additional component in MPC. Thus, we propose a new intelligent trajectory planning method that takes into account the special shape of obstacles, which in turn significantly improves the metrics of intelligent agent movement on the well-known Moving AI benchmark. The challenging aspect of MPC planning is collision avoidance on large and complicated grid maps. We propose the Polygon Segmentation for obtaining Artificial Potential Field (PolySAP). This local planner approximates the obstacles on the map with a set of polygons. We address the question of how to partition a map with polygons to make it fast and effective for a practical MPC planner. We propose a decomposition algorithm based on Straight Skeleton. Our algorithm returns a set of polygons, which are then convexified. Numerical experiments show that our method outperforms basic algorithms in performance and provides sufficient partition quality for effective planning. We propose an artificial potential function calculated for polygonal obstacles and added to the MPC objective for collision avoidance. We evaluate our approach on city map dataset and on a real robotic platform. Numerical experiments show that PolySAP allows for polygon decomposition that is five times faster than Interior Extensions. Our MPC solver provides a fast solution for the MPC task compared to the state-of-the-art MPC planners. Our planner ensured the safe motion of the real mobile robot through a narrow indoor environment. Our code is available at https: //github. com/alhaddad-m/PolySAP.

Details DOI

AAAI Conference 2024 Conference Paper

Decentralized Monte Carlo Tree Search for Partially Observable Multi-Agent Pathfinding

Alexey Skrynnik
Anton Andreychuk
Konstantin Yakovlev
Aleksandr Panov

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.

PDF Details DOI

EAAI Journal 2024 Journal Article

Hierarchical waste detection with weakly supervised segmentation in images from recycling plants

Dmitry Yudin
Nikita Zakharenko
Artem Smetanin
Roman Filonov
Margarita Kichik
Vladislav Kuznetsov
Dmitry Larichev
Evgeny Gudov

Reducing environmental pollution with household waste and emissions from the computing clusters is an urgent technological problem. In our work, we explore both of these aspects: the deep learning application to improve the efficiency of waste recognition on recycling plant’s conveyor, as well as carbon dioxide emission from the computing devices used in this process. To conduct research, we developed an unique open WaRP dataset that demonstrates the best diversity among similar industrial datasets and contains more than 10, 000 images with 28 different types of recyclable goods (bottles, glasses, card boards, cans, detergents, and canisters). Objects can overlap, be in poor lighting conditions, or significantly distorted. On the WaRP dataset, we study training and evaluation of cutting-edge deep neural networks for detection, classification and segmentation tasks. Additionally, we developed a hierarchical neural network approach called H-YC with weakly supervised waste segmentation. It provided a notable increase in the detection quality and made it possible to segment images, learning only having class labels, not their masks. Both the suggested hierarchical approach and the WaRP dataset have shown great industrial application potential.

Details DOI

AAAI Conference 2024 Conference Paper

Learn to Follow: Decentralized Lifelong Multi-Agent Pathfinding via Planning and Learning

Alexey Skrynnik
Anton Andreychuk
Maria Nesterova
Konstantin Yakovlev
Aleksandr Panov

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 possesses all the information on the agents' locations and goals is absent and the agents have to sequentially decide the actions on their own without having access to the 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. The code is available at https://github.com/AIRI-Institute/learn-to-follow.

PDF Details DOI

PRL Workshop 2023 Workshop Paper

Learn to Follow: Lifelong Multi-agent Pathfinding with Decentralized Replanning

Alexey Skrynnik
Anton Andreychuk
Maria Nesterova
Konstantin Yakovlev
Aleksandr Panov

Multi-agent Pathfinding (MAPF) problem generally asks to find a set of conflict-free paths for a set of agents confined to a graph. In conventional MAPF scenarios, the graph and the agents' start and goal locations are known in advance. Thus, a centralized planning algorithm can be utilized to generate a solution. In this work, we investigate the decentralized MAPF setting, in which the agents can not share the information and must independently navigate toward their goals without knowing the other agents' goals or paths. We focus on the 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 (RL) through policy optimization. Planning is utilized to maintain an individual path, while RL is employed to discover the collision avoidance policies that effectively guide an agent along the path. This decomposition and intrinsic motivation specific for multi-agent scenarios allows leveraging replanning with learnable policies. We evaluate our method on a wide range of setups and compare it to the state-of-the-art competitors (both learnable and search-based). The results show that our method consistently outperforms the competitors in challenging setups when the number of agents is high.

PDF Details

AAAI Conference 2023 Conference Paper

TransPath: Learning Heuristics for Grid-Based Pathfinding via Transformers

Daniil Kirilenko
Anton Andreychuk
Aleksandr Panov
Konstantin Yakovlev

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 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, learning the correction factor utilizes the knowledge of the instance-independent heuristic. 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 employed 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 4x while producing the solutions, whose costs exceed those 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. The project web-page is: https://airi-institute.github.io/TransPath/.

PDF Details DOI