Towards General Computer Control with Hierarchical Agents and Multi-Level Action Spaces

• URL: http://arxiv.org/abs/2509.18230v1
• Date: Mon, 22 Sep 2025 13:14:47 GMT
• Title: Towards General Computer Control with Hierarchical Agents and Multi-Level Action Spaces
• Authors: Zihan Dong, Xinyu Fan, Zixiang Tang, Yunqing Li,
• Abstract summary: We introduce a lightweight hierarchical reinforcement learning framework, ComputerAgent, that formulates OS control as a two-level option process.<n>On a suite of 135 real-world desktop tasks, ComputerAgent attains 92.1% success on simple tasks and 58.8% on hard tasks.<n>Results demonstrate that hierarchical RL offers a practical, scalable alternative to monolithic MLLM-based automation for computer control.
• Score: 5.258138614911196
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Controlling desktop applications via software remains a fundamental yet under-served problem. Existing multi-modal large language models (MLLMs) ingest screenshots and task instructions to generate keystrokes and mouse events, but they suffer from prohibitive inference latency, poor sample efficiency on long-horizon sparse-reward tasks, and infeasible on-device deployment. We introduce a lightweight hierarchical reinforcement learning framework, ComputerAgent, that formulates OS control as a two-level option process (manager and subpolicy), employs a triple-modal state encoder (screenshot, task ID, numeric state) to handle visual and contextual diversity, integrates meta-actions with an early-stop mechanism to reduce wasted interactions, and uses a compact vision backbone plus small policy networks for on-device inference (15M parameters). On a suite of 135 real-world desktop tasks, ComputerAgent attains 92.1% success on simple tasks (<8 steps) and 58.8% on hard tasks (>=8 steps), matching or exceeding 200B-parameter MLLM baselines on simple scenarios while reducing model size by over four orders of magnitude and halving inference time. These results demonstrate that hierarchical RL offers a practical, scalable alternative to monolithic MLLM-based automation for computer control.

Related papers

• AgentCgroup: Understanding and Controlling OS Resources of AI Agents [2.8139711959925244]
  AI agents are increasingly deployed in multi-tenant cloud environments, where they execute diverse tool calls within sandboxed containers.<n>We present a systematic characterization of OS-level resource dynamics in sandboxed AI coding agents.<n>Preliminary evaluation demonstrates improved multi-tenant isolation and reduced resource waste.
  arXiv  Detail & Related papers  (2026-02-10T02:37:42Z)
• OS-Symphony: A Holistic Framework for Robust and Generalist Computer-Using Agent [58.07447442040785]
  We introduce OS-Symphony, a holistic framework that comprises an Orchestrator coordinating two key innovations for robust automation.<n>Results demonstrate that OS-Symphony delivers substantial performance gains across varying model scales.
  arXiv  Detail & Related papers  (2026-01-12T17:55:51Z)
• UltraCUA: A Foundation Model for Computer Use Agents with Hybrid Action [77.63125913907771]
  We present UltraCUA, a foundation model that bridges the gap between GUI primitives and high-level programmatic tool calls.<n>Experiments with our 7B and 32B models demonstrate substantial improvements over state-of-the-art agents.
  arXiv  Detail & Related papers  (2025-10-20T17:48:26Z)
• Multi-Agent Tool-Integrated Policy Optimization [67.12841355267678]
  Large language models (LLMs) increasingly rely on multi-turn tool-integrated planning for knowledge-intensive and complex reasoning tasks.<n>Existing implementations typically rely on a single agent, but they suffer from limited context length and noisy tool responses.<n>No existing methods support effective reinforcement learning post-training of tool-integrated multi-agent frameworks.
  arXiv  Detail & Related papers  (2025-10-06T10:44:04Z)
• Scaling Synthetic Task Generation for Agents via Exploration [67.70129766322985]
  Post-Training Multimodal Large Language Models (MLLMs) to build interactive agents holds promise across domains such as computer-use, web navigation, and robotics.<n>Existing approaches for task generation rely heavily on human annotation or prompting MLLM with limited downstream environment information.<n>We present AutoPlay, a scalable pipeline for task generation that explicitly explores interactive environments to synthesize environment-grounded tasks.
  arXiv  Detail & Related papers  (2025-09-29T17:00:02Z)
• UFO2: The Desktop AgentOS [60.317812905300336]
  UFO2 is a multiagent AgentOS for Windows desktops that elevates into practical, system-level automation.<n>We evaluate UFO2 across over 20 real-world Windows applications, demonstrating substantial improvements in robustness and execution accuracy over prior CUAs.<n>Our results show that deep OS integration unlocks a scalable path toward reliable, user-aligned desktop automation.
  arXiv  Detail & Related papers  (2025-04-20T13:04:43Z)
• PC-Agent: A Hierarchical Multi-Agent Collaboration Framework for Complex Task Automation on PC [98.82146219495792]
  In this paper, we propose a hierarchical agent framework named PC-Agent.<n>From the perception perspective, we devise an Active Perception Module (APM) to overcome the inadequate abilities of current MLLMs in perceiving screenshot content.<n>From the decision-making perspective, to handle complex user instructions and interdependent subtasks more effectively, we propose a hierarchical multi-agent collaboration architecture.
  arXiv  Detail & Related papers  (2025-02-20T05:41:55Z)
• DeeR-VLA: Dynamic Inference of Multimodal Large Language Models for Efficient Robot Execution [114.61347672265076]
  Development of MLLMs for real-world robots is challenging due to the typically limited computation and memory capacities available on robotic platforms. We propose a Dynamic Early-Exit Framework for Robotic Vision-Language-Action Model (DeeR) that automatically adjusts the size of the activated MLLM. DeeR demonstrates significant reductions in computational costs of LLM by 5.2-6.5x and GPU memory of LLM by 2-6x without compromising performance.
  arXiv  Detail & Related papers  (2024-11-04T18:26:08Z)
• CAMPHOR: Collaborative Agents for Multi-input Planning and High-Order Reasoning On Device [2.4100803794273005]
  We introduce an on-device Small Language Models (SLMs) framework designed to handle multiple user inputs and reason over personal context locally. CAMPHOR employs a hierarchical architecture where a high-order reasoning agent decomposes complex tasks and coordinates expert agents responsible for personal context retrieval, tool interaction, and dynamic plan generation. By implementing parameter sharing across agents and leveraging prompt compression, we significantly reduce model size, latency, and memory usage.
  arXiv  Detail & Related papers  (2024-10-12T07:28:10Z)
• Sparse Diffusion Policy: A Sparse, Reusable, and Flexible Policy for Robot Learning [61.294110816231886]
  We introduce a sparse, reusable, and flexible policy, Sparse Diffusion Policy (SDP) SDP selectively activates experts and skills, enabling efficient and task-specific learning without retraining the entire model. Demos and codes can be found in https://forrest-110.io/sparse_diffusion_policy/.
  arXiv  Detail & Related papers  (2024-07-01T17:59:56Z)
• CAAP: Context-Aware Action Planning Prompting to Solve Computer Tasks with Front-End UI Only [21.054681757006385]
  We propose an agent that perceives its environment solely through screenshot images.<n>By leveraging the reasoning capability of the Large Language Models, we eliminate the need for large-scale human demonstration data.<n>Agent achieves an average success rate of 94.5% on MiniWoB++ and an average task score of 62.3 on WebShop.
  arXiv  Detail & Related papers  (2024-06-11T05:21:20Z)
• Synapse: Trajectory-as-Exemplar Prompting with Memory for Computer Control [23.115574119132507]
  Building agents with large language models for computer control is a burgeoning research area, where the agent receives computer states and performs actions to complete tasks. Previous computer agents have demonstrated the benefits of in-context learning (ICL), but their performance is hindered by several issues. We introduce Synapse, a computer agent featuring three key components: i) state abstraction, which filters out task-irrelevant information from raw states, allowing more exemplars within the limited context, ii) trajectory-as-exemplar prompting, which prompts the LLM with complete trajectories of the abstracted states and actions to improve multi-step decision-
  arXiv  Detail & Related papers  (2023-06-13T15:49:41Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.