The Trinity:Capability-Driven Recursive Self-Improvement for Multimodal Agents

Computer Science > Artificial Intelligence arXiv:2610.36077 (cs) [Submitted on 26 Oct 2026] Title: The Trinity: Capability-Driven Recursive Self-Improvement for Multimodal Agents. Authors: Weidong Tang, Ruiheng Wang, Jingxuan Zhang, Hexin Wang, Haoyuan Chen, Zhengfeiyang Zhang, Yixing Ma, Zichen Jiang, Ziyu Guo, Haodong Li, Zhengbo Jiao, Zhiqing Cui, Xinxiang Yin, Shuo Liang, Zhenglin Wan, Hongzhan Lin, Wangbo Zhao, Kaipeng Zhang, Pengfei Zhou, Odin Zhang, Yang You.

计算机科学 > 人工智能 arXiv:2610.36077 (cs) [提交于 2026 年 10 月 26 日] 标题：《三位一体：面向多模态智能体的能力驱动递归自我改进》。作者：Weidong Tang, Ruiheng Wang, Jingxuan Zhang, Hexin Wang, Haoyuan Chen, Zhengfeiyang Zhang, Yixing Ma, Zichen Jiang, Ziyu Guo, Haodong Li, Zhengbo Jiao, Zhiqing Cui, Xinxiang Yin, Shuo Liang, Zhenglin Wan, Hongzhan Lin, Wangbo Zhao, Kaipeng Zhang, Pengfei Zhou, Odin Zhang, Yang You。

Abstract: Multimodal intelligence can be viewed as human-guided co-evolution between worlds and agents through interaction. However, scaling this process faces two bottlenecks: constructing heterogeneous, verifiable environments is costly, and manual Harness iteration cannot keep pace with growing task diversity and complexity.

摘要：多模态智能可以被视为通过交互在世界与智能体之间进行的、由人类引导的协同进化。然而，扩展这一过程面临两个瓶颈：构建异构且可验证的环境成本高昂，且手动进行的 Harness 迭代无法跟上日益增长的任务多样性和复杂性。

A key insight is that tasks with different modalities and surface forms can share the same underlying capability requirements. We therefore introduce The Trinity, a recursive self-improvement framework coupling world generation with Harness evolution through three roles: Creator, Explorer, and Evolver.

一个关键的见解是，具有不同模态和表面形式的任务可以共享相同的底层能力需求。因此，我们引入了“三位一体”（The Trinity），这是一个通过创造者（Creator）、探索者（Explorer）和进化者（Evolver）这三个角色，将世界生成与 Harness 进化相结合的递归自我改进框架。

Drawing on verified execution trajectories, Evolver shifts the basis of task organization from surface categories to shared requirements, reorganizing the original task tree into a hierarchical capability tree. We introduce Capability-Driven Real2Sim2Real: for heterogeneous tasks at the same node, Creator abstracts their common requirements into simplified, verifiable simulated worlds for capability learning, while evaluation on original tasks measures transfer.

进化者利用经过验证的执行轨迹，将任务组织的基准从表面类别转向共享需求，将原始任务树重组为分层的能力树。我们引入了“能力驱动的 Real2Sim2Real”：对于同一节点下的异构任务，创造者将其共同需求抽象为简化的、可验证的模拟世界以进行能力学习，同时通过在原始任务上的评估来衡量迁移效果。

Along this capability tree, Explorer progressively evolves reusable capability modules through selective inheritance, forming a Harness tree whose validated updates accumulate across iterations within a fixed topology. The resulting Harnesses specialize in different capabilities, motivating a Capability-Aware Harness Router that uses public task information to select compatible module configurations from the tree.

沿着这棵能力树，探索者通过选择性继承逐步进化出可重用的能力模块，形成了一个 Harness 树，其经过验证的更新在固定拓扑结构内随迭代不断积累。由此产生的 Harness 专注于不同的能力，从而促使我们设计了一个“能力感知 Harness 路由器”，它利用公共任务信息从树中选择兼容的模块配置。

Our evolved Harnesses improve mean performance by 34 points across 28 training benchmarks and 26 points across 18 unseen benchmarks, including two real-world embodied tasks.

我们进化后的 Harness 在 28 个训练基准测试中平均性能提升了 34 个点，在 18 个未见过的基准测试中提升了 26 个点，其中包括两项现实世界的具身任务。