HumanoidArena: Benchmarking Egocentric Hierarchical Whole-body Learning

2026-06-16

AI Summary

This work addresses the tight coupling between task-level decision-making and dynamic control in humanoid robots performing whole-body interactive tasks in human-centric environments, which limits policy scalability and lacks systematic evaluation of the interface between high-level policies and motion trackers. To this end, the authors propose HumanoidArena, a simulation benchmark that formulates control as a hierarchical decision problem: a high-level policy, conditioned on first-person vision, proprioception, and language instructions, outputs compact whole-body actions executed by a general motion tracker (GMT). The benchmark emphasizes lower-limb coordination through seven interaction tasks where leg synergy is critical. For the first time, it highlights the structural role of leg coordination and evaluates the hierarchical system from two perspectives—perturbation generalization and GMT transferability. Experiments demonstrate that the architecture can accomplish diverse leg-intensive tasks, yet performance remains highly dependent on the specific GMT, revealing the core challenge of transferability in intermediate action representations.

Abstract

Humanoid robots promise whole-body interaction in human-centered environments, but scalable policy learning remains difficult because task-level decision-making and whole-body dynamic execution are tightly coupled. A practical solution is hierarchical control, where a high-level policy predicts intermediate whole-body actions and low-level general motion trackers (GMTs) execute them as stable humanoid motion. However, existing benchmarks rarely evaluate the policy-tracker interface itself, leaving open whether intermediate whole-body actions are executable, robust under task distribution shifts, and transferable across different GMT backends. We introduce HumanoidArena, a simulation-first benchmark for egocentric hierarchical whole-body learning. The benchmark formulates policy learning as a hierarchical decision making problem: a high-level policy converts egocentric vision, proprioception, and instructions into a compact whole-body action, which is subsequently executed by a low-level GMT. Instead of treating the legs as planar transport tools, HumanoidArena emphasizes interactions where lower-body coordination is structurally necessary in task completion. We therefore design 7 leg-critical HOI/HSI tasks in which success requires foot placement, balance maintenance, posture adjustment, and whole-body reorientation. To further diagnose the hierarchical system, we evaluate policies from two complementary perspectives: perturbation-conditioned generalization and GMT-conditioned transfer. Experiments show that hierarchical control enables learned policies to solve diverse leg-critical interactions, but performance is strongly tracker-conditioned and cross-GMT transfer remains fragile. These results position HumanoidArena as a benchmark for studying transferable intermediate action representations and scalable egocentric whole-body policy learning.

Problem

• hierarchical control
• humanoid robots
• whole-body learning
• general motion trackers
• transferability

Innovation

• HumanoidArena
• hierarchical whole-body control
• general motion trackers
• leg-critical tasks
• GMT-conditioned transfer