DexJoCo: A Benchmark and Toolkit for Task-Oriented Dexterous Manipulation on MuJoCo

2026-05-15

AI Summary

Existing dexterous manipulation benchmarks lack tasks that effectively capture the unique capabilities of dexterous hands—such as tool use, bimanual coordination, and long-horizon reasoning—and suffer from inconsistent evaluation protocols. This work proposes the first standardized task suite specifically designed to assess these distinctive abilities, comprising 11 functional tasks, along with an accompanying simulation toolkit and a dataset of 1.1K human demonstrations. Built on the MuJoCo platform, the framework integrates visual and dynamic domain randomization, multi-task training, and adaptive action heads to enable low-cost data collection and multidimensional robustness evaluation. Large-scale empirical analysis reveals critical limitations of current policies in complex dexterous manipulation and delineates key challenges for future research.

Abstract

Achieving human-level manipulation requires dexterous robotic hands capable of complex object interactions. Advancing such capabilities further demands standardized benchmarks for systematic evaluation. However, existing dexterous benchmarks lack tasks that reflect the unique manipulation capabilities of dexterous hands over parallel grippers, as well as comprehensive evaluation pipelines. In this paper, we present DexJoCo, a benchmark and toolkit for task-oriented dexterous manipulation, comprising 11 functionally grounded tasks that evaluate tool-use, bimanual coordination, long-horizon execution, and reasoning. We develop a low-cost data collection system and collect 1.1K trajectories across these tasks, with support for domain randomization to assess robustness. We benchmark modern models under diverse settings, including visual and dynamics randomization, multi-task training, and action-head adaptation. Through extensive empirical analysis, we identify several important insights and common limitations of current policies in dexterous manipulation, highlighting key challenges for future research in dexterous hand robot learning. Project page available at: https://dexjoco.github.io

Problem

• dexterous manipulation
• benchmark
• task-oriented
• robotic hands
• evaluation pipeline

Innovation

• dexterous manipulation
• benchmark
• MuJoCo
• domain randomization
• bimanual coordination