Industrial Dexterity Benchmark: A Hardware-Software Benchmarking Platform for Industrial Dexterous Manipulation
2026-07-15 • Robotics
Robotics
AI summaryⓘ
The authors address the challenge of complex robot tasks in factories, like managing cables and assembling parts, which still need humans despite many advances. They created new test boards mimicking real industrial tasks and developed a learning system called DAG-ROS that teaches robots using multiple types of sensory data, including images and force measurements. Their best robot system, AG-iDP3, learned to handle cables better by using multiple camera views, achieving a 78% success rate compared to 36% with a simpler single-camera setup. This work shows that robots can learn these tricky tasks more effectively than traditional programming by watching and imitating demonstrations.
dexterous manipulationimitation learningmultimodal datadiffusion policyRGB imagespoint cloudsrobotic benchmarksteleoperationindustrial automationrobot vision
Authors
Honglu He, Jacob Laufer, Zhiwu Zheng, David Elkan-gonzalez, Raman Goyal, Xinyi Li, Su Lu, Mishek Musa, Berke Saat, Nicolas Tan, Colm Prendergast
Abstract
Dexterous manipulation remains a critical bottleneck in industrial automation; tasks such as cable routing, connector insertion, and precision assembly still rely heavily on manual labor despite decades of robotics research. This work presents a progression from classical, modular robotics pipelines toward an end-to-end multimodal imitation-learning framework for industrial dexterous manipulation. As a part of this work, we introduce three key contributions: a set of Industrial Dexterity Benchmark (IDB) boards aimed to mimic datacenter cable management, automotive cable harnesses, and gearbox assembly tasks; a scalable imitation learning framework (DAG-ROS); and a multimodal diffusion-based policy framework (AG-iDP3) that creates models fusing RGB images, point clouds, joint positions, and wrist-frame wrench data. Focusing on the datacenter cable manipulation board, we evaluate the performance of a task involving cleaning a single cable over variations of an end-to-end AI policy using 48 trials per configuration. The best performing configuration, a multimodal expansion Diffusion Policy (DP), includes a multi-view RGB image source passed through an R3M encoder and reaches a 78% grasp and insert combined task success rate. This performance marks a significant improvement over the 36% observed from the single-camera RGB DP baseline. Each of the tested configurations requires only approximately 100 teleoperated demonstrations per task phase. These results indicate that the correct learned policy can outperform classical vision and control robotic methods in robustness, generalization, and deployment efficiency, justifying a shift toward scalable robotic automation for high up-time industrial environments.