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.
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.
The quantified requirements of industrial robots enabled by EAI4I are analyzed and recent research progress is reviewed, covering core technologies for single- and multi-robot systems, dedicated hardware platforms, high-fidelity simulators, task-specific datasets, representative industrial application scenarios, and critical deployment challenges.
Hai-Bin Yu, Chunhe Song, Yinlong Zhang et al.· National Science Review· 0 citations
Custom-built quadrupedal platforms offer an accessible path for robotics research, yet researchers frequently encounter a “framework gap” when porting complex open-source control software to custom hardware. This paper documents the integration and validation process of Scotty, a custom quadrupedal robot, evaluating two distinct open-source control frameworks: the torque-based MIT Mini Cheetah and the position-based CHAMP architectures. Due to minimal documentation, hardware-dependent complexities, and a tightly coupled architecture, it was difficult to achieve a satisfactory result using the MIT framework within our system’s scope. Conversely, CHAMP’s structured integration documentation enabled the deployment of a locomotion pipeline validated in Gazebo simulation alongside a functional hardware interface middle layer. To overcome CHAMP’s lack of native operational state management, we developed a custom state-based controller with a web-based GUI that safely orchestrates transitions across Idle, Ready, Down, Stand, and Walk configurations. While walking was successfully validated in the simulation environment to verify the control software pipeline, physical hardware evaluation was restricted to individual joint control, localized leg movements, and GUI-based parameter tuning. Full closed-loop hardware locomotion was not achieved, as the extensive tuning of gait parameters and controller gains under full system weight was bounded by project constraints. The integration logs, practical hurdles, and architectural lessons documented in this work are shared openly to provide a clear, transferable roadmap for future developers of robotic systems.
Vishnudev Kurumbaparambil, Subashkumar Rajanayagam, Stefan Twieg· Italian National Conference...· 0 citations
High-precision insertion remains a fundamental challenge in robotic manipulation due to the strict alignment requirements and contact-rich interactions involved. Although peg-in-hole tasks are widely used for evaluation, existing bench- marks often rely on fixed task configurations, limiting their ability to assess robustness and generalization across different insertion scenarios. This paper introduces a reconfigurable peg-in-hole benchmark designed to evaluate task generalization in high-precision insertion. The benchmark consists of a set of fully 3D-printable modular components, including multiple peg geometries, tolerance levels, and configurable base structures that can be combined to generate a large variety of insertion and assembly tasks. By varying object layouts, orientations, and task structures while maintaining controlled physical conditions, the benchmark enables systematic evaluation of adaptation to unseen scenarios. To support reproducibility, we additionally provide a scenario generation tool capable of producing standardized task configurations and machine-readable task descriptions. The scenario generation tool and the STL files of the benchmark pieces are available through the project repository: https://github.com/aistairc/peg-in-bench.
Y. Delgado, José G. Buenaventura-Carreón, Floris Erich et al.· 0 citations
The transition toward high-mix low-volume manufacturing demands flexibility in robotic manipulation. However, conventional vision systems remain a bottleneck, requiring extensive data collection and model retraining whenever a new object is introduced to the production line. To overcome this rigidity, we present xperception, a zero-shot 6D pose estimation technology that eliminates the need for object-specific fine-tuning and laborious data annotation. By directly utilizing typical CAD models and integrating the rich semantic features of foundation models (e.g. DINOv2, GeDi), xperception achieves millimeter-accurate 6D pose estimation. xperception showed robustness against severe occlusions in industrial tasks like bin picking and is engineered for deployment on industrial edge hardware, such as NVIDIA Jetson Thor. Validated at a TRL of 6, the core methodology behind xperception is based on the FreeZe algorithm, which won the international BOP Challenge 2024, paving the way for scalable, plug-and-play robotic automation in unstructured high-mix low-volume manufacturing industries.
M. Bortolon, Andrea Caraffa, Alice Fasoli et al.· arXiv.org· 0 citations
This hands-on course presents a workflow for configuring robot simulations using OpenUSD and PhysX in NVIDIA Isaac Sim, with an emphasis on physics tuning, stability, and asset authoring best practices, and situates OpenUSD’s composability within the broader physical AI ecosystem.
Ji Yuan Feng, Alexandra Kissel· Proceedings of the Special I...· 0 citations
Robustness under test-time shift remains a major bottleneck for learning-based robot manipulation in cameraguided industrial workcells. Policies trained only in nominal simulation often lose performance when lighting, background appearance, table appearance, or object appearance changes at deployment. This paper studies this problem in a fixedcamera Franka Panda workcell using RLBench and Colosseum. We compare three training regimes under the same front, left-shoulder, and right-shoulder camera setup: a standard Robotic View Transformer (RVT) baseline, RVT with expanded targeted domain randomization (TDR), and RVT with expanded targeted domain randomization plus adaptive hard-example sampling (AHS). Expanded TDR broadens the training distribution over deployment-relevant Colosseum perturbation identifiers while keeping the backbone and sensor configuration fixed. AHS then updates replay weights online so that persistently low-scoring perturbation identifiers receive more attention without collapsing training onto only a few difficult cases. Across the four-task benchmark, expanded TDR improves the overall mean score from 32.4 to 50.0, and AHS further improves it to 58.0. The worst-ID score increases from 0.0 to 28.0, the Tail-2 score increases from 9.8 to 42.0, and the summary robustness-curve area increases from 14.6 to 44.0. The largest gains occur on the harder Close Box and Insert Onto Square Peg tasks. These results show that broadening the coverage of perturbation and adapting the replay toward the hard-tail is an effective and practical recipe for robust industrial manipulation.
Abdullah Matar, Yara Altamimi, Y. M. Alrawashdeh· IEEE Jordan Conference on Ap...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
