OMRON SINIC X Corporation (HQ: Bunkyo-ku, Tokyo; President and CEO: Masaki Suwa; hereinafter “OSX”) will present the latest research findings at IEEE/RSJ International Conference on Intelligent Robots and Systems (hereinafter “IROS 2026”).

Along with ICRA¹⁾, IROS is one of the largest and most influential top-tier conferences in the field of robotics. In 2026, 1,585 papers (approximately 36%) were accepted out of 4,947 submissions, and the conference will be held from September 27 to October 1, 2026, in Pittsburgh, United States (local time).
^{1) IEEE International Conference on Robotics and Automation The following five research papers on OSX have been accepted.}

Research Papers Accepted at IROS 2026

■ Grounded Vision-Language Interpreter for Long-Horizon Bimanual Task and Motion Planning

Jeremy Siburian (The University of Tokyo / OSX Intern), Keisuke Shirai (AIST2) / OSX Intern), Cristian Camilo Beltran-Hernandez (OSX), Masashi Hamaya (OSX), Michael Görner (Constructor University Bremen), Atsushi Hashimoto (OSX)
2) National Institute of Advanced Industrial Science and Technology

While recent advances in vision-language models have accelerated language-guided robot planning, their blackbox nature lacks the safety guarantees and interpretability crucial for real-world deployment. Conversely, classical symbolic planners offer rigorous safety verification but require significant expert knowledge for setup. Moreover, most existing methods are limited to single-arm pick-and-place tasks, leaving bimanual manipulation largely underexplored, despite its tightly interdependent subtasks and the need to explicitly manage interarm collisions.
To bridge this gap, this paper proposes ViLaInTAMP, a hybrid planning framework for enabling verifiable, interpretable, and autonomous bimanual robot behaviors. ViLaIn-TAMP comprises three main components: (1) a VisionLanguage Interpreter (ViLaIn) adapted from a prior work that converts multimodal inputs into structured PDDL problem specifications, (2) an integrated Task and Motion Planning (TAMP) system that grounds these specifications in actionable trajectory sequences through symbolic and geometric constraint reasoning, explicitly verifying feasibility before execution, and (3) a corrective planning (CP) module which receives structured motion failure feedback and feeds them as constraints back to ViLaIn to refine the specification.
We design challenging bimanual manipulation tasks in a cooking domain to evaluate our framework, where experimental results show that ViLaInTAMP outperforms a VLM-as-a-planner baseline by 18% in mean success rate, with the CP module boosting it further by 32%. We further validate ViLaIn-TAMP on a physical dual-arm robotic system.

<Notice from the AIST>
https://eart.airc.aist.go.jp/news/iros2026-presentations-ws/

■ PHASE: Compliance-Enabled Tactile Phase Retrieval for Few-Shot Insertion Learning

Jeremy Siburian (The University of Tokyo / OSX Intern), Cristian Camilo Beltran-Hernandez (OSX), Tatsuya Matsushima (The University of Tokyo), Yusuke Iwasawa (The University of Tokyo), Masashi Hamaya (OSX), Mai Nishimura (OSX)

Contact-rich assembly tasks such as peg-in-hole insertion　remain difficult to learn from limited demonstrations. While retrieval-augmented imitation learning, which augments target demonstrations with relevant prior data, offers a promising direction, its applicability to contact-rich manipulation remains largely unexplored. Contact-rich insertion unfolds over multiple phases from search to insert, and retrieving phase-specific experience from prior data in principled ways remains an open question.
Our key insight is that a compliant wrist enables the robot to sustain contact throughout execution,　producing rich tactile and force signals that naturally reveal the phase structure of insertion and inform what should be retrieved. Based on this insight, we present PHASE (PHase-Aware Segmentation and REtrieval), a framework for compliance-enabled tactile phase retrieval that integrates multimodal contact-aware representation learning, variable-length phase segmentation from tactile signals, and phase-consistent retrieval for policy learning.
We evaluate PHASE on real-world peg-in-hole insertion across five peg geometries, comparing against　retrieval strategies drawn from state-of-the-art methods under a shared policy architecture. PHASE outperforms all baselines that do not account for phase structure by +13% or more on seen and unseen geometries, and by +30% under state distribution shift. These results demonstrate that aligning retrieval with interaction-defined contact phases substantially improves robustness in few-shot insertion learning.

■ SHAFT: A Slack-Compensating, Helical-Buckling-Attenuating Flexible-Shaft Transmission for Lightweight Multi-DoF Manipulation

Tomoya Takahashi (OSX), Moses Gladson Selvamuthu (Yamagata University), Riichiro Tadakuma (Tohoku University), Kazutoshi Tanaka (OSX)

To ensure safe robot operation in human environments, we propose SHAFT: a Slack-compensating, Helical-buckling-Attenuating Flexible-shaft Transmission mechanism for ultra-lightweight manipulators.
While proximal actuation via flexible shafts minimizes arm inertia and friction losses, large joint bending generates path slack, causing helical buckling and significant angular transmission errors under torque loading. The proposed SHAFT mechanism incorporates a passive, spring-based linear tensioner at the base to dynamically eliminate slack without increasing the arm’s moving mass. In a path with four 90° bends, SHAFT improved transmission efficiency by approximately 30% and reduced angular error by 65%.
We successfully validated the system by fabricating a 6-DoF arm weighing just 280 g, which carried a 500 g payload, presenting a highly efficient remote actuation system for collaborative robots. 

■ EgoReveal: Learning Active Vision and Object Manipulation under Partial Occlusions from Egocentric Demonstrations

Hiromu Kurihara (Keio University), Kazutoshi Tanaka (OSX), Masashi Hatano (Keio University), Tadashi Kozuno (OSX), Hideo Saito (Keio University), Mariko Isogawa (Keio University)

Learning from Demonstrations (LfD) using egocentric human videos from smart glasses enables scalable data collection for robot manipulation, but faces challenges under partial occlusions where visual cues are obscured. Conventional egocentric methods assume a fixed camera viewpoint at inference time, frequently leading to task failure when target objects are hidden.
To overcome this limitation, we propose EgoReveal, a framework that co-learns occlusion-aware viewpoint-repositioning and object manipulation directly from natural, robot-free human demonstrations. EgoReveal integrates two imitation policies: an Active Vision policy implemented via Action Chunking with Transformers (ACT) to actively adjust the camera-arm pose for visibility, and a Manipulation policy (EgoZero) to complete the pick task using 3D keypoints tracked after viewpoint optimization. A visibility-based Occlusion Resolution Assessment (ORA) module governs the dynamic switching between these two phases.
In tabletop experiments with complex and unseen occlusions, EgoReveal reliably moved the camera to eliminate occlusions and achieved higher task success rates than fixed-camera baselines, demonstrating robust generalization to everyday cluttered environments. 

■ Symmetry-Breaking in Multi-Agent Navigation: Winding Number-Aware MPC with a Learned Topological Strategy

Tomoki Nakao (Kyoto University), Kazumi Kasaura (OSX), Tadashi Kozuno (OSX)

In distributed multi-agent navigation without explicit communication, agents can fall into symmetry-induced deadlocks because each agent must autonomously decide how to pass others. To address this problem, we propose WNumMPC, a hierarchical navigation method that quantifies cooperative symmetry-breaking strategies via a topological invariant, the winding number, and learns such strategies through reinforcement learning.
The learning-based Planner outputs continuous-valued signed target winding numbers and dynamic importance weights to prioritize critical interactions in dense crossings. Then, the model-based Controller generates collision-free and efficient motions based on the strategy and weights provided by the Planner.
Simulation and real-world robot experiments indicate that WNumMPC effectively avoids deadlocks and collisions and achieves better performance than the baselines, particularly in dense and symmetry-prone scenarios. These experiments also suggest that explicitly leveraging winding numbers yields robust sim-to-real transfer with minimal performance degradation.

※Author information is current as of the date of writing or submission. Please be advised that the information may become outdated after that point.

For any inquiries about OSX, please contact us here.