Verified Events
DateTypeCompanyHeadlineSegment
Dyna Robotics details the million-hour infrastructure behind Dyna-2
- A new episode format with topic-group chunking reduced storage by about 68% and made sample reads approximately 2.9 times faster.
- An Airflow-based ingestion architecture increased throughput from 14,000 to 440,000 episode-hours per week, cutting million-hour processing from roughly 16 months to under three weeks.
- Warehouse-backed curation and memory-mapped tables reduced time to first training batch from about 48 hours to under one minute across a 43-million-episode dataset.
Human-Robot Shared Control for Humanized End-Effector Teleoperation
- The study proposes a real-time shared-control framework for human-like end-effector trajectories during teleoperation.
- The framework preserves operator intent while applying movement profiles based on the two-thirds power law.
- The authors validate the approach in real-world experiments on a 6-DoF Dobot CR10 robotic arm.
UR20 pose-accuracy paper evaluates Universal Robots arm under ISO 9283 testing
- The paper reports experimental pose-accuracy testing of the UR20 robot from Universal Robots.
- The experiments follow ISO 9283 guidelines and use an OptiTrack vision system to record robot positions.
VASO paper evaluates formally verifiable physical-AI skills on Clearpath Jackal
- The framework couples a formal interface for model checking with a planner-facing interface for generating executable physical behaviors.
- When a generated skill or plan violates temporal specifications, VASO turns the counterexample trace into textual feedback that updates the skill while keeping foundation-model weights frozen.
- On Clearpath Jackal and PX4 quadcopter tasks, the paper reports 97.2% formal-specification compliance using fewer than 100 optimization samples.
MoDex evaluates sequential multi-object dexterous grasping on Allegro Hand hardware
- MoDex introduces a diffusion-policy approach for sequentially grasping multiple objects with a single dexterous hand.
- The authors evaluate the method in simulation on a Franka Emika Panda robot with an Allegro Hand and run 90 real-world grasping sequences on matching hardware.
External force-torque sensor feedback on the ABB GoFa collaborative robot
- The paper studies external force-torque sensor feedback on the ABB GoFa collaborative robot.
- The study’s named evaluation platform is GoFa CRB 15000 from ABB Robotics.
PhAIL: A Real-Robot VLA Benchmark and Distributional Methodology
- Launches a real-robot VLA benchmark on Franka FR3 with distributional metrics beyond binary success rates.
- Includes analysis code, paper sources, annotation pipeline, dataset metadata, and public dataset download instructions.
- Reports throughput-style comparisons across public VLA systems using time-to-success curves, bootstrap confidence intervals, and KS tests.
Video-model policy paper includes Allegro Hand dexterous cube reorientation benchmarks
- The paper studies turning video generative models into generalist robot policies across simulated and real robot settings.
- Its evaluation includes 16-DOF Allegro-hand dexterous cube reorientation benchmarks.
RAI Institute demonstrates AthenaZero vision-based robot juggling
- AthenaZero uses onboard visual feedback rather than an external motion-capture system.
- The multi-fingered system transitions between multiple juggling patterns.
- The work forms part of RAI Institute research in advanced learning for control.
When Agents Control Robots: A Zero Trust Policy Model for Agentic Cyber-Physical Systems
- Multi-agent systems powered by large foundation models (LFMs) are increasingly deployed to control industrial robots through natural language, creating deployments in which security failures produce physical consequences. We analyse this threat landscape through Cobot-Claw, a deployed four-agent system for UR3e robotic arm control, and identify five attack classes specific to agentic cyber-physical systems.
- Experiments use UR3e as the physical evaluation platform.
- The reported results provide an independent evaluation on UR3e hardware.
SafePBDS demonstrates safe geometric motion policies on a Franka Panda–Allegro platform
- SafePBDS proposes steerable geometric motion policies for dexterous manipulation with safety constraints.
- The paper reports hardware experiments on a 23-DOF Franka Emika Panda arm and Wonik Allegro 16-DOF dexterous hand platform.
- This is external product-use research and is product-bound to Allegro Hand.
Safety-constrained RL paper validates reachability-checked navigation on Clearpath Jackal
- The framework combines Conditional Value-at-Risk optimization on an off-policy TD3 backbone with neural reachability verification under bounded observation uncertainty.
- Across ten navigation scenarios and six baselines, the method reports a 98.3% success rate and the highest safety-verification rate while showing that average-cost rankings can miss tail-risk behavior.
- A physical Clearpath Jackal validation demonstrates successful sim-to-real transfer and confirms that reachability-based safety checks can expose risks hidden by average empirical cost.
SECOND-Grasp reports transfer results for Allegro Hand dexterous grasping
- SECOND-Grasp introduces semantic contact-guided dexterous grasping for physically stable and task-aware grasps.
- The paper reports transfer results on Allegro Hand alongside other robotic-hand embodiments.
- Allegro Hand serves as an evaluation platform for testing cross-embodiment dexterous-grasp transfer.
GA3T dataset uses Clearpath Husky for air-ground traversability research
- The GA3T paper presents a real-world multi-robot perception dataset collected with a Clearpath Husky UGV and an Autel EVO II UAV.
- The dataset includes 3D LiDAR, stereo camera, IMU, GPS, RGB, thermal and GPS streams across forest trails, rocks, mud, snow piles and grass.
- The work positions Husky as the ground platform for cross-view perception, terrain traversability and air-ground collaborative scene understanding.
Genesis AI unveils GENE-26.5 robotics foundation model and dexterous hand demo
- Genesis AI unveiled GENE-26.5, a robotics foundation model for dexterous physical manipulation.
- The company said the system combines a human-scale dexterous robotic hand with a data engine for skill transfer, training and evaluation.
- Genesis AI demonstrated the model across tasks including cooking eggs, lab pipetting, smoothie making, Rubik’s cube solving, wire harnessing and piano performance.
Cooptimizing Safety and Performance Using Safety Value-Constrained Model Predictive Control
- The paper augments MPC with a safety value terminal constraint and validates the method on a Flexiv Rizon 10s manipulator.
- The study’s named evaluation platform is Rizon Adaptive Robot Series from Flexiv Robotics Ltd..
Franka GELLO & GELLO Duo open-source teleoperation hardware
- Released STL hardware design files and ROS 2-compatible software resources for the single-arm GELLO and bimanual GELLO Duo teleoperation devices.
- Enables low-latency joint and gripper control plus reproducible demonstration collection workflows.
- Reduces hardware barriers for teleoperation, imitation learning, and bimanual manipulation data pipelines.
ROBOGATE: Adaptive Failure Discovery for Safe Robot Policy Deployment via Two-Stage Boundary-Focused Sampling
- The UR3e configuration uses a suction SurfaceGripper and 10,000 Latin-hypercube-sampled Isaac Sim experiments.
- UR3e achieved 9.6% success in the shared workspace; the study attributes the low rate primarily to its 0.50-metre reach placing tasks at the reachable boundary.
- All UR3e failures were grasp misses; the reported evaluation is simulation-only and does not establish physical deployment performance.
RoboCritics: Enabling Reliable End-to-End LLM Robot Programming through Expert-Informed Critics
- End-user robot programming grants users the flexibility to re-task robots in situ, yet it remains challenging for novices due to the need for specialized robotics knowledge. Large Language Models (LLMs) hold the potential to lower the barrier to robot programming by enabling task specification through natural language.
- Experiments use UR3e as the physical evaluation platform.
- The reported results provide an independent evaluation on UR3e hardware.
DexFormer studies cross-embodied dexterous manipulation transfer to Allegro Hand
- DexFormer introduces a history-conditioned transformer for cross-embodied dexterous manipulation.
- The paper evaluates zero-shot transfer across hand embodiments including LEAP Hand, Allegro Hand and Rapid Hand.
DISF: Disentangled Iterative Surface Fitting for Contact-stable Grasp Planning with Grasp Pose Alignment to the Object Center of Mass
- In this work, we address the limitation of surface fitting-based grasp planning algorithm, which primarily focuses on geometric alignment between the gripper and object surface while overlooking the stability of contact point distribution, often resulting in unstable grasps due to inadequate contact configurations. To overcome this limitation, the study proposes a novel surface fitting algorithm that integrates contact stability while preserving geometric compatibility.
- UR3e is used in the study's data-collection or experimental dataset workflow.
YOPO-Nav paper benchmarks visual navigation on Clearpath Jackal trajectories
- YOPO-Nav uses one-pass exploration video as visual reference and avoids relying on a metric map by aligning current observations with a compact 3D Gaussian Splatting representation.
- Its hierarchy combines visual place recognition for coarse localization with local 3DGS models to refine poses and generate navigation actions.
- The YOPO-Campus dataset contains four hours of egocentric video and controller inputs from more than six kilometres of human-teleoperated trajectories; benchmarks use a physical Clearpath Jackal.
ContactRL paper validates tactile insertion learning on Universal Robots UR3e
- ContactRL presents a reinforcement-learning approach for contact-rich insertion tasks using tactile perception.
- The real-world validation uses a Universal Robots UR3e arm in the robotic insertion setup.
Heterogeneous robot collaboration paper runs real-world missions with Clearpath Jackal
- SPINE-HT uses a three-stage process in which an LLM generates grounded subtasks, validates feasibility, assigns work based on robot capabilities and refines the plan using online feedback.
- In simulation with closed-loop perception and control, the framework achieves nearly twice the success rate of prior LLM-enabled heterogeneous-teaming approaches.
- Real-world missions with a Clearpath Jackal, Clearpath Husky, Boston Dynamics Spot and a high-altitude UAV achieve an 87% success rate when tasks require capability reasoning and online refinement.
Development of a Linear Guide-Rail Testbed for Physically Emulating ISAM Operations
- In-Space Servicing, Assembly, and Manufacturing (ISAM) is a set of emerging operations that provides several benefits to improve the longevity, capacity, mo- bility, and expandability of existing and future space assets. Serial robotic ma- nipulators are particularly vital in accomplishing ISAM operations, however, the complex perturbation forces and motions associated with movement of a robotic arm on a free-flying satellite presents a complex controls problem requiring addi- tional study.
- Experiments use UR3e as the physical evaluation platform.
- The reported results provide an independent evaluation on UR3e hardware.
Amazon demonstrates ResMimic training for humanoid box-carry tasks
- Amazon FAR and academic collaborators introduced ResMimic, a residual-learning framework for humanoid whole-body loco-manipulation.
- A Unitree G1 carried a 4.5 kg box and handled irregular objects using whole-body contact.
- The work was published as arXiv paper 2510.05070 and demonstrated on physical hardware.
Kinodynamic-planning paper reports off-road field experiments with Clearpath Warthog UGV
- The paper studies kinodynamic motion planning across inconsistent world models for mobile robot navigation.
- The field experiment uses a Clearpath Robotics Warthog unmanned ground vehicle in an unstructured off-road environment.
ISyHand: A Dexterous Multi-finger Robot Hand with an Articulated Palm
- The rapid increase in the development of humanoid robots and customized manufacturing solutions has brought dexterous manipulation to the forefront of modern robotics. Over the past decade, several expensive dexterous hands have come to market, but advances in hardware design, particularly in servo motors and 3D printing, have recently facilitated an explosion of cheaper open-source hands.
- Experiments use Allegro Hand V5 Plus as the physical evaluation platform.
- The evaluation connects the reported method to Allegro Hand V5 Plus hardware in manipulation and dexterity tasks.
GD2P paper reports nonprehensile pushing and pulling experiments with Allegro Hand
- The GD2P paper studies geometry-aware nonprehensile pushing and pulling with high-DOF dexterous hands.
- The authors report real-world experiments with Allegro Hand and LEAP Hand embodiments.
- This is external product-use research, so the event is product-bound to Allegro Hand.
Zero to Autonomy paper evaluates online dynamics adaptation on Clearpath Jackal
- The method treats function-encoder coefficients as latent states and updates them from streaming odometry, enabling constant-time adaptation without gradient-based inner-loop updates.
- The authors evaluate it in a Van der Pol system, a high-fidelity Unity off-road simulator and on a Clearpath Jackal at a local ice rink where terrain dynamics shift abruptly.
- Across these settings, the paper reports improved model accuracy and downstream planning with fewer collisions than static and meta-learning baselines.
Learning to Assemble the Soma Cube with Legal-Action Masked DQN and Safe ZYZ Regrasp on a Doosan M0609
- The arXiv paper studies Soma cube assembly using a Doosan M0609 collaborative robot from the M Series family.
- The study’s named evaluation platform is M Series from Doosan Robotics.
HALO paper deploys human-preference reward learning on Clearpath Husky
- The HALO paper introduces an offline reward-learning method that turns human navigation preferences into a vision-based reward model.
- The authors deploy the reward model on a Clearpath Husky across diverse real-world navigation scenarios.
- The paper reports improved success rate, normalized trajectory length and Frechet distance against state-of-the-art vision-based navigation baselines.
Development of Educational Applications with ABB GoFa Collaborative Robot using Externally Guided Motion
- The ICCAD paper develops educational applications with the ABB GoFa collaborative robot using externally guided motion.
- The study’s named evaluation platform is GoFa CRB 15000 from ABB Robotics.
TACT paper uses RHP7 Kaleido for tactile whole-body contact manipulation
- The TACT paper introduces tactile-modality extended ACT for humanoid whole-body contact manipulation from human teleoperation data.
- The system is integrated with retargeting and locomotion control and demonstrated on the life-size RHP7 Kaleido humanoid.
- The work gives Kaleido/RHP a concrete tactile whole-body manipulation benchmark.
ROS 2 interface paper evaluates FANUC CRX collaborative robots
- The paper introduces a ROS2 control hardware interface for the FANUC CRX robot family and integrates it with MoveIt2.
- Experiments cover step response, trajectory tracking, collision avoidance, and dynamic velocity scaling.
- The results show usable path tracking and collision avoidance while documenting communication delay and velocity-limit constraints.
YuMi redundancy-parameterization paper validates arm-angle kinematics
- The paper defines and validates the Shoulder-Elbow-Wrist angle used by the 7-DOF ABB YuMi arm.
- It supplies the SEW-angle Jacobian, singularity conditions, and IK-Geo examples for finding all inverse-kinematics solutions.
- The result is useful context for YuMi path planning beyond RobotStudio-only workflows.
Distributed tactile-sensor control paper demonstrates multi-contact motions on RHP Kaleido
- The paper develops whole-body multi-contact motion control using deformable distributed tactile sensors mounted on humanoid limbs.
- The authors demonstrate motions including forearm-supported stepping and thigh-contact balancing on the life-sized RHP Kaleido humanoid.
- The work gives Kaleido/RHP a concrete benchmark for multi-contact humanoid control with distributed tactile sensing.
Preference-Driven Active 3D Scene Representation for Robotic Inspection in Nuclear Decommissioning
- The study proposes a real-time shared-control framework for active 3D scene representation in robotic inspection.
- The work addresses operator-specific objectives such as safety-critical coverage and task-driven viewpoints during inspection, manipulation and telepresence.
- The research is tied to UR3e usage, adding external research coverage beyond company-authored product releases.
Singularities of ABB’s YuMi 7-DOF robot arm
- The Mechanism and Machine Theory paper studies singularities of ABB’s YuMi 7-DOF robot arm.
- The study’s named evaluation platform is YuMi (IRB 14000 Dual-arm) from ABB.
A strictly predefined-time convergent and anti-noise fractional-order zeroing neural network for solving time-variant quadratic programming in kinematic robot control
- The paper evaluates a noise-resilient neural-network solver for time-variant quadratic programming with experiments on a Flexiv Rizon robot.
- The evaluation uses a physical Rizon Adaptive Robot Series platform supplied by Flexiv Robotics Ltd..
Cuspidal-manipulator path-planning paper identifies ABB GoFa kinematic behavior
- The paper studies path planning for cuspidal 6R manipulators, where inverse-kinematics branch choices can make a task-space path infeasible.
- The authors identify ABB GoFa as cuspidal and propose graph-based planning over inverse-kinematics solutions.
- The work adds research context for GoFa motion planning and optimization in industrial toolpath settings.
Flexiv Rizon-Based Multitasking Dual-Arm Robot Platform
- The book chapter presents a multitasking dual-arm robot platform based on Flexiv Rizon arms.
- The study’s named evaluation platform is Rizon Adaptive Robot Series from Flexiv Robotics Ltd..
Multimodal HRI paper evaluates LLM-guided voice and posture interaction on Universal Robots UR3e
- The paper presents a natural multimodal fusion system combining voice and deictic posture for human-robot interaction via a large language model.
- The system is evaluated on real-world tasks of varying complexity using a Universal Robots UR3e manipulator.
A Predefined-Time Convergent and Noise-Tolerant Zeroing Neural Network Model for Time Variant Quadratic Programming With Application to Robot Motion Planning
- The paper applies a noise-tolerant zeroing-neural-network model to robot motion planning and tests it on a Flexiv Rizon robotic arm.
- The study’s named evaluation platform is Rizon Adaptive Robot Series from Flexiv Robotics Ltd..
TrojanRobot paper evaluates physical robotic backdoor attacks on Universal Robots UR3e
- Researchers published TrojanRobot, a study of physical-world backdoor attacks evaluated on a Universal Robots UR3e.
- TrojanRobot studies physical-world backdoor attacks against robotic manipulation policies.
- The experimental setup includes a Universal Robots UR3e arm carrying a Robotiq Hand-E gripper.
Jailbreaking LLM-Controlled Robots
- RoboPAIR is designed to jailbreak LLM-controlled robots and target harmful physical actions, extending the risk beyond text-only chatbot attacks.
- The study evaluates white-box, gray-box and black-box settings involving NVIDIA Dolphins, a Clearpath Robotics Jackal with a GPT-4o planner and a Unitree Go2 with GPT-3.5.
- Across three datasets of harmful robotic actions, RoboPAIR and static baselines often reach 100% attack success, exposing a material safety risk for LLM-enabled robot deployment.
Failure Prediction from Limited Hardware Demonstrations
- Prediction of failures in real-world robotic systems either requires accurate model information or extensive testing. Partial knowledge of the system model makes simulation-based failure prediction unreliable.
- Experiments use UR3e as the physical evaluation platform.
- The reported results provide an independent evaluation on UR3e hardware.
Robotic ultrasound paper studies image acquisition with Universal Robots UR3e
- The paper studies robotic ultrasound image acquisition using diffusion models and partially observable Markov decision processes.
- The robotic imaging setup uses a Universal Robots UR3e arm for ultrasound probe positioning.
VLM-GroNav evaluates physically grounded outdoor navigation on Clearpath Husky
- The VLM-GroNav paper combines vision-language models with physically grounded proprioceptive signals for outdoor terrain navigation.
- The authors validate the method on a Clearpath Husky and a Ghost Vision 60 across deformable and slippery outdoor terrain.
- The reported experiments show navigation-success improvements by grounding semantic terrain understanding in real physical interaction data.
Reinforcement Learning Approach to Optimizing Profilometric Sensor Trajectories for Surface Inspection
- High-precision surface defect detection in manufacturing is essential for ensuring quality control. Laser triangulation profilometric sensors are key to this process, providing detailed and accurate surface measurements over a line.
- Experiments use UR3e as the physical evaluation platform.
- The reported results provide an independent evaluation on UR3e hardware.
ROS 2 Nav2 agricultural-navigation paper validates real-time costmap updates on Husky
- The paper extends the ROS 2 Navigation Stack with depth-camera classification and real-time local costmap corrections for outdoor agriculture settings.
- The system is implemented and validated on a Clearpath Husky to traverse tall grass and weeds that standard LiDAR costmaps may treat as obstacles.
- The work adds a Husky research example for practical outdoor Nav2 adaptation in orchards, vineyards and field-like environments.
Implementación de un entorno virtual para teleoperación de robot colaborativo ABB GOFA usando movimiento guiado externamente (EGM)
- The paper implements a virtual environment for teleoperation of an ABB GoFa collaborative robot using EGM.
- The study’s named evaluation platform is GoFa CRB 15000 from ABB Robotics.
Speech-guided sequential planning paper evaluates Llama3 navigation on Clearpath Jackal
- Llama3 interprets voice commands, extracts task details and converts them into sequential actions for scenarios such as pickup and delivery.
- The navigation stack uses a DRL-VO control policy, with simulation experiments spanning Turtlebot 2 in ROS1 and Turtlebot 3 in ROS2, including social spaces with infrastructure and crowds.
- Hardware trials on a Clearpath Robotics Jackal UGV demonstrate the system’s potential for flexible, interactive real-world navigation; the paper was accepted at the 16th International Conference on Social Robotics and AI.
Sensor-Based Distributionally Robust Control for Safe Robot Navigation in Dynamic Environments
- We introduce a novel method for mobile robot navigation in dynamic, unknown environments, leveraging onboard sensing and distributionally robust optimization to impose probabilistic safety constraints. Our method introduces a distributionally robust control barrier function (DR-CBF) that directly integrates noisy sensor measurements and state estimates to define safety constraints.
- Experiments use Jackal UGV as the physical evaluation platform.
- The evaluation connects the reported method to Jackal UGV hardware in navigation and sim_to_real tasks.
ROS2swarm - A ROS 2 Package for Swarm Robot Behaviors
- The package targets reusable, hardware-independent swarm-robotics software and provides ready-to-use behavioral primitives within ROS 2.
- The paper demonstrates aggregation, dispersion and collective decision-making behaviors across TurtleBot3 Burger, TurtleBot3 Waffle Pi and the Clearpath Jackal UGV.
- The authors emphasize modularity and platform independence as a way to simplify multi-robot communication, prototyping and swarm experiments.
Behavioral-entropy exploration paper evaluates Clearpath Warthog UGV in ROS-Unity simulation
- The paper introduces generalized behavioral entropy as an exploration objective for autonomous robots.
- The proof-of-concept evaluates exploration behavior with a Clearpath Warthog robot in a ROS-Unity simulation environment.
LEAP Hand: Low-Cost, Efficient, and Anthropomorphic Hand for Robot Learning
- Dexterous manipulation has been a long-standing challenge in robotics. While machine learning techniques have shown some promise, results have largely been currently limited to simulation.
- Experiments use Allegro Hand V5 Plus as the physical evaluation platform.
- The evaluation connects the reported method to Allegro Hand V5 Plus hardware in locomotion and manipulation tasks.
One-4-All: Neural Potential Fields for Embodied Navigation
- One-4-All uses self-supervised and manifold learning to create a graph-free navigation pipeline in which the goal is specified by an image.
- The system is trained offline on non-expert RGB sequences and controls, then greedily minimizes a continuous potential function without requiring depth or pose measurements.
- The authors report long-range navigation in eight simulated Gibson indoor environments and demonstrate successful real-world navigation with a Jackal UGV; the paper was accepted to IROS 2023.
Hunan Quanyu publishes 3T3R casting-cleaning robot patent
- The architecture combines three linear axes and three rotary axes around a controller.
- Its rotary-axis geometry is designed to retain an analytic kinematic solution while supporting complex casting poses.
- The disclosed system targets casting cleaning, rough machining and compound processes with a compact footprint.
Loc-NeRF: Monte Carlo Localization using Neural Radiance Fields
- Loc-NeRF uses a pre-trained NeRF as the environment map and Monte Carlo localization to estimate robot pose from an RGB camera as its only exteroceptive sensor.
- The method is designed to localize in real time without a good initial pose estimate, addressing limitations of earlier NeRF-based localization approaches.
- The authors evaluate synthetic data and real data collected by a Clearpath Jackal UGV, report real-time global localization and release the code publicly.
3D publishes AI weld-inspection case study
- The case study applied object detection to visual inspection of an automotive central-carrier weldment.
- The system assessed weld geometry, discontinuities, burn-through, porosity, spatter and missing fasteners.
- Detected object positions and counts supported dimensional measurements and automated weld-quality decisions.
3D publishes maintenance guide for continuous emissions monitoring systems
- The guide compared in-situ and extractive measurement architectures used in automatic emissions monitoring systems.
- It identified critical maintenance points including probes, filters, pumps, sample coolers, seals and energy supplies.
- It outlined preventive checks and QAL3 quality procedures intended to support the required monitoring availability.
Real-time collision-detection paper evaluates self-collision checking on Meca500
- The paper presents a high-precision real-time collision-detection system for robotics using URDF models and an extended Gilbert-Johnson-Keerthi approach.
- The evaluation checks self-collision performance on the Mecademic Meca500 and KUKA LBR iiwa 7 R800 robots.
X-Ray: Mechanical Search for an Occluded Object by Minimizing Support of Learned Occupancy Distributions
- For applications in e-commerce, warehouses, healthcare, and home service, robots are often required to search through heaps of objects to grasp a specific target object. For mechanical search, we introduce X-Ray, an algorithm based on learned occupancy distributions.
- Experiments use YuMi (IRB 14000 Dual-arm) as the physical evaluation platform.
- The reported results provide an independent evaluation on YuMi (IRB 14000 Dual-arm) hardware.
Learning Dense Visual Correspondences in Simulation to Smooth and Fold Real Fabrics
- Robotic fabric manipulation is challenging due to the infinite dimensional configuration space, self-occlusion, and complex dynamics of fabrics. There has been significant prior work on learning policies for specific deformable manipulation tasks, but comparatively less focus on algorithms which can efficiently learn many different tasks.
- YuMi (IRB 14000 Dual-arm) is the evaluation platform used to test the method and reported findings.
Learning Rope Manipulation Policies Using Dense Object Descriptors Trained on Synthetic Depth Data
- Robotic manipulation of deformable 1D objects such as ropes, cables, and hoses is challenging due to the lack of high-fidelity analytic models and large configuration spaces. Furthermore, learning end-to-end manipulation policies directly from images and physical interaction requires significant time on a robot and can fail to generalize across tasks.
- YuMi (IRB 14000 Dual-arm) is the hardware platform used for the reported demonstration.
Tactile Dexterity: Manipulation Primitives with Tactile Feedback
- This paper develops closed-loop tactile controllers for dexterous robotic manipulation with a dual-palm robotic system. Tactile dexterity is an approach to dexterous manipulation that plans for robot/object interactions that render interpretable tactile information for control.
- Experiments use YuMi (IRB 14000 Dual-arm) as the physical evaluation platform.
- The reported results provide an independent evaluation on YuMi (IRB 14000 Dual-arm) hardware.
Bayesian Optimization in Variational Latent Spaces with Dynamic Compression
- Data-efficiency is crucial for autonomous robots to adapt to new tasks and environments. In this work we focus on robotics problems with a budget of only 10-20 trials.
- Experiments use YuMi (IRB 14000 Dual-arm) as the physical evaluation platform.
- The reported results provide an independent evaluation on YuMi (IRB 14000 Dual-arm) hardware.
Dexterous Manipulation Graph paper demonstrates dual-arm regrasping on ABB YuMi
- The paper introduces Dexterous Manipulation Graphs for planning in-hand manipulation and regrasping with parallel grippers.
- The authors execute coordinated pushing and regrasping sequences on an ABB YuMi dual-arm robot.
- It adds a concrete YuMi research example for dexterous bimanual manipulation rather than simple pick-and-place.
Neural Network Model for Identifying Workspace, Forward and Inverse Kinematics of the 7-DOF YuMi 14000 ABB Collaborative Robot
- The IFAC paper models workspace and kinematics for the ABB YuMi 14000 collaborative robot.
- The study’s named evaluation platform is YuMi (IRB 14000 Dual-arm) from ABB.
Preparatory manipulation-planning paper evaluates single- and dual-arm tasks on NEXTAGE
- The paper proposes a regrasp-graph planner for reorienting objects through single-arm, dual-arm, and mixed manipulation sequences.
- The authors evaluate the planner across humanoid platforms including NEXTAGE, HRP2Kai, and HRP5P in simulation and real-world experiments.
- It gives NEXTAGE a concrete research record for bimanual manipulation planning rather than only industrial demonstration context.
Global Search with Bernoulli Alternation Kernel for Task-oriented Grasping Informed by Simulation
- Hardware experiments use an ABB YuMi, a rigid parallel gripper and Microsoft Kinect point clouds to test novel objects across six tasks.
- The simulated training set covers 605 object meshes from 13 categories and 4,500 grasps per object.
- Bayesian optimization found successful task-appropriate grasps for the spatula and mug handover tasks on its fourth trial and the pan task on its fifth.
Dex-Net 3.0: Computing Robust Robot Vacuum Suction Grasp Targets in Point Clouds using a New Analytic Model and Deep Learning
- Vacuum-based end effectors are widely used in industry and are often preferred over parallel-jaw and multifinger grippers due to their ability to lift objects with a single point of contact. Suction grasp planners often target planar surfaces on point clouds near the estimated centroid of an object.
- YuMi (IRB 14000 Dual-arm) is used in the study's data-collection or experimental dataset workflow.
Dual quaternion based kinematic control for Yumi dual arm robot
- The URAI paper presents dual-quaternion kinematic control for the YuMi dual-arm robot.
- The study’s named evaluation platform is YuMi (IRB 14000 Dual-arm) from ABB.
Dex-Net 2.0: Deep Learning to Plan Robust Grasps with Synthetic Point Clouds and Analytic Grasp Metrics
- To reduce data collection time for deep learning of robust robotic grasp plans, we explore training from a synthetic dataset of 6. 7 million point clouds, grasps, and analytic grasp metrics generated from thousands of 3D models from Dex-Net 1.
- Experiments use YuMi (IRB 14000 Dual-arm) as the physical evaluation platform.
- The reported results provide an independent evaluation on YuMi (IRB 14000 Dual-arm) hardware.
DLR publishes overview paper for torque-controlled humanoid TORO
- DLR published an overview of TORO at the 2014 IEEE-RAS International Conference on Humanoid Robots.
- The paper describes TORO's mechanical design, sensors, electronics, computer hardware, walking strategy, and multi-contact balancing approach.
- The paper gives TORO a technical baseline for torque-controlled humanoid design, walking, and multi-contact balance.
Improvement Trend of a Humanoid Robot Platform HUBO2+
- The paper “Improvement Trend of a Humanoid Robot Platform HUBO2+” was published on March 1, 2014.
- The paper reviews improvement trends for the HUBO2+ humanoid robot platform.
- The study’s named evaluation platform is HUBO2 from Rainbow Robotics.
Kawada and AIST announce HRP-4 research platform for working humanoid robots
- HRP-4 has 34 degrees of freedom, including seven per arm, in a 151 cm, 39 kg body.
- Each arm carries 0.5 kg, while OpenRTM-aist and a real-time Linux stack support research software.
- Kawada Industries and AIST designed the platform for human-scale environments and cooperative robot research.