Abstract:In precision pick-and-place tasks, robotic grasp errors can cause the failure of subsequent placement steps. Therefore, in-hand object pose measurement has attracted the attention of the intelligent manufacturing industry. However, existing algorithms have limited accuracy and insufficient practicality. Moreover, they often assume that the actual object geometry is identical to the CAD model. To alleviate the problems in terms of accuracy, practicality, and uncritical assumptions, this paper proposes an in-hand measurement solution based on multi-view reconstruction for objects with simple features. In this concept, multiple monocular cameras are deployed at the robot end to calculate the object’s actual geometry and 6D pose by reconstructing its key geometric features. An in-hand pose compensation strategy and a robot terminal system integrated with manipulation and measurement are developed. To verify the solution’s effectiveness, several experiments were carried out. First, grasp errors caused by multiple sources are intuitively demonstrated in a simulation environment. The in-hand measurement is validated to compensate for grasp errors through comparison experiments. Then, robots can rearrange various irregularly piled objects in real-world experiments based on the developed terminal system. Results demonstrate that maximum remaining position errors are within 2 mm with maximum angle errors of less than 1°. The maximum error compensation rate reaches 95%, which is superior to existing in-hand measurement algorithms. It can be observed that the proposed solution is valid for multiple objects. Geometry measurement can help robots calculate the in-hand object pose and identify the object’s category and availability. Therefore, the proposed measurement solution seems accurate and universal for robot applications in smart manufacturing industries, enhancing the accuracy and flexibility of robot manipulation.

2D Rotation-Angle Measurement Utilizing Least Iterative Region Segmentation

A Systemic Model for Simultaneously Measuring Three dimensional Rotation Angles of the Linear Guide Track [J]

A Novel Method for Analyzing the Influence of Relative Attitude of Laser Profile Scanner on 3D Measurement System

A New Error Measurement Method to Identify All Six Error Parameters of A Rotational Axis of A Machine Tool

Four-Point Method in the Measurement and Separation of Spindle Rotation Error

Method of Binocular Rotation Angle Measurement Based on Machine Vision

A Novel Method and System Implementation for Precise Estimation of Single-Axis Rotational Angles

Machining tests to identify kinematic errors of machine tool table rotation axis based on sensitive directions

Practical Rotation Angle Measurement Method by Monocular Vision

A Visual-Based Angle Measurement Method Using the Rotating Spot Images: Mathematic Modeling and Experiment Validation

An Online Spindle Rotation Error Measurement System Based on Improved Three Point Method

Online approach to measuring relative location of spatial geometric features of long rotating parts

A Novel Method of Measuring Spatial Rotation Angle Using Mems Tilt Sensors

Accurate Angle Estimation Based on Moment for Multirotation Computation Imaging

Method Of Rotation Angle Measurement In Machine Vision Based On Calibration Pattern With Spot Array

Angular error measurement of workpiece repositioning using a full-scale rotation detection method

A Measurement and Separation Method for Perpendicular Errors of Rotary Axes

Optimization Scheme of Measuring Accuracy of Bearing Seat Size Based on Machine Vision

Three-point Method for Measuring the Geometric Error Components of Linear and Rotary Axes Based on Sequential Multilateration

A Novel Combined Method for Measuring the Three-Dimensional Rotational Angle of a Spherical Joint

Precise Geometry and Pose Measurement of In-Hand Objects with Simple Features Using a Multi-Camera System