TheSixdegrees of freedom 6DoF pose estimation of texture-less objects provides a spatial understanding of industrial scenes and is the basis for accurate object manipulation. Recent studies have shown that the introduction of known model information and initial poses helps CNNs-based methods achieve better performance in complex scenes. However, the mapping from image space to pose space learned by neural networks is dimensionally lifting. Due to the lack of depth information, it is difficult for neural networks to capture local clues on texture-less surfaces and directly regress the relative 3-D translation and 3-D rotation. Instead, we propose a novel framework named REG-Net, which transforms the 6DoF pose estimation task into a 2-D keypoint registration problem. The proposed network first encodes regional prior information using multi-representation, utilizes the globally-consistent offset attention module to align 2-D keypoint features in a long range, and then estimates offsets and potential regions of keypoints. The proposed regional PnP simultaneously adjusts the keypoint locations in a short range and outputs the pose. This framework compresses the learning space of the network from 3-D to 2-D. Extensive experiments on two benchmark datasets demonstrate the robustness and accuracy of REG-Net. We further demonstrate the effectiveness of REG-Net in the reflective industrial part grasping applications.

The discontinuities of a five-axis linear-segment toolpath result in fluctuation in the feedrate, acceleration and jerk commands that lead to machine tool vibration and poor surface finish. For path smoothing, with the global curve-fitting method it is difficult to control fitting error and the local corner-smoothing method has large curvature extreme. For path synchronization, the parameter synchronization method cannot ensure smooth rotary motion. Aiming at these problems, this paper proposes a double B-spline curve-fitting and synchronization-integrated feedrate scheduling method. Two C2-continuous and error-bounded B-spline curves are produced to fit tool-tip position and tool orientation, respectively. The fitting error is controlled by locally refining the curve segments that exceed the fitting tolerance. The tool-tip position trajectory is firstly planned to address axial kinematic constraints in the feedrate scheduling process. Then the feedrate is deformed for the tool orientation to guarantee smooth rotary motion as well as to share the same motion time with the tool-tip position segment by segment. The feasibility and effectiveness of the proposed method have been validated by simulations and experiments on the S-shape test piece. Simulations show that the proposed curve-fitting method can generate smooth toolpath and constrain fitting error. The proposed feedrate scheduling method can guarantee smooth rotary motion and keep axial motions under kinematic limits, compared with the method that does not consider axial kinematic constraints and the parameter synchronization method. Experimental results verify that the proposed curve-fitting method can generate smooth tool path under fitting tolerance, and the proposed feedrate scheduling method can produce smooth and restricted axial motions.

Vibration of a high-speed elevator is one of the major factors contributing to a safe, stable and comfortable ride. Vibration control and systematic optimization rely on the establishment of an accurate and reliable model. An energy-based vibration model (EVM) is proposed in this paper to describe the multi-directional coupling properties of high-speed elevators. After analyzing the structural and kinematic characteristics, a combination of the kinematic energy, elastic potential energy and virtual works is implemented to describe the vibrational characteristics. The Gaussian precise integration method is implemented to obtain accurate solutions because of the infinite degrees of freedom and the time-varying parameters. An example study is conducted using the KLK2 high-speed elevator. EVM is established to obtain the $Max(P_k)/P_k$ and A95 indices of the X, Y, Z axis vibrational acceleration through simulations. A comparison is proposed between the proposed EVM method, conventional methods based on differential-equation vibration model (DVM) and real prototype experimental data. Results show that the proposed EVM method could achieve more accurate results than the conventional DVM method. Deviations of these indicators are less than 5%, which suggest the efficiency, precision and reliability of the proposed EVM.

Recently, deep learning-based defect inspection methods have begun to receive more attention—from both researchers and the industrial community—due to their powerful representation and learning capabilities. These methods, however, require a large number of samples and manual annotation to achieve an acceptable detection rate. In this paper, we propose an unsupervised method of detecting and locating defects on patterned texture surface images which, in the training phase, needs only a moderate number of defect-free samples. An extended deep convolutional generative adversarial network (DCGAN) is utilized to reconstruct input image patches; the resulting residual map can be used to realize the initial segmentation defects. To further improve the accuracy of defect segmentation, a submodule termed “local difference analysis” (LDA) is embedded into the overall module to eliminate false positives. We conduct comparative experiments on a series of datasets and the final results verify the effectiveness of the proposed method.

Accuracy allocation is crucial in the accuracy design of machining tools. Current accuracy allocation methods primarily focus on positional deviation, with