Aimed to reduce surface roughness (Ra) and improve surface quality of mold steel, the optimizations of process parameters for robotic polishing, such as polishing pressure, feed speed and rotating speed of tool, are accomplish in this research. The optimum range of each parameter is obtained according to a single factor experiment, and the central composite design experiments on the three polishing parameters are conducted to establish a prediction model of surface roughness. Furthermore, a significance test of the prediction model is carried out through variance analysis. The optimum polishing parameters are obtained based on the analysis of response surface, and are then adopted in the polishing experiments of mold steel for validation. The experiment result of model verification indicates that the relative errors of predicted Ra ratio and actual Ra ratio are within the allowable range (maximum is 13.47%). It proves the accuracy of the roughness prediction model. Meanwhile, the experimental results of multipath polishing show that the surface roughness decreased effectively after polishing with the optimum polishing parameters. The prediction model of surface roughness and optimum polishing parameters are helpful to improve surface quality in robotic polishing for mold steel.

Estimating the 6D pose of objects is crucial for robots to interact with the environment. 6D Object pose estimation from RGB images in a cluttered scene and heavy occlusions is a critical issue. Most existing methods use two stages to estimate object pose: First, extract the object features, and then use the PnP/RANSAC method to estimate object pose. However, most of these techniques merely localize a group of key-points by regressing their coordinates, which are vulnerable to occlusion and have poor performance for multi-object pose estimation. These methods cannot directly regress the 6D pose estimation from a loss during training. In this paper, we propose a framework based on convolutional neural network (CNN) and self-attention mechanism as an end-to-end method for single and multi-object 6D pose estimation using RGB images with low computational cost. Our method utilizes feature fusion to extract local features and combines multi-head self-attention (MHSA) with iterative refinement to improve pose estimation performance. Furthermore, our method can be scaled according to computational resources. Our experiments illustrate that our method performs in benchmark datasets the Linemod and Occlusion Linemod and achieves 97.45% and 84.84% in terms of the ADD(-S) metric in both datasets, respectively.

In this paper, we present an in-built N⁺ pocket electrically doped tunnel FET (ED-TFET) based on the polarity bias concept that enhances the DC and analog/RF performance. The proposed device begins with a MOSFET like structure (n-p-n) with a control gate (CG) and a polarity gate (PG). The PG is biased at −0.7 V to induce a P⁺ region at the source side, leaving an N⁺ pocket between the source and the channel. This technique yields an N⁺ pocket that is realized in the in-built architecture and removes the need for additional chemical doping. Calibrated 2-D simulations have demonstrated that the introduction of the N⁺ pocket yields a higher I_ON and a steeper average subthreshold swing when compared to conventional ED-TFET. Further, a local minimum on the conduction band edge (E_C) curve at the tunneling junction is observed, leading to a dramatic reduction in the tunneling width. As a result, the in-built N⁺ pocket ED-TFET significantly improves the DC and analog/RF figure-of-merits and, hence, can serve as a better candidate for low-power applications.