Zhijie Yang

An experimental system for dismantling waste printed circuit boards (WPCBs) is developed based on YOLOv5. The system integrates a non-contact hot-air convection heating module with a 4-degree-of-freedom SCARA robotic arm to support vision-guided selective component removal. An STM32-based control unit and a Jetson Nano–based vision module are employed to coordinate motion control, temperature regulation, and visual perception in a closed-loop workflow. YOLOv5 is used to detect, classify, and localize components on WPCBs, providing target positions for robotic operations. A host-computer interface supports online human–machine interaction and flexible task configuration during operation. Experimental results demonstrate accurate SCARA trajectory tracking and stable temperature control of the heating module, enabling repeatable component detection, classification, and separation under controlled conditions. Overall, the system exhibits convenient operation and promising disassembly performance in laboratory trials, offering a baseline reference for component-level recovery and secondary reuse.

Composite surface structures inspired by the functional surface of Scapharca subcrenata shells can improve the tribological properties effectively, composed of the ordered “U”- shape micro-grooves and solid lubricant Sn-3.0Ag-0.5Cu (SAC305) alloys. A series of wear tests were conducted to further investigate the formation characteristics of the self-compensating lubrication film, and then the mathematical model of the spreading tribofilm could be proposed. The results showed that the appropriate surface texturing parameters (NBCSS-28) had a great effect on the formation of the self-compensating lubrication film, which exhibited a lower friction coefficient (0.386) and wear volume (0.682 mm³) than the other NBCSS samples. The tribofilm, with a thickness of a few microns, was deposited on the contact surface after the wear tests. The interfacial reactants (the Ni/Ni₃Sn₂ interface) of the SAC305 alloys, and Ni₃Al alloys confirmed by the wetting experiments and the thermogravimetric analysis, could promote the deposition and diffusion of the tribofilm during the sliding process. Hence, distinguishable layered structures could be observed on the fractured surfaces of the NBCSS samples. Moreover, the formation process of the tribofilm exhibited an obvious relationship with the reduction in the dynamic friction coefficient. The tribofilm formation model was proposed by the accumulation behaviors of the spreading tribofilm randomly in the form of multiple discrete irregular film shapes on the worn surface, which could predict the formation characteristics of the self-compensating lubrication film to improve the optimization design of the parameters.

In the field of computer vision, deep convolution neural network plays an increasingly important role, and its accuracy in various tasks has gradually exceeded the traditional method. However, real-time is important in autonomous driving and mobile robot scenarios. Most processes need to be run locally and completed in a short time with sufficient precision. Semantic segmentation based on deep convolutional neural networks consumes huge computing resources, and the balance of accuracy and speed has been a difficult problem in industrial deployment. We propose a general elastic computing acceleration method for semantic segmentation, which makes full use of the correlation between the front and back frames of the input video. We use optical flow, fuzzy prediction and the combination of these two methods to accelerate the inference of semantic segmentation. Experiments show that our method greatly saves computing resources and retains high accuracy. On platforms with limited computing resources, it can well balance accuracy and speed.

Based on Fischertechnik, this paper proposes and designs an alkaline soil restoration and vegetation planting machine. This device prossess the chemical restoration mode and plant restoration mode, and realizes self-adaptive soil detection and treatment. This paper also proposes an efficient workflow. The four-station conversion module can realize mechanism scheduling and the switch of restore modes, which save space and improve flexibility.