Visual recognition is the most critical function of a harvesting robot, and the accuracy of the harvesting action is based on the performance of visual recognition. However, unstructured environment, such as severe occlusion, fruits overlap, illumination changes, complex backgrounds, and even heavy fog weather, pose series of serious challenges to the detection accuracy of the recognition algorithm. Hence, this paper proposes an improved YOLO v4 model, called YOLO v4+, to cope with the challenges brought by unstructured environment. The output of each Resblock_body in the backbone is processed using a simple, parameterless attention mechanism for full dimensional refinement of extracted features. Further, in order to alleviate the problem of feature information loss, a multi scale feature fusion module with fusion weight and jump connection structure was pro-posed. In addition, the focal loss function is adopted and the hyperparameters α, γ are adjusted to 0.75 and 2. The experimental results show that the average precision of the YOLO v4+ model is 94.25% and the F1 score is 93%, which is 3.35% and 3% higher than the original YOLO v4 respectively. Compared with several state-of-the-art detection models, YOLO v4+ not only has the highest comprehensive ability, but also has better generalization ability. Selecting the corresponding augmentation method for specific working condition can greatly improve the model detection accuracy. Applying the proposed method to harvesting robots may enhance the applicability and robustness of the robotic system.

Tool frame calibration has been widely used in robot-assisted printing, welding, and grinding, but it is not suitable for ultrasonic testing because the robot is submerged in water. The purpose of this paper is to present a tool frame calibration method, which is suitable for improving the precision of ultrasonic testing. In uniform mediums, sound travels along a straight line like ray. A reflector is fixed in water to reflect ultrasound, which makes it possible to measure distances between incidence points on a reflector and tool center point (TCP) on an ultrasound transducer. In addition, the positions and poses of the end flange are recorded through a robot controller. Finally, an optimization method is applied to calculate the position and pose errors of the tool frame relative to the end flange according to such records. The presented method was implemented in an ultrasonic testing system. We selected 100 incidence points on the reflector to calculate the assembly errors of the transducer. The pulse amplitude rose obviously after calibration, which verifies that this is an effective method. Considering that ultrasonic transducers can be used as a measuring tool, this paper proposes a tool frame calibration method for ultrasonic testing robots without introducing other measuring devices, which draws the conclusion that tool frame can be calibrated through ultrasound.

Diverse model factories have been established in universities and enterprises to support practical education across various fields. Increasingly stringent health and safety regulations have made practical equipment more complex and costlier. With the advancement of digital infrastructure, virtual reality (VR) technology has been widely adopted in education to simulate real-world environments. This study explores the application of VR technology in enhancing manufacturing process education. To achieve this, an interaction methodology based on the OPC UA standard is proposed to enable data exchange between virtual and physical environments. Additionally, a detailed workflow of the practice course, conducted in a physical model factory at North China University of Technology, is presented. This approach is particularly noteworthy because it allows students to validate simulated results using a physical system, rather than relying solely on virtual scenes to mimic real-world settings. Students were divided into two groups: a practice group using the proposed digital method, and a control group without digital tools. The number of mistakes from the practice group was 37% less than that of the control group. Statistical analysis of students’ grades and questionnaire responses concludes that the proposed methodology is valuable to improve students’ engagement and practical skills. The presented course is replicable for other training institutions.

In recently years, diverse education platforms have been progressively established across various disciplines to support engineering practice. In addition, the teaching method to improve students’ practical ability is equally important. This paper presents a novel education platform from NCUT (North China University of Technology) and introduces corresponding teaching methods based on the platform to promote the practical ability in engineering students. Taking the manufacturing process of a rotary part as an example, this paper explains the virtual reality integrated method, which is an innovative approach to enhance students’ engagement during the practice sessions. Students are divided into several groups and collaborate to complete the practice. The study concludes that the virtual environment simulated by CAD (computer-aided design) software, serve as an excellent complement to the real-world equipment, enabling students to complete operations repeatedly without the constraints of time and space. The proposed virtual reality integrated education platform and corresponding teaching method significantly enhance students' engagement and practical skills.