Mold polishing is one of important processes of mold manufacturing, and it is a bottleneck in mold industry that needs to be broken. At present, mold polishing is mainly performed by traditional manual methods, which is labor-intensive, low-efficiency, and time-consuming, it is a job that requires professionals with rich technical experience. In order to improve production efficiency and product quality, solve shortage of professional workers, advanced industrial countries have begun research work on automatic mold polishing very early. As a kind of automated production equipment, robot has a good flexible structure, and is especially suitable for polishing free-form surface of mold. Therefore, research of robot-based mold automatic polishing platform has received extensive attention.

 

Flexible structure of industrial robot makes end effector unable to withstand a large load, mold polishing is a finishing process with little or no material removal, polishing force is small. When industrial robot is used for mold polishing, it just avoids disadvantage of its low load, exerts advantages of flexible structure, automation and high degree of freedom. A complete set of robot mold polishing system has been successfully built and used for mold polishing test,

Industrial robot itself is not tailor-made for mold surface polishing, so its structure and system must be modified, expanded to adapt to mold polishing process.

Due to flexible structure of robot body and complexity of mold polishing work, robot mold polishing process cannot establish an accurate dynamic model; at the same time, traditional PID and PD control-based robot controllers have poor dynamic performance, low positioning accuracy, and poor robustness. Industrial robots usually use PID control algorithms, but this algorithm does not provide ability to compensate, especially for frequent sudden parameter changes. In order to accurately control polishing speed, polishing position and tracking trajectory, researchers seek a control method based on modern control theory and artificial intelligence, which overcomes inability to establish a dynamic system model due to nonlinear friction, load changes in an open system and environment.

Polishing force is pressure in specified direction exerted by polishing tool on the surface of mold. Difference in polishing tool leads to different control strategies and algorithms. In order to maintain a constant or variable polishing force throughout polishing process, a force sensor is required to measure contact force between polishing tool and mold surface, at the same time feed back measurement data to controller and adjust actual polishing force according to different process conditions.

Different robot mold polishing systems have different polishing process parameters due to different system designs and polishing methods, but ultimate goal is to improve mold polishing efficiency, surface quality and surface uniformity. Main process parameters are: equivalent radius of contact area between polishing tool and mold surface, polishing force, polishing feed rate, polishing line speed, polishing posture, and number of polishing. Among them, polishing force and polishing posture are the two most important parameters in polishing process. If polishing force is too large or too small, mold surface will be over-polished or under-polished; posture of polishing tool is related to distribution of polishing linear velocity in contact area and uniformity of polishing effect.

As requirements for surface quality of mold become higher and higher, since contact between commonly used mechanical polishing tools and mold surface is a rigid contact, requirements for trajectory tracking are relatively high. A small error in trajectory will result in a significant change in polishing force, thus new types of flexible contact and indirect contact polishing methods have emerged. These polishing methods mainly start with design of polishing tools. Changing way of interaction between polishing tool and mold surface can largely avoid irreversible damage due to trajectory tracking errors. There are three main types of polishing tools:

The most commonly used method for buffering rigid impact of polishing tools and mold processing surfaces is to coat surface of polishing tools with a layer of wear-resistant, high-friction and high-elastic fibers (such as non-woven fabrics, wool felt, etc.). This method can protect machining surface of mold within a certain impact range, obtain better surface quality without leaving traces of tool path edge. Its disadvantage is that friction between polishing tool and surface of mold causes temperature in contact area to rise, which reduces friction with processed surface of mold, thereby reducing polishing efficiency.

Based on concept of flexible polishing, bladder polishing technology can improve stability, efficiency and quality of polishing of free-form surface of mold. A rubber bladder with controllable air pressure is installed at the end of polishing tool, and a soft polishing material is coated on it at the same time. In addition to characteristics of soft material wrapping method, pressure inside airbag can be adjusted according to curvature of surface to be polished and depth of local material removal, polishing tool can fit well with mold surface.

Elastic polishing tool is fixed at the end of corresponding position of robot, and does not directly contact processing surface of mold. High-frequency longitudinal displacement is generated under drive of ultrasonic vibrator, so that free abrasive and machining surface of mold interact to perform polishing operations.

First problem to be solved by robot mold polishing system is path planning of mold processing surface. Path planning in existing research mainly uses CAM-based CL data and surface feature visual recognition technology.

The most commonly used in research is to modify APT toolpath generated in CAM, and then transplant it into robot system. Since most of molds are designed through CAD/CAM software, APT toolpaths generated by CAM software are stored as text files, which is convenient for data processing and conversion. APT tool path file contains CL data and tool-moving methods. CL data contains coordinates of points on processed surface and direction of corresponding tool axis. Robot path program can be obtained by corresponding processing for different robot platforms.

CAM toolpath planning usually adopts topological structure of line cutting or circular cutting. Although CAM is quite mature in path planning, it is designed to solve processing path problem of five-axis and five-axis CNC systems. It is not completely suitable for industrial robots with six degrees of freedom. If line cutting or ring cutting is used, it will also cause polishing protuberances and uneven edges on processed surface of mold after polishing. Polishing mold surface from as many directions as possible can obviously overcome this defect.

This technology integrates path planning into industrial robot mold polishing platform, and develops paths suitable for robot platform system; it does not use CAM to generate tool paths, but only uses CAD model of mold.

This technology does not rely on any CAD/CAM model. It captures surface features and analyzes surface quality through a vision sensor, then selects different polishing processes for each area with different surface quality.