Research Highlights
On-machine Time-Series Tool Wear Dataset Generation
A microscope-based on-machine image acquisition system is developed to capture high-resolution cutting tool images and generate a time-series labeled dataset capturing tool wear evolution from the start to the end of tool life. The sequential information reduces subjectivity in labeling, as decisions are based on observing multiple images rather than a single image, assisting both accurate labeling and understanding of wear progression. Learn more
HG-XAI: Human Guided eXplainable Artificial Intelligence
Contrived Tool Wear Methodology
The develope contrived wear methodology generate wear on the tool artificially using a grinding process to avoid such time-consuming and cost-intensive cutting tests. The tools are worn by taking several passes over a grinding wheel in a controlled environment. The performance of contrived and naturally worn tools is compared by analyzing various parameters such as process force, wear topography and chip formation, which shows a good agreement. Learn more
Stochastic Toolpath
The concept of stochastic machining aims to introduce stochastic nature to the toolpath. The toolpath strategy involves random movement of the tool over the surface. It is hypothesized that the random motion of the tool results in movement of tool over already machined regions of the workpiece and enables rapid dissipation of heat and thereby reduces tool wear. Learn more
Control of Geometric Tolerances
The rigidity of the thin-walled component varies considerably with the change of workpiece curvature and reduces as machining progresses due to material removal. This variation leads to a violation of geometric tolerances envisaged by the designer. The research work devised a Rigidity Regulation Approach (RRA) to obtain the semifinished geometry at the end of roughing operation. The finish cutting sequence is performed subsequently on the geometry for achieving optimal geometric tolerances. Learn more
Designing-as-a-Service (DaaS)
The work proposes Designing-as-a-Service (DaaS) model for enabling systematic product innovations. The DaaS model proposes to connect skilled human resources with enterprises interested in transforming an idea into a product or solution. It is established that the DaaS has the potential for rapid and economical product discovery and can be readily accessible to SMEs or independent individuals.. Learn more
The HG-XAI integrates human expertise with the CNN-based Efficient-Net-b0 model to identify tool wear states efficiently. In the event of lower classification confidence of the model, HG-XAI will seek the assistance of human expertise in deciding the tool wear state. The HG-XAI displays the captured image, feature map, and probability values of the two highest-ranked wear states to the human for making the final decision. Learn more
