An approach to semi-automatically determine mechanical hazards in VR models

Abstract

A risk assessment is typically required by legal regulations in production automation. These regulations vary from country to country. Risk assessment is often incomplete in industrial practice. Not all hazards are detected and risks are estimated insufficiently. Furthermore, risks are mostly assessed after the design of the product and/or production process. As result additional safety measures to reduce risks will cause higher costs and delays. This paper proposes an approach to semi-automatically derive risk from a virtual reality (VR) model of a machine. The focus is on the specific category of crushing hazard. These include situations where a human being might be hurt by being crushed. The VR model is based on a CAD model which mostly has been developed during the design phase of almost all technical products. The core idea is to identify moving parts of the machine and to calculate possible crushing zones by automatic distance approximations. The used algorithm relies on geometric CAD data. It uses axis aligned bounding volumes (AABB's) and the octree data structure. The concept is implemented in a prototype which especially determines crushing and shearing hazards and visualizes the result in the VR model for the experts. The software supports the process of risk assessment a precise and methodical approach and hardly minimizes the possibility of non-considering potential safety hazards. The prototype permits an effective, exact and methodical practice to the experts responsible in the process of the risk assessment by giving targeted advices. Furthermore, it allows the analysis of crushing hazards already during the design phase.