1 - INTRODUCTION
Computer-based visual inspection is nowadays employed in several production processes, either at end of the process, to check the quality of each product, or during the production itself, to analyse the status of the product and enable adaptive processing. The vision system is usually an industrial camera coupled with some structured light projector. An example is EyeT+ from IT+Robotics (Figure 1). The assembly line of an engine or the molding process can be taken as examples. The inspection can be carried out by an inspection robot that moves a 3D sensor to perform the quality control of the engine (Figure 2). Nowadays, the robot trajectories for visual inspection are usually generated manually or with semi-automatic motion-planning/calibration technologies. These inspection processes have been proved to be difficult, time-consuming and – consequently – expensive.
2 - OBJECTIVE
IT+Robotics’ objective within the EU ROCHI project, is to develop the core of a generic framework to tackle the coverage path planning problem for inspection tasks. The framework is meant to be a kind of “swiss army knife” for system integrators or manufacturing companies that need to deploy inspection robots. The framework deals either with inspection robots, which are used to move the vision system, and manipulators, which are used to move the product in front of the vision system (Figure 3). The inspection process is easily specified by selecting the region of interest on a 3D cad of the product (Figure 4). The framework comprises models of the robot kinematics, solvers for the motion planning and collision detection problems to generate the robot trajectory for accomplish the desired inspection task. As the end of the simulation, the framework generates the robot code and send it to the real robot to run the inspection process. The most innovative and important features of IT+Robotics’ framework are probably the accurate models of both the vision system components, i.e. cameras, optics and light projectors, and the inspection process (Figure 5). These models enable to have a precise measurement of the quality of the inspection trajectory, which is used to run the simulation until a satisfactory trajectory has been found (Figure 6). This is the key point to have a framework that really works in practise, even in scenarios with complex shapes. Here is where IT+Robotics wants to play an important role.
3 - IMPACT
IT+Robotics’ framework simplify notably the task of creating an inspection trajectory. Instead of requiring a couple of days spent for generating and testing the trajectories on the real robot, the task will take few hours mostly spent on the simulator. The main advantages are the following. First, a new inspection task can be crated on the simulator without stopping the production. This enables to reduce the cost required to implement an inspection process. The person that create the inspection task needs to have expertise on the product that need to be inspected and a basic knowledge about the robot used for the task, so crating new roles for the employees of the company using the framework. The overall reduction of both the costs and novel expertise required to implement an inspection process would make it widespread in the manufacturing.
The framework will be particular useful for industries where either the batch size is really small, e.g. foundries, high quality plastic and injection moulding, or the product geometry is very complex, e.g. aerospace and automotive industry. IT+Robotics’ framework is an important tool to generate efficient trajectory to increase the throughput of the production line.
