Project Description

This project consists in a design methodology and cost-effective implementing solution for reverse engineering of complex-shaped objects using interlaced numeric and adaptive machining control.

The scanning of the object's surfaces is performed using a short-range laser device moved by the robot gripper along configurable patterns. Image enhancement techniques will be developed for noise removal, data thinning and multiple point clouds registration elliminating data overlapping. The 3D model of the object of interest is then reconstructed using depth map images.

An innovative, low computation effort solution is proposed for generating the toolpaths for complex 3D surface processing on milling machines. The generated adaptive toolpaths are optimized in order to increase accuracy, reduce tool wear and machining time. Tool compensation is obtained using an original method: the height map of the active tool profile is generated, and by applying image morphology techniques, the surface to be machined is obtained.

An autonomous, self-learning adaptive cnc machine control system is proposed in order to optimize the cutting parameters. Two strategies of adaptive control are implemented: manual (MAC) and automatic (AAC). MAC is designed as a knowledge-based, self-learning system based on sensor fusion, observation-prediction implemented by fuzzy techniques, and generating best machining strategies. The second approach proposed for real-time updating of machining parameters is based on nonlinear, inequality constrained multivariable optimization of the cutting process.

The embedded CNC-AC (adaptation and contouring) machine tool controller will be integrated as a set of functional holons in a multi-agent, service oriented manufacturing architecture.