Parametric tool correction algorithm for 5-axis machining

Abstract

Today, 5-axes CAM systems generate tool paths as a sequence of tool postures, which define the position of the tool tip and the orientation of the tool axis. This tool posture data file is usually called CLDATA, because it stores the position of the Cutter Location. In current practice, if the tool correction problem arises on the shop-floor, or at the machining simulation level, it can only be solved by sending back the code to the CAM system for re-generation. This iterative procedure is slow and very inefficient.

In this article, a data model is specified that stores machine independent and dependent information separately. Based on this segmented data model a general method is presented to solve the tool correction problem for 5-axis milling machines. The new data model and the provided algorithms allow us to solve tool correction and possible other setup problems on the shop floor level. The algorithm is developed for ball-, flat- and radius-ended tools and is valid for 5-axis milling machine with open kinematic chain.

Introduction

One major advantage of 5-axis milling compared to conventional 3-axis milling is the better quality of the obtained surfaces [1], [5], [9], due to the fact that the tool cannot only be positioned, but also oriented with respect to the workpiece. This advantage implies a higher processing effort compared to conventional 3-axis machining. In today′s manufacturing systems that effort is always handled at the CAM level. In a 5-axis CAM system, the toolpaths are generated as cutter location (CL) paths. These paths describe the position of the tool center and the orientation of the tool axis. CLDATA is the input to the post-processor (PP) of a specific 5-axis machine, that transform them to axis commands (see [8], [14]).

In current practice of data exchange between CAM and CNC systems [4], there are many stages, where adaptation or correction should be applied. This follows from the fact that certain machine specific information might not be known at the CAM level. Moreover certain adaptation-like tool correction in 3-axis milling-does not require considerable computational resources, which implies that they can be applied at the CNC level too.

Unlike the 3-axis CLDATA, the 5-axes one is not machine independent. Changing the tool orientation implies changes of the tool position, provided that the cutter contact point remains the same. This tool position correction depends on the kinematic structure of the machine and the geometry of the tool. Therefore, the 5-axis CLDATA can be compensated if the inverse kinematic map of the machine is known. This correction is an inherent part of the process, because the exact tool dimensions are known only before the manufacturing starts. Since the tool wearing changes the tool′s nominal dimensions, the CNC must have the ability to correct the tool path generated with a nominal tool dimension.

Several articles address this problem. In [15], a collection of tool-correction methods for a wide variety of tool types are given. But the author assumed that the tool-correction is part of the CAM system and is separated from the kinematic transformations necessary to calculate the axis commands. In [10], the tool correction was included into the controller to improve the surface quality. The author of this article used a simplified kinematic model for the 5-axis milling machine and also a numerical method to perform the inverse of the kinematic transformation.

Besides the need for tool correction, collision avoidance often requires to change the technological optimal tool posture data. This case the technologically optimal toolpath data could be spoiled in the vicinity of the corrected segment, but in return big time saving is absolved. In [7], the collision avoidance operation uses the CC (Cutter Contact) point and an optimized tool orientation, which is checked for collision using simulation. In case, the tool orientation changes an iterative algorithm is used to adjust the tool inclination angle in order to minimize the deviation between actual and the specified CC point.

In this article, a complete and general solution is proposed for the toolpath adaptation problem. First, a new data model is defined, where the exchange data is separated into machine dependent and machine independent parts. The machine independent part of the data model stores the prescribed path and the nominal machine parameters generated by the CAM system. The machine dependent part contains the adapted machine parameters and the adapted path. The data model is discussed in Section 2.

In Section 3, the inverse kinematic map calculation is shown, since the tool-correction is included completely in the kinematic model of the 5-axis milling machine. In the first step, the direct kinematic problem (transformation from the axis values to the tool placement) is generated including the tool correction for three types of tools (ball-, flat- and radius-end). The transformation from prescribed path to adapted path (inverse kinematic map) is generated using the Template Equation Method [2], [11]. This method allows not only to find all solutions of the inverse kinematic problem analytically, but also to handle parameterized kinematic models. These parameters are used to represent geometric values of the machine and of the tool (like diameter, tool length, etc), that-together with the inverse kinematic map-constitutes the machine parameters data segment.

This new data interface between the CAM system and the CNC is completely machine independent. This way the same prescribed path and nominal machine parameters can be applied on different machines, provided that the adapted machine parameter values are updated accordingly (see in Fig. 1). The proposed solution leads to a considerable rise in flexibility at the shop floor.