Advanced methods for robotic machining of hard materials

MasterCAM NX G-CODEEasyRob PDL2 language

C4G controller with PDL2 real-time condition-handler and override adjustment

Introduction

• The removal of materials is one of those least-utilized techniques within robotic applications.

• Almost all robotic producers offer today robots for manufacturing applications, however, they are mainly for robot finishing.

• CAM packages do not consider specific performance and limitations in robotic system kinematics, dynamics and control.

Experimental set-up• Comau robots are chosen because

of innovative C4G Open Control interfaces (access controller (servo and programming) via UDP/IP in real time with a rate of 1 ms)

• Algorithms for force, impedance control and path governor are running real-time on RT patches (RTAI, RTNet) on external Linux (Ubuntu) PC. New commands are integrated in the standard robot language (PDL2) to manage and monitor external control algorithms at the programming layer

Machining off-line planning• Geometries are designed in a CAD

program or scanned by a 3D scanner, converted in surface model before processed by CAM software

• Any standard CAM software can be used (NX and MasterCAM programs were applied to generate standard NC-machine G-Code in this project)

• G-Code is then processed and simulated by the robot simulation software (EasyRob) and converted into PDL2 language after testing

• PDL2 is then executed by the C4G controller

Machining off-line planning• Before the final generation of robot

program, several specific robot milling problems (detection and avoidance of collisions, singular configuration, restricted work space, bad manipulability, etc. must be addressed)

• Algorithms are implemented in C++ and MATLAB and integrated in the EasyRob environment (by means of open APIs) as advanced robot machining Toolbox. The Toolbox involves kinematic, dynamic and control robot models (implemented in MATLAB). Specific robot signature is taken into account

On-line compensation controls• Despite careful planning, initial

experiments with standardized parts demonstrate higher position errors at specific robot configurations. During direction changes, one axes remains jammed while other axes achieve maximum speed to keep required Cartesian velocity causing unaccepted quality of work piece.

• To cope with this problem on-line algorithm referred to a real-time velocity path governor is implemented.

On-line compensation controls• The control algorithm has been

implemented in the C4GOpen robot control system on the external PC in C++ and using language control interface. It is running at 10 ms and is adjusting programmed real-time override within range 5-100% of the nominal speed using C4GOpen functions (PDL2 language real-time condition-handler and real-time override adjustment functionalities)

• Blue - without path-governor,• Red - with path-governor

Conclusion• Very hard material milling (e.g. Inconel, granite) is beyond the scope of

this project. Chattering effects cause additional and dominant errors due to oscillations caused by elastic effects in robot structure. Path-governor algorithm becomes insufficient to cope with such kind of errors.

• Robot grinding and polishing operations requires even higher precision (hundredth of mm) than milling to achieve good quality. Which is beyond robot positioning accuracy.

• Advanced approaches address an inclusive set of robotic machining operations providing a systematic methodology to meet requirements of end-users, imposing “all or nothing” mind-set and requiring complete solution for hard materials machining by robots, rather than only pre-machining.