1. IMPLEMENTATION ISSUES REGARDING A 3D ROBOT – BASED LASER SCANNING SYSTEM Theodor Borangiu, Anamaria Dogar, Alexandru Dumitrache University Politehnica of Bucharest Dept. of Automation and Applied Informatics

2. Overview Project Description Integrating SLP with Robotic Arm Calibration Issues Constraints-based Motion Planning Adaptive Scanning Strategies Platform Simulator Conclusions

3. Project Description RE Platform Components Laser probe 6-DOF Robot Arm Turntable PC Milling Machine

4. Integrating SLP with Robotic Arm The Data acquisition module communicates with the laser probe, being able to take measurements. Encoder latching server module, receives the encoder readings from the robot arm and rotary table controllers using a TCP/IP connection. The Scanning trajectory generator is able to compute either predefined or adaptive scanning paths, based on the type and approximate dimensions of the scanned workpiece

5. Calibration Issues Robot-Laser Probe Transformation There are two transformation matrices that are only known with approximation. the transformation between the 6th link of the robot and the laser probe, established by robot – laser probe calibration the transformation between the rotary table and robot base, determined by robot – rotary table calibration

6. Defined Constraints Hard Constraints: known obstacles in the robot workspace articulated robot singularities articulated robot joint angle limits. Soft Constraints: surface avoiding (keeping the minimum allowed scanning distance towards the modelled object) flexible reach (avoiding “un-comfortable” positions of the robot arm) following the computed path.

7. Constraints-based Motion Planning

8. Input data - scanning toolpaths Output data - joint values of the robotic arm and the rotary table angle The problem presented here is the inverse kinematics for a 7-DOF mechanism. The computed solution has to satisfy the following requirements: minimize the accelerations and limit the speed of the rotary table; avoid collisions with any obstacles.

9. Adaptive Scanning Strategies It is proposed a two stage scanning strategy. In the first stage of the scanning process, there is no information regarding the object; this laser scanning will be done at high speed rates, which implies a low resolution data set. In the second stage of the scanning process, there is an approximate model of the object and it can be computed an efficient scanning strategy. There appears an occluded region due to one vertical wall, for which a supplementary trajectory must be generated. If in the first stage of the scanning process, the obtained cloud point is not analyzed, the next data acquiring pose is X

10. Adaptive Scanning Strategies In order to generate an optimal strategy for the second stage of the scanning process, the occlusions must be analyzed in the first stage. So a possible solution for the proposed profile is shown in the Figure, based on the following rules: if there is an occluded area, generate trajectory for scanning this area; otherwise, generate the next path so that the up-left corner of the trapeze have the same horizontal coordinate as the last point detected in the previous scan.

11. Adaptive Scanning Strategies In the second stage there is available an approximate model of the scanned object. This model is used for modeling the optimal strategy that will lead to the complete model of the object. In the a) case the intersection of the two trapezes sides on the profile surface does not create any occlusions. In the b) case the intersection of the trapezes sides on the mould surface will lead to an occlusion, in this case the trajectory must be placed so to be tangent to the vertical wall.

12. Platform Simulator-Static Simulator

13. The static simulation module allows specifying the robot pose, using one of the three input methods: specify the angular values for each joint and for the rotary table (direct kinematics) specify the position in Cartesian coordinates and the orientation in ZYZ’ Euler angles, in the robot’s reference frame (inverse kinematics with respect to robot) specify the angle of the rotary table, and the position and the orientation of the robot end point in the rotary table’s reference frame (inverse kinematics with respect to rotary table)

14. Motion Simulation Module

15. The motion simulation module lets the user simulate and analyze the behaviour of the robot using a sequence of user- defined trajectories. The user interface has an editor for the motion sequence, and controls for generating the animation.

16. Example

17. Conclusions An essential feature of the planning algorithm will be its ability to run in real time, while the scanning process takes place. The close-future work includes developing the algorithms for complex 3D path following, optimizing and speed – up methods of the implemented algorithms. This work has been done in the Robotics and AI Laboratory of the Faculty of Automatic Control and Computers. www.scanrobot.cimr.pub.ro

18. Thank you! Questions?