Accurate Robot Assistant

Información del proyecto

AccuRobAs

Identificador del acuerdo de subvención: 045201

Proyecto cerrado

Fecha de inicio 1 Octubre 2006

Fecha de finalización 30 Septiembre 2009

Financiado con arreglo a

• Information Society Technologies: thematic priority under the specific programme "Integrating and strengthening the European research area" (2002-2006).

Coste total € 3 793 431,00

Aportación de la UE € 3 047 000,00

Coordinado por

Karlsruher Institut fuer Technologie
Germany

Exploitable results

Universal robotic assistant system

The Accurobas project aimed to develop an innovative and universal robotic assistant system to support a human in dextrous manipulation. Methods to increase accuracy for lightweight compliant robotic systems during surgical procedures supporting different levels of autonomy were focus of the research in Accurobas. These levels vary from telemanipulation to autonomous behaviour. The Accurobas approach focuses on adaptive control by exhibiting rich sensory-motor skills and multi-sensor measurement to distinctly increase the system accuracy. The software architecture developed uses realtime CORBA for the communication between all components. For the implementation TAO is used. TAO generates C++ class files from the IDL specifications for the interfaces. The system was tested with the Open CORBA benchmarking suite. The test revealed with standard hardware a ping time of lower than 0.5 ms achieved on gigabit ethernet. The robot has an update cycle of 12 ms and the tracking system delivers the data with 16 ms. The communication delay compared to standard socket IPC is approximately 10% higher with TAO and is not resulting in any advantages in this scenario. The demonstrator and the performance analysis show the feasibility of the architecture. It is possible to control the hardware with a 12 ms update rate. It is even possible to read the data from the laser distance sensor with 1 ms update rate. To lighten the implementation interfaces for Matlab/Simulink have been developed. For the laser system an end-effector was designed. A body consisting of marker spheres is attached to the end-effector of the robot to measure the actual position with the optical tracking system. The end-effector consists of a colour camera, a scanhead, a flange for the laser mirror arm, a laser distance sensor and the already mentioned body. One approach in the Accurobas project was to combine a standard vision camera at the front of the end-effector with the optical tracking system. Different trials have been performed but until know the accuracy is not sufficient for laser ablation. The appendage will be continued in the FP7 project Safros. While a standard point to point transformation for the transformation between optical tracking system and robot is inadequate, a special registration algorithm was developed and therefore is used. This allows an automatic registration by moving the robot to different positions and store the position of a body attached to the robot TCP. The scan head is registered to the optical tracking or to the robot using a camera that tracks the prototype laser. Only the patient must be registered manually. With the known registration between robot and tracking system the robot can also be controlled via visual servoing. The marker-based standard registration method is used for the registration. Marker spheres are attached to the skull to track it. Additional titanium screws are drilled into the skull and a CT scan is performed. The titanium screws can be identified inside the segmented CT data. On the real bone a standard pointer device is used that is tracked by the same tracking system. To acquire the position of the marker the points acquired by the tracking system have to be fitted to a sphere. The following method describes the estimation of the points via using a tracked pointing device. The tip pointer device is put to the screw and moved. The tracked position of this pointer is fitted to sphere and the middle point of this sphere is the position of the screw. The Gauss-Newton algorithm is a standard method to solve such. Each titanium screw is pivotised using the pointer and the resulting point cloud is then registered to the point cloud from the CT data set to acquire the transformation between the two coordinate systems. The method used for the registration between the two point clouds, one from the CT data and one from the tracking system, is the method from Horn et al. For the optimisation of the registration two different approaches have been tested. The first calibration method uses the FARO measurement arm, this method was developed for the static case. In the second case a tracked pointer device is used for the dynamic case. The accuracy for the dynamic case is 0.5 mm that is outperformed by the registration for the static case, because the FARO measurement arm has a better accuracy. The drawback is that this method can not be used for the dynamic case.