Public Summary Month 3/2012

Within the last two months we have focused on integration and evaluation which took place at office and lab environments and at the SCHUNK production facilities. Within the onging integration we were able to fix many of the open technical issues. Besides the developed mobile manipulator itself the main results are:


Localization according the ambient magnetic field

Overall, the localization results were satisfying, even though the recording of sensor data for larger environments still poses a difficulty. A global search for the best matching position of a sensor path history found the robots position with an accuracy of 10cm to 5cm in 8 to 9 out of 10 tests. The few failing tests can be attributed to particularly bad path movement data – if the robot performs mostly on-the-spot rotations, the matching of the path data structure will fail, because there is no path to speak of. We can therefore conclude that magnetic localization will only provide good results if the robot has driven a suitable distance before evaluating the sensor readings.

 

Low cost box detection

We have designed and realized special gripper fingers that are able to grasp standard crates while carrying 12 MAGMODULEs and a small camera, too. We are able to coarsely position the arm over the box. From there, the magnetic box detection has to take over. We found out that in case of an empty crate only disturbances of the earth's magnetic field the measurements reaches standard deviations of approximately one millimeter. This result is to be interpreted as very positive and a stable use under a real production surrounding seems to be possible. But, we also found out that metal plates act as a disturbance. The largest error of three millimeters were identified in those cases.

 

In addition, we have presented the project results at 5th International Expert Days on Service Robotics and at the Innorobo 2012 in Lyon. The next events for demonstration and exploitation are Hannover Messe within the area of mobile robots and autonomous systems in April 2012 and the Automatica within the ECHORD booth in May 2012. With KANMAN we have applied for the German “Innovationspreis IT”, a yearly contest about the best new IT products. The contest is managed by the Initiative Mittelstand under the patronage of the Federal Commissioner of the German Federal Government Cornelia Rogall-Grothe and IBM Germany. Under in total 2,500 applications KANMAN was awarded under the TOP 20 in the section logistics.

 

Finally,... thanks to our experiment moderator Germano Veiga and the whole ECHORD-team for their support during our experiment!


Public Summary Month 1/2012

Within December 2011 and January 2012 we have integrated the developed technologies to the mobile manipulator prototype and its evaluation. We have finished the hardware setup of the prototype, improved the magnetic sensor software and we have focused on the processing and evaluation of processing and evaluation of magnetic sensor data in respect of the localization of the mobile robot base. Initial measurements revealed a distribution of magnitudes that hardly changed, regardless of where the measurement took place. Test measurements in our lab environment have shown a reproducible magnetic footprint that looks promising in regards to localization. As the localization algorithm needs discrete data, we have written a conversion tool that takes raw magnetic sensor data and yields a grid map.

Furthermore, we have been adapting the robot arm to the new control hardware, but there are still communication issues to be solved within further integration meetings. We plan to do the final evaluation at the SCHUNK plant in March 2012.

Tags: public summary  

Public Summary Month 11/2011

For the magnetic localization of the mobile robot we have mainly finished the evaluation sensor array based on 30 MAGMODULES. Regarding the software implementation of the MAGMODUL for the magnetic localization we have created a tool to insert reference points to eliminate odometry errors. The maps are now odometry error free. The hardware set-up of the whole mobile manipulator is almost finished and integration meetings have taken place.

The work performed for the magnetic tracking of boxes was described in the deliverable KANMAN.4 which is submitted in the meantime.

Tags: public summary  

Public Summary Month 09/2011

We have been continuing the work in this task "Safe manipulation in industrial environments" and tested mechanical hazards of the robot arm and prepared the evaluation of bumpers and tactile sensors. The final hardware build up of the MAGMODUL for the ambient magnetic field measurement is finished. Regarding the software implementation of the MAGMODUL we have implemented a driver for magnetic field sensors that automatically corrects/normalizes measurements and outputs them as vectors. Furthermore, we have implemented a mapping module that takes these vectors as input and combines them with the robot's odometry/position. With the MAGMODUL measuring units it is also possible to realize a precise positioning. For two-dimensional precision positioning it is necessary to build also a two-dimensional array of MAGMODUL measuring units. A holder which fixes the modules on the KANMAN robotic platform is under construction.

 

For the set-up of the mobile manipulator we have evaluated the battery system, finalized the design of the robot base including drive system, battery storage, and two laser scanners, developed and integrated a novel unit for the control of low level robot functionalities and improved the usability of the robot platform with a special focus on installation and transportation.

Tags: public summary  

Public Summary Month 7/2011

In the last two months we have been continuing our work on the tasks save manipulation in industrial environments, magnetic localization of mobile manipulators & tracking of boxes and integration and adaptation of the mobile platform. We have submitted the deliverable KANMAN.3 in which we have described the progress towards a safe manipulation of modular kinematics in a collaborative way without traditional safety devices. We have implemented a bumper to eliminate sharp edges. We realized a safe robot arm concerning stop 0 using components according to performance level d. We were able to finish the development of the MAGMODULS which we use to measure the ambient magnetic field.The development and the composition of the module is described in the deliverable KANMAN.2. With a 3D Helmholtz coil we have measured the slope of the AMR sensor elements and their alignment. The maximum error was 1,34 %. This deviation are passable compared with the linearity error of the AMR sensor elements. With one calibration the system has the possibility to switch to different gains. This situation allows to switch active if the ambient magnetic filed changes on different basements. The next step is the evaluation of the sensor array on the SCITOS G5 mobile base to prove their applicability to support the localization. Furthermore, we have been continuing the assembling of the KANMAN mobile base and mechanically integrated the arm structure. The development of the main control unit of the mobile base is almost finished and we except to have a driveable mobile manipulator by end of September.

Tags: public summary