FreeMotionHandling Self-learning system for the future working world 01: Intelligent monitoring: with its installed infrared cameras, the indoor GPS is used to pinpoint the ball in an enclosed space 02: Integrated camera technology: using the camera on the outer sleeve, the handling system detects its surroundings even during the gripping process 03: Autonomous object detection: the camera in the gripper enables the system to identify objects and pick them up safely 04: Flying assistance system: conceivable application scenarios include working overhead, at dizzying heights or in spaces that are difficult to access Exact positioning using indoor GPS In order for the FreeMotionHandling to fly stably and with process reliability, a permanent communication system is necessary. The pinpointing of the flying object is ensured by the wireless sensor technology on board combined with the installed guidance and monitoring system, which has already proven itself on the Festo eMotionSpheres and eMotionButterflies. An important part of this indoor GPS is a camera system, as could also be used in the factory of the future. Several infrared cameras installed in the space record the FreeMotionHandling using eight infrared LEDs, which are attached to the flying ring as active markers. The cameras transmit the position data to a central master computer, which acts like an air traffic controller and coordinates the ball from outside. They are set up so that they map out the space as a whole and each flying object is recorded by at least two cameras at all times. Due to their special filters, they only capture infrared light and are not sensitive to other light. In this respect, the camera system measures the exact actual position of the flying object 160 times per second, upon which the computer readjusts each deviation. The flight path planning is therefore constantly updated. Besides the position, the system also directly determines the orientation of the gripping sphere. In addition, the speeds and accelerations are recorded. Precise object detection using onboard cameras Once the ball approaches the object being gripped, it takes over its flight-path planning independently, but with the aid of two integrated cameras: the first camera is located in the gripper and the second one directly on the outer sleeve next to it. The handling system thus perceives its environment even during the gripping process and can react to it in any given situation. Besides the item being gripped, the ball also detects possible obstacles on the ground that it can avoid. After being put into operation, an external control system is no longer required. The constant information exchange guarantees the process stability of the entire system. It is possible, however, to monitor all the parameters wirelessly and to make a regulating intervention. To enable the FreeMotionHandling to independently identify the objects it is looking for and then grip them safely, the developers rely on another scientific phenomenon. A mini PC is located on the outer sleeve of the ball and constantly analyses the signals from the two onboard cameras. By means of neural networks, its intelligent software is trained in advance to detect the objects to be gripped in any light and from all possible perspectives by itself. Object detection by means of machine learning For this particular purpose, a large number of images of the target object are transmitted to the system in various sizes, views and lighting conditions, which it saves as an array of positive examples. Just like the human brain, the computer can link the individual empirical values to generate a coherent overall image and subsequently use it to extract the key features of the object in question. Once the system has collected sufficient positive samples, it can then recognise and handle the learned object or comparable items in all situations. Taking the strain off people With the ability to be able to react independently to its surroundings, the FreeMotionHandling already meets one possible criterion for the workplace of the future. Self-learning subsystems and other intelligent components are taking on an increasingly important role in production. With Industry 4.0, humans and technology are working even closer together. An important element for future production facilities are human assistance systems that can adjust flexibly to a wide range of production scenarios. In this respect, the technology is able to react at all times to interventions by humans and other variable parameters. Machines are becoming increasingly harmless to people and also help to take the strain off them during their daily work. The concept of FreeMotionHandling could therefore be used anywhere and in all kinds of situations where people require additional support from machinery – for instance, when doing ergonomically one-sided tasks in assembly, during sorting as well as in warehousing. 04 03 01 02 6 Festo AG & Co. KG 7 FreeMotionHandling: autonomously flying gripping sphere