AI at the edge

(Unfortunately, this presentation is currently only available in German.)

In naher Zukunft wird KI zu unserem Alltag gehören. So werden selbstoptimierende Maschinen, Spracherkennung sowie diagnostizierende Medizingeräte uns täglich unterstützen. Die zugrunde liegenden Algorithmen sind extrem leistungsfähig aber auch leistungshungrig. Was steckt wirklich hinter der KI? Wie kann diese «at the edge», also eingebettet im Endgerät laufen, ohne Daten in der Cloud zu rechnen? In diesem Talk erklären David Gschwend und Florentin Marty, wie Ingenieure von SCS für Ihre Kunden erfolgreich KI in eingebettete Systeme packen.

Dieser Vortrag wurde am 9. Juni 2020 im Rahmen der Embedded Computing Conference 2020 mit dem “Audience Choice Award” ausgezeichnet.

You have an idea, a project or a problem you’d like to discuss? Get in touch with me now!

    Contact David Gschwend

    Protected by reCAPTCHA. The Google Privacy Policy and Terms of Service apply.

    We are extremely proud to announce that the world-renowned design prize iF DESIGN AWARD has been awarded to the rc_visard of our customer Roboception!

    From the very beginning, it has been one of Roboception’s goals to not only optimize our sensors in terms of functionality, but to make them visually attractive as well. In the end, they are typically mounted to their customer’s robots in very prominent positions.

    And obviously, the rc_visard’s combination of functionality and design convinced the iF DESIGN AWARD’s 63-member jury, made up of independent experts from all over the world.

    Further Infos at Roboception Website.

    Around 3 years after the first ideas, we were able to realize the rc_visard, a 3D-stereo camera for robots, on behalf of the project partners KUKA and Roboception within just 9 months. For this project SCS could rely on their extensive expertise in the electrical development and camera systems. This camera with the on board NVIDIA Tegra K1 chip allows realtime 3D measurements and positioning and enables a multitude of further developments in robot automation. For the development of the sensor, Roboception was focusing on intuitive usability and seamless integration with standard interfaces. At the Hannovermesse in April 2017, Roboception presented the solution to the general public and has worked since then on the ramp up of serial production. From end of September 2017, the rc_visard solution will be delivered to customers across the world. We congratulate for the successful product launch.

    Further information:

    The German company GLP Systems Ltd., based in Hamburg, is an innovative specialist in the area of information and automation systems for clinical laboratories. GLP Systems has revolutionized the sample transfer using a new approach: Similar to a Carrera race track, the sample vessels are individually moved through the laboratories and cold-storage rooms in intelligent CARs along lanes. Worldwide, lots of those systems are already being used, also in the Center of Laboratory Medicine at the Insel Hospital in Bern.

    Figure 1: The pool of empty CARs allows for an efficient and automatic filling on the right lane. In the Tube Assessment Center (TAC), at the right rear of the picture, the samples are classified reliably by the SCS computer vision system and start their individual route through the laboratory.
    Source: Center of Laboratory Medicine – Insel Hospital, Bern
















    Incorrect analysis results are being avoided thanks to the SCS computer vision solution: The samples are being classified in purely visual terms using their shape and color only. Practically, the world wide range of vessel types is a challenge, since their characteristics oftentimes only differ slightly. Using a statistical evaluation; those samples can be reliably identified despite their variations in production batches. Since the system automatically withdraws samples with unsure classifications, confusions are hence avoided. In those rare cases, employees check on the samples manually and guarantee for the mandatory safety.

    SCS Service Tool
    Figure 2: The SCS Service Tool allows a detailed status analysis. Thanks to rapid prototyping using MATLAB, the agile development took place quickly in a cost efficient manner and was soon ready for use.














    A service technician can quickly detect causes of uncertainties thanks to the SCS Service Tool: It analyzes and visualizes diagnostic images along with other data of the TAC. If required, a new series of images can be acquired and thus, the data pool used for the machine learning can be continually expanded. Applying this method, new types of vessels can easily be learned by the system. Moreover, new variations of known vessels can be understood better, therefore the recognition rate is steadily optimized.

    Picture 3: Thanks to swarm intelligence, the CARs drive autonomously through a variety of modules (TAC, centrifuges, cap removers, analysis devices etc.) This allows for an individual and cost efficient analysis of the blood samples according to the Industry 4.0 idea.  Source: Center of Laboratory Medicine – Insel Hospital, Bern

    The safety and availability of SBB railway infrastructures is ensured by the Infrastructure Monitoring business unit. SBB requires a “towed diagnostics vehicle” (gDFZ) to fulfill this duty and at the same time to provide an enhancement to the existing self-propelled diagnostic vehicle. The gDFZ can also measure the vehicle dynamics and overhead line. When in use, the gDFZ is drawn in a train composition consisting of a locomotive, a braking and a control car.

    The central system forms the heart of the new vehicle. It not only adopts the interaction with the user, but also the analysis and the persistent storage and display of all measurement data.

    The SBB has accepted the tender from SCS for the development of the central system (hardware and software). The successful concept in the public call for tender combines for the first time modular measurement equipment via open interfaces with the technology of a modern data centre and integrates them into a rail car. This flagship project meets all the essential features of an “Industry 4.0” application.

    The Microservice architecture allows a modular, flexible, extensible and highly scalable solution. The result is a clear and extensible software environment, which is extremely well suited for further development over the entire life of the gDFZ and which can also be operated efficiently.

    SCS also takes on the role of integrator for the overall system. System integration is the key success factor for the entire project. The chosen solution therefore supports the integration of all current diagnostic and support systems through a clear and systematic approach and through open, transparent and scalable system architecture. SCS has extensive experience both in the field of open systems, the integration of third-party systems and the development of “mission critical applications” including maintenance and support throughout the entire life cycle (Life Cycle Management – LCM), for example with the REGA communication system.

    We are delighted with this award and the prospect of working with SBB and the suppliers of the other work packages, peripheral systems and vehicle modifications.


    Background information on the application:

    The SBB infrastructure monitoring business unit is responsible for ensuring the safety and availability of railway infrastructure in compliance with the applicable regulations. The services provided by the monitoring unit are fulfilled using a mix of human expertise (route inspectors) and machine-determined diagnostic and prognostic data (Measurement and Diagnostic Technology). In the medium term the proportion of machine performance will increase due to the expansion of the industrial monitoring of the network and the commissioning of new high-speed lines (Gotthard and Ceneri Base Tunnel GBT/CBT).

    Measured test runs are state of the art in railway technology and essential for professional investment management, particularly in the track and traction current asset classes. The measurements in the field of driving technology are mandatory because of the legal and regulatory requirements (including R 22070). Reliable, consistent and meaningful measurement data also provides an important input for short-term diagnostics (monitoring), preventive maintenance and the prognosis of the substance preservation. They form the basis for the medium and long-term maintenance control of the asset classes.

    Industry 4.0, Internet of Things, Cloud – all these terms are currently being discussed all over. However, most of the time, a common understanding is missing about the meaning of these terms and the way they are to be implemented in the industry. Also, the benefit for the customer is not always clear.

    During a recent round table of the forum “Laufenburger Talks”, Dr. Johannes Gassner was invited as an expert to talk about this topic.

    The article „Swiss KMU is ready – but the benefit is still unclear” was subsequently written about this roundtable and was published twice:

    • Article in the “Technische Rundschau”, the Swiss industry magazine
    • Article in the “Polyscope”, the magazine for electronics and automation

    The “Internet of Things”, abbreviated IoT, has the intention to equip many electrical appliances with the ability to communicate with each other via the internet. Industry 4.0 stands for the pooling of technology trends and a vision of the way the manufacturing business may look in the future.

    The „Laufenburger Talks“ offer a platform to discuss interesting, controversial and current topics among experts.

    Contact person at SCS
    Johannes Gassner

    Wir demonstrieren ein Entwicklungssystem für Stereo-Kamera-Fahrerassistenzsysteme auf der Basis des Zynq All Programmable SoC. Optimiert für die Entwicklung von Bildanalysealgorithmen demonstriert dieses System sowohl eine Bildentzerrung als auch die Berechnung des SGM Stereo um die Lage und den Abstand von Objekten abzubilden.

    Um mehr zu erfahren, besuchen Sie uns an der Embedded World in Nürnberg.
    Wir sind zu Gast am Xilinx-Stand 1-205.

    SGM Stereo on SCS Zynq Box


    As an ecological pioneer, isofloc has been producing high-quality insulation from recycled newspapers since 1982. In both retrofits and new buildings, the injection of cellulose flakes allows for seamless and therefore very efficient insulation.

    Robofloc automatically fills pre-fabricated wood components (image on the right: reconstructed top view) with isofloc insulating material.

    In the field of sectional construction, isofloc is pursuing a process of innovative, user-orientated rationalisation: It has developed the robofloc filling machine which semi-automatically injects cellulose flakes into industrially manufactured wood construction components. Existing production lines can also be retrofitted with the robot system easily. High quality injection can only be achieved if the injection positions have been precisely defined on the wood components. The open-top elements are sometimes not aligned and might be delivered in any order on roller conveyors.

    To meet its quality specifications, isofloc has developed a camera system, together with SCS, which performs three-dimensional laser scans of the wood components in order to measure their shape and position very accurately. The precise 3D data allows for accurate, geometric modelling of the elements: the numbers and positions of the corners and side lengths, element depths and recesses in the walls are determined. The system also recognises spaces which are not to be filled and in this way prevents the unwanted egress of insulation flakes. Afterwards, the individual injection positions are determined, the robot advances to them, and the elements are filled in accordance with the injection data that has been calculated.

    The camera systems enable reproducibly high quality for the insertion of cellulose flakes into industrially manufactured prefabricated components, irrespective of the software interface, such as to CAD construction programs. It does not affect investments that may already have been made by manufacturers.

    Ferag AG is a leading provider of conveyance and processing technology for print media. Ferag was looking for a solution for the monitoring and optimisation of various processes at its facilities. By way of response, the SCS team has developed a computer platform based on an Analog Devices BlackFin DSP digital signal processor operated by uClinux. The CMOS sensor by Aptina is particularly suitable for moving scenes and guarantees long-term availability. The computer platform communicates directly with the controller via parallel interfaces (24V), Ethernet or EtherCAT. The open software framework provides the basis for efficiently implementing today’s tasks and can be expanded to meet future requirements.

    Ferag QualiEye in front of a Unidrum Conveyor System

    The ‘Machine Vision’ – solution by SCS allows badly positioned sheets to be detected automatically. In this way, it helps to prevent heavy financial losses and delays in print production. The computer platform can be used extensively thanks to low manufacturing costs. Furthermore, numerous servomotors and light barriers can be dispensed with in the construction of the machinery, since the image processing solution can handle various formats automatically and without mechanical modifications.

    Viasuisse is the Swiss traffic information centre based in Biel. A trilingual editorial team refines various data sources into high quality traffic information, which is disseminated through various channels.

    The left lane is 95% free, the right one with rush-hour evening traffic is 15% free.

    Viasuisse evaluates footage with regard to the traffic conditions and enters the results into your production system. Viasuisse wishes to optimise these processes. For this purpose, it requires automatic image analysis and a means of integrating the results into the existing framework without media discontinuities.

    Together with SCS, Viasuisse has demonstrated an innovative way to identify traffic conditions reliably and automatically, such as traffic jams or heavy traffic, using footage from greatly varying sources. This is then effectively integrated into the existing workflow.

    Integration of automatic traffic jam recognition into the work process of the editorial team

    The scalable platform for automatic image evaluation and the generation of client-specific traffic reports also manages the camera positions including meta-information such as geo-references. Operating costs can be saved through a centralised operation involving their distribution across several parties.