Getting a Grip on Robotics
How powerful software modules and high-performance embedded systems can drive the widespread adoption of autonomous, cooperative, and collaborative robots.
The Service Robotics Research Center of Ulm University of Applied Sciences is developing a modular software framework to make it easier to program robots. The goal is to provide software modules that can be used universally, for instance to swap robotic gripping arms from different manufacturers as required to generate new robotics solutions via plug and play. The team at Ulm University relies on congatec for highly scalable and standardized embedded computing hardware.
Today’s modern robots are highly complex constructions with numerous subsystems. They use manipulators with various axes and drive units, at the ends of which specific tools, gripper systems or measuring instruments are installed. Controlling the kinematics takes additional sensor systems, as does object and position recognition in pick-and-place applications for example. With the advent of autonomous and collaborative robots—sharing the same workspace with humans —many more tasks and building blocks are added. Examples include localizing and navigating mobile robots in industrial settings and safe human machine interaction (HMI). Industry 4.0 environments also need an M2M interface to the surrounding machines and systems. The goal is mutual task coordination. All of these different robot types—from autonomous to cooperative to collaborative—require powerful software modules and high-performance embedded systems.
High Market Demand for Smart Robots
Market demand for smart robots will grow rapidly in the coming years. For example, the market for autonomous robot systems is expected to grow at a CAGR of 23.7% until 2023, while the new market segment of collaborative robots is due to grow twice as much at an average 59% per annum. OEMs are under immense pressure to develop and to bring such new systems to market maturity as quickly as possible in order to participate in this high market growth. But the software development is a particularly great challenge for OEMs, system integrators, and users: More subsystems have to be integrated into the already complex autonomous robotics solutions if they are to become collaborative and/or cooperative.
Why the Closed System Approach Isn’t Working
Today, the software for robots is frequently still implemented as a closed system—usually with individually tailored x86 or other hardware including ASICS or FPGAs. Often, the software is even individually tailored for each robot, making reuse difficult. All tasks such as manipulator control, navigation, machine vision, task coordination, and HMI are programmed as a unit. It is therefore currently nearly impossible to exchange software components even for the most frequently required functions or to use them on another hardware platform. This means that for every new design, the robotics software has to be re-implemented. This is both error-prone and time-consuming and can significantly delay the rollout of much-needed innovative solutions—not to mention the hassle this causes operators who have to program each robot initially for its specific task.
Modular and Reusable
The development team of the Service Robotics Research Center of Ulm University of Applied Sciences under Professor Schlegel is now replacing this closed system approach, which burdens system integrators and users with additional work, with a modular software approach. Adopting this approach divides the complex overall robot system into several independent functional units. In a second step, it specifies the interaction between the individual units via fully and transparently defined interfaces.
This concept, SmartSoft, is now being expanded and widely marketed at the European level (EU H2020 project “RobMoSys – Composable Models and Software for Robotic Systems”) and national level (BMWi PAiCE project “SeRoNet – a platform for the joint development of service robot solutions”) in cooperation with partners from industry and research. Essentially, this approach aims to make it possible to assemble robotic systems from fully developed and tested modular software building blocks. This allows software developers to focus on individual function modules without having to consider the internals of the other components. More importantly, it makes it possible to combine functions such as the cooperative or collaborative elements as well as the logic for specific manipulators and a lot more in a modular way—even across manufacturers. Ultimately, this also reduces the effort required for system integrators and end users to make customer-specific adaptations and will significantly drive the widespread adoption of robotics.
So, let’s assume you have a manipulator from company A, combined with a chassis from manufacturer B, and a stereoscopic machine vision system from manufacturer C. The dedicated control software, like that in intralogistics applications for instance, is then easily assembled from the ready-made software components thanks to the high level of abstraction. Only minor adjustments are needed. No pipe dream, this application is already being tested in the real world. For example, the Ulm team has implemented the service robotics duo Larry and Robotino, which, in a pharmaceutical intralogistics application for Transpharm Logistik GmbH, assembles drug packages from individual trays completely autonomously and takes them to a specified delivery point.
In a slightly different configuration, the two robots have autonomously taken coffee orders and delivered them to the customer’s table. Thanks to the ready-made, freely combinable software components, the redesign took only a few hours. The video to see the two robots in action can be found at https://www.youtube.com/watch?v=Zz66I4NVtNU.
Containers with Clearly Defined Interfaces
To enable virtually any assembly of elements, the team from the Service Robotics Research Center of Ulm University of Applied Sciences has developed a software model with individual service-oriented components and a model-driven open-source software toolchain for the Eclipse development environment. This environment provides component developers with tools that they can use to build their own code for each functional unit and then secure those algorithms by automatically generated component containers. These containers communicate with other containers based on uniform communication interfaces. In addition, the wrapping also protects the component developer’s IP. The team has already developed several such functional modules and makes them available for use in their own projects. These include navigation modules, machine vision, HMI, manipulator control and task coordination, to name just a few examples. As a unifying communication interface, SmartSoft also relies on OPC UA. This allows manufacturers to focus on specific containers and build their core competencies here. Customers benefit from a much more flexible offering.
Generic Embedded Hardware Over Proprietary Designs
For the logic hardware, the Ulm team uses Intel® x86 technology to decouple the software development as far as possible from any specific hardware. With the appropriate glue logic, such an approach is also particularly easy to implement with x86 technology as far as the later migration of such systems is concerned.
Embedded x86 hardware is also particularly apt in this context because of the high standardization and comprehensive documentation. The form factors are standardized not only as regards dimensions, but also with regard to the application programming interface. Standardization facilitates replacement of hardware provided the boards comply with the EAPI specification of the PICMG or SGET UIC standard. Under those circumstances, it is even possible to vary freely between different form factors such as motherboards and Computer-on-Modules, depending on the requirements of the application, without having to significantly change how hardware is accessed during the migration. One supplier who attaches great importance to this standardization and its documentation as well as the simplest possible hardware integration is congatec, whose products the Service Robotics Research Center of Ulm University of Applied Sciences uses in its projects.
“Next to basic requirements such as maximum computing power, energy efficiency, and reliability, we also attach great importance to high standardization and the capability to migrate universally,” explains Matthias Lutz from Ulm University of Applied Sciences. “Every additional abstraction level in the software requires additional computing performance, so we’re currently working with powerful dual-core technology. A standardized approach to board components and GPIOs to control the robotics modules also gives us the abstraction required for independence at the embedded computing level.”
The choice ultimately fell on the fully industrial Mini-ITX carrier board conga-IC175. That’s because the standardized Mini-ITX form factor offers many advantages for developing the prototypes of the innovative software modules into real systems: It already integrates all interfaces on a standardized board, and the power supply is via standard ATX power supplies, industrial 12V feed-in, or SMART batteries, which is essential for mobile robots such as Robotino and Larry. Using PCIe expansion cards make implementing extensions quick and efficient. The board is highly energy efficient and uses robust embedded components, so it can be operated without expensive cooling.
Future commercial robot designs from Ulm will be implemented on Computer-on-Modules. But regardless of whether it’s a Mini-ITX motherboard, module with standard Mini-ITX carrier, module and individual carrier, or full-custom design: It is the Total cost of Ownership (TCO) that ultimately matters to OEMs, and when using modular software this is also determined by the software support of the hardware. To make it even easier to integrate more functionalities in the future, comprehensive support for real-time hypervisor technology can bring added benefits. This will give customers the option to integrate additional functionalities, such as their own IoT gateway, without having to use a dedicated hardware platform, which saves hardware costs.
Use Case: Pharamacy Intralogistics
Picking tasks are performed by a heterogeneous robot fleet in an intralogistics application at congatec’s industrial partner Transpharm Logistik GmbH. The autonomous picking robot Larry is equipped with a UR5 manipulator module and uses a Segway chassis. The transport robot Robotino has a conveyor belt instead of a manipulator to take the picking robot to any point. Orders are received directly from the warehouse management system via WLAN. The fleet management system selects two picking robots, which then execute the order. The application is based on results from the BMBF project “LogiRob – Multi-Robot Transport System in a Shared Human-Machine Workspace” and “ZAFH Intralogistics – Collaborative Systems to Increase Intralogistics Flexibility” (Baden-Württemberg and EU ERDF 2014-2020).
“We see clear benefits in such modular approaches as they mirror the modular approach of our software. In this respect, it is very interesting to see that with the acquisition of Real-Time Systems congatec now has virtually direct access to the hypervisor technology of these robotics and automation experts,” concludes Lutz.
Coupled with the Technical Solution Center (TSC), in which congatec consolidates all its OEM services, this results in a complete package for customers such as the Service Robotics Research Center of Ulm University of Applied Sciences or Transpharm Logistik GmbH.
Zeljko Loncaric is Marketing Engineer, congatec.
Prof. Dr. Christian Schlegel is in the Service Robotics Research Group, Ulm University of Applied Sciences