On the way to fully flexible logistics, autonomy in decision-making and the self-organisation of the material flow are no longer conceivable without them. Completely new, highly dynamic classes of transport vehicles and transport robots are emerging that pose new challenges for decentralised control. Today, virtually no driverless transport vehicle (AGV) or transport robot is fully autonomous. The few autonomous systems are proprietary and do not offer a standard interface. Against this background, the “OpenDynamics” project aims to develop a new class of FTF or robots that actually act autonomously. All components will be published as open source. In this way, the “OpenDynamics” project also follows the guiding idea of the Silicon Economy to provide companies with software and hardware for standard logistical processes and tasks – so-called commodities that are not market-differentiating.
Four subprojects
There are four sub-projects in the “OpenDynamics” project: In the area of software, these are navigation & simulation and localisation & sensor technology; in the area of hardware, two different vehicle or robot platforms – one for pallet transport, one for KLT/parcel transport.
+ Navigation & Simulation
The use of artificial intelligence is moving into the focus of the automation of highly complex systems. New types of simulations offer new possibilities: Highly complex processes – including physical ones – can be simulated in real time using the strongly parallel processing of modern graphics cards, thus forming the basis for a new class of algorithms – “simulation-based AI”.
Simulations use models for abstraction. For the development of such highly dynamic systems, the behaviour of the simulated transport vehicles is compared with that of the real ones in a special, particularly suitable test environment, thus optimising the simulation model. If the difference between model and reality is reduced, the simulation turns into a Digital Reality for the AI and the robot becomes the CPS twin (CPS = Cyber-Physical System) of the simulation.
The AI-based simulation serves to accelerate the development of transport vehicles or robots, as they can be continuously adapted to changing conditions during the development period in the simulation without much effort.
Developments for the Silicon Economy
- Navigation Toolbox: Analysis and comparison of different methods for navigation and control, implementation of further, generic planning algorithms if necessary.
- Physical simulation models for two vehicle platforms
The results will be shown both in simulation and on the real vehicle platforms for different use cases.
+ Sensors & Localisation
The complexity of mobile robot systems is constantly increasing – especially with regard to the speed of the systems and indoor/outdoor applications. In the project, individual components are developed that can be used alone (“stand alone”) or in combination with other components. Use in combination with already existing open source software is also possible (keyword: interoperability). For some components, artificial intelligence methods are used, e.g. for detecting objects, which enable a robot to be positioned relative to these objects.
Developments for the Silicon Economy
As a result, a Mobile Robot Localisation Toolbox is to be developed, which includes the following “tools”:
- GPSFuser and DGPS Server
- Landmark Detection (orientation to fixed objects in the environment of a robot, still under development)
+ Dynamically stable robot
For this platform on two wheels, the driving principle of the “inverted pendulum” is used. The platform can be commercially equipped with a wide variety of gripper solutions for holding, positioning and moving goods. Thanks to the pendulum motion, a robot based on the platform can lift objects directly from the ground and deliver them at different heights. This is a very smart way to handle many intralogistics tasks that previously required the use of multiple robots. This improves the handling of goods and maximises pick rates. The navigation and localisation software from the project is used in the platform.
The blueprint for the first stage of the platform will be made open source. The platform can thus be used as a basis for expansion to use-case specific applications by companies.
+ Highly dynamic outdoor robot
Many automated guided vehicles today are designed for low dynamics and are designed either for indoor or outdoor use on company premises. They have either high performance, dynamics or flexibility. All three characteristics together have hardly been realised in a single driverless transport vehicle so far. In the project, an automated guided vehicle has now been developed that can transport large loads in the format of a pallet omnidirectionally with high dynamics and at speeds of up to 14 m/s. The vehicle is designed to transport the protected area of a factory site either indoors or outdoors. It is designed to leave the protected and defined environment of warehouses and operate on the respective company premises.
With the help of the localisation toolbox, the vehicle’s navigation functions seamlessly at the transition between indoor and outdoor areas. It is implemented using environment-based and radio-based localisation algorithms. The vehicle is based on the research results and the vehicle base of the LoadRunner®.
For the development of the vehicle control system, strong emphasis is placed on the use of artificial intelligence for the realisation (see Navigation and Simulation).
Contact:
Dr.-Ing. Sören Kerner, soeren.kerner@iml.fraunhofer.de
Dipl.-Ing. Guido Follert, guido.follert@iml.fraunhofer.de
