In the professional environment and due to the growing trading volume of consumer products, the logistics sector is under high cost and time pressure. Robotic innovations enable companies to react to these requirements in a way that makes economic sense, allowing them to automate previously manual tasks in an intelligent and flexible way. A core requirement of robot systems is their ability to recognize and handle a broad spectrum of single objects. For example, they can put together a number of different goods for a customer order. This key function can also be implemented in the retail industry: For instance, robot systems are capable of conducting inventories, as well as continuously checking that goods are sorted according to store requirements and that sufficient quantities are on the shelves.
Expertise and sample projects
Fraunhofer IPA develops system solutions for retail logistics that are tailored to customer needs. These automation solutions include processes such as grasping, transporting and packaging goods. Depending on requirements, the institute develops complete systems as well as the various core technologies, such as navigation, recognition or mobile manipulation for the scenario concerned.
The diverse robot technologies offered by Fraunhofer IPA are suitable for a wide range of retail logistics applications. Starting with free navigation, automated guided vehicles (AGV) can perform transport processes autonomously, bringing goods to picking stations, for example. The advantage of IPA navigation software: elaborately-planned infrastructures in the operational environment that were often necessary in the past are no longer required. Instead, transport pathways are based on sensor information, making them highly-accurate and flexibly adaptable at the same time.
In order for robots to recognize individual products as well as standardized objects such as pallets or boxes, the scientists at IPA have developed a technology for 3-D object recognition. This allows new products to be learned, classified and recognized quickly by robot systems. Since the system compares color, shape and texture features, it is also capable of distinguishing between similar objects. It is furthermore not a problem if light conditions change or if products are partially covered. The system also calculates the direct gripping options it could use to pick up a target object with the gripper concerned.
Calculating gripping movements is a must when it comes to manipulation. A mobile manipulator, i.e. a combination of AGV and robot arm, not only picks up and transports goods but also packs customer orders. Fraunhofer IPA has developed a flexible software environment which combines algorithms for kinematic calculations with different components to achieve collision-free manipulation.
Besides the application in warehouses, these technologies can also be implemented in retail logistics: For a customer, the experts have developed a system for checking that shelves are fully stocked with the correct items.
The applications mentioned can be integrated locally or via an in-house or external cloud architecture: For instance, this would make new objects centrally available once they have been learned by the object recognition system. In the same way, all AGVs could be navigated cooperatively by analyzing the sensor data.