This paper presents a novel approach for detection and localization of standardized euro pallets, which are orientated up to 90° in relation to the sensor plane. There is no a priori information about the pallets pose needed. We use a time-of-flight camera. Our algorithm is based on finding surfaces in the point cloud, which represent the three wooden blocks of a euro pallet. Different kinds of geometrical checks set up our detection pipeline, where no artificial markers on the pallets are needed. Since we perform the detection while driving a forklift, the algorithm must process the point cloud within a set time limit. The detection and localization result in the pallets position and orientation in relation to the camera coordinate system. This information can be provided to higher-level systems, like advanced driver assistance systems. The results show that the localization of pallets is possible in the scenario considered.

The increasing demand for flexibility in automated production systems also affects the automated material flow systems (aMFSs) within these systems and, thus, demands reconfigurable systems. However, the centralized control concept usually applied in aMFSs hinders easy adaptation, as the entire control software has to be retested when subparts of the control are manually changed. As adaption and subsequent testing are time-consuming tasks, concepts are required for splitting the control from one centralized node to multiple, decentralized control nodes. Therefore, this article presents a holistic, agent-based control concept for aMFSs, whereby the system is divided into so-called automated material flow modules (aMFMs), each controlled by a dedicated module agent. The concept allows reconfiguring an aMFS consisting of heterogeneous, stationary aMFMs, during runtime. Furthermore, it includes aspects such as uniform agent knowledge bases through metamodel-based development, a communication ontology considering different information types and properties, strategic route optimization in decentralized control architectures, and a visualization concept to make decisions of the module agents comprehensible to operators and maintenance staff. We performed the concept evaluation using material flow simulations and a prototypical implementation on a lab-sized demonstrator.

As in manufacturing with its Industry 4.0 transformation, the enormous potential of artificial intelligence (AI) is also being recognized in the construction industry. Specifically, the equipment-intensive construction industry can benefit from using AI. AI applications can leverage the data recorded by the numerous sensors on machines and mirror them in a digital twin. Analyzing the digital twin can help optimize processes on the construction site and increase productivity. We present a case from special foundation engineering: the machine production of bored piles. We introduce a hierarchical classification for activity recognition and apply a hybrid deep learning model based on convolutional and recurrent neural networks. Then, based on the results from the activity detection, we use discrete-event simulation to predict construction progress. We highlight the difficulty of defining the appropriate modeling granularity. While activity detection requires equipment movement, simulation requires knowledge of the production flow. Therefore, we present a flow-based production model that can be captured in a modularized process catalog. Overall, this paper aims to illustrate modeling using digital-twin technologies to increase construction process improvement in practice.