Urban traffic management (UTM) relies on accurate vehicular flow data to optimize infrastructure and reduce congestion. However, existing video-based methods struggle with occlusions and complex trajectories in multi-directional intersections, limiting their applicability. This paper presents a novel Speed Calculation and Trajectory Observation Algorithm (SCTOA) for robust vehicle counting, classification, and speed estimation in visually obstructed urban environments (e.g., tunnels, overpasses). Our approach integrates drone-captured aerial footage with YOLOv8 detection, ByteTrack tracking, and LoFTR-based motion compensation. To address occlusion challenges, we propose strategically placed Observation Areas (OAs), in points of traffic bifurcation, improving the overall performance. By mapping vehicle trajectories between OAs we achieve up to 96.73% accuracy on vehicle counting in unobstructed zones and 87.8% in complex scenarios, —outperforming prior single-direction methods. Experimental validation across 27 hours of drone footage from high-traffic intersections demonstrates the efficacy of the method for multi-directional flow analysis. The results enable precise input for traffic simulators (e.g., PTV-Vissim), supporting data-driven UTM decisions while minimizing costly real-world experimentation.

Traditionally, manufacturing systems that employ industrial robots restrict the workspace of the robot with the objective of avoiding collisions with humans or other obstacles during operations. A recent trend in manufacturing is to design even more flexible manufacturing systems to adapt quickly to market demand. This manufacturing tendency leads to a system in which humans and robots collaborate, sharing the tasks and workspace. This type of hybrid manufacturing combines human and robot abilities and is based on using the advantages of each participant, focusing on the tasks each element is superior at. A human being, for instance, has superior sensorimotor skills to those of the robot in complex manipulation tasks, but does not possess large force capabilities. On the other hand, the robot presents superior abilities in repetitive tasks, like precision and repetitiveness. In this paper, a manufacturing framework to manipulate irregular products between a human operator and a robot is presented. The objective is to allow the robot and the human to interact in a common task sharing the same workspace, eliminating traditional physical barriers. The development of this collaborative module involves the design of strategies for grasping and kitting operations of irregular products that also includes the recognition of those products. The experimental platform is composed by a 12-camera motion capture system, KUKA KR60 industrial robot, which is equipped with an end-effector with deformable fingers and an RGBD sensor (depth sensing devices that work in association with an RGB camera). A lexicon based on some human gestures has been designed and tested on-line during human-robot interaction tasks. Results showed the usefulness of the method to be implemented with real industrial robots and further work has been envisaged related to security issues that have to be met during its deployment in other similar applications.

In this paper we propose a novel set of techniques for real-time Multi-Target Multi-Camera (MTMC) tracking of vehicles in congested, low speed environments, such as those of drive-thru scenarios, where metrics such as the number of vehicles, time of stay, and interactions between vehicles and staff are needed and must be highly accurate. Traditional methods of tracking based on Intersection over Union (IoU) and basic appearance features produce fragmented trajectories of misidentifications under these conditions. Furthermore, detectors, such as YOLO (You Only Look Once) architectures, exhibit different types of errors due to vehicle proximity, lane changes, and occlusions. Our methodology introduces a new tracker algorithm, Multi-Object Tracker based on Corner Displacement (MTCD), that improves the robustness against bounding box deformations by analysing corner displacement patterns and several other factors involved. The proposed solution was validated on real-world drive-thru footage, outperforming standard IoU-based trackers like Nvidia Discriminative Correlation Filter (NvDCF) tracker. By maintaining accurate cross-camera trajectories, our framework enables the extraction of critical operational metrics, including vehicle dwell times and person–vehicle interaction patterns, which are essential for optimizing service efficiency. This study tackles persistent tracking challenges in constrained environments, showcasing practical applications for real-world surveillance and logistics systems where precision is critical. The findings underscore the benefits of incorporating geometric resilience and delayed decision-making into MTMC architectures. Furthermore, our approach offers the advantage of seamless integration with existing camera infrastructure, eliminating the need for new deployments.

Reinforcement Learning has become a very powerful technique in the last decade thanks to the technological advances in computational power and efficient learning algorithms in neuronal networks. Reinforcement Learning allows an autonomous agent to learn through experience how to solve a problem in an optimal way, with a minimal information of its environment, just a reward signal. This approach has lightened the way to an artificial intelligence capable of reaching superhuman performance in games such as Go and Backgammon, beating world champion players. Unfortunately this is far from a real world robot application due to many technological challenges. So in this paper we propose how to get a low-dimensional state representation for industrial robot manipulators, using visual information and Convolutional Autoencoders to speed up the information processing in the Reinforcement Learning training phase.

Industrial production systems for smart factories or the so-called Industry 4.0 will demand high interoperability and connectivity between production modules, so that modules could be monitored in real-time. Production modules should make decisions on their own, with no human intervention, they must be modular and adaptable to changing circumstances and customers' requirements. The autonomous operation of production modules in smart factories imposes asynchronous delays due to several reasons. For instance, object recognition time, grasping time or welding delays that change due to a newly reoriented or positioned component. Consequently, production modules need to be speeded up to compensate for delays in previous production stages. In this paper, we present a novel Distributed Controller that autonomously compensate production delays. The proposed controller compensates for major production delays that affect the total production time of a product with the objective of maintaining the mean production time. Experiments are reported in this paper which show the feasibility of the algorithm working in a real manufacturing scenario.