One of the most frequently used approaches to represent collaborative mapping are probabilistic occupancy grid maps. These maps can be exchanged and integrated among robots to reduce the overall exploration time, which is the main advantage of the collaborative systems. Such map fusion requires solving the unknown initial correspondence problem. This article presents an effective feature-based map fusion approach that includes processing the spatial occupancy probabilities and detecting features based on locally adaptive nonlinear diffusion filtering. We also present a procedure to verify and accept the correct transformation to avoid ambiguous map merging. Further, a global grid fusion strategy based on the Bayesian inference, which is independent of the order of merging, is also provided. It is shown that the presented method is suitable for identifying geometrically consistent features across various mapping conditions, such as low overlapping and different grid resolutions. We also present the results based on hierarchical map fusion to merge six individual maps at once in order to constrict a consistent global map for SLAM.

The primary goal of this research is to develop a deep learning-powered robotic solution to address labor shortages and optimize harvesting processes in strawberry greenhouse farms. By incorporating this system into the development process, the aim is to provide continuous, 24/7 operational efficiency for strawberry harvesting in greenhouse environments. This study is grounded in a comprehensive literature review of simulated environments, such as ROS-Gazebo, deep learning detection models, and mobile robot platforms, with a focus on developing innovative robotic solutions for strawberry detection and grasping in a simulated digital twin greenhouse environment. The YOLOv9-GLEAN deep learning model, with super-resolution capabilities, is introduced to enhance strawberry detection accuracy by generating high-resolution image features. We developed a digital twin model of the SILAL strawberry greenhouse farm in Abu Dhabi, UAE, within the ROS-Gazebo environment, to validate our algorithm and test the MARTA (Mobile Autonomous Robot with Telescopic Arm) robot. The dataset used to improve model performance includes both real strawberry images from greenhouse farm and synthetic CAD-generated images. ROS-MoveIt was employed to implement visual servoing, allowing the robot to generate precise motion trajectories to approach and grasp identified strawberries, with visual feedback enhancing accuracy. Empirical results show that our proposed detection model outperforms other existing models, achieving a precision of 0.996 and a recall of 0.991. The model’s adaptability to varied datasets, including real and synthetic images, is notable, and it performs exceptionally well in the simulated digital twin model of the greenhouse farm. The model is uniquely trained on both real and synthetic strawberry images to ensure robust detection performance. It is compared to state-of-the-art models and deployed on a telescopic arm-based robotic platform, which is simpler to control than an articulated arm for strawberry harvesting and grasping tasks.

Human–robot interaction is of the utmost importance as it enables seamless collaboration and communication between humans and robots, leading to enhanced productivity and efficiency. It involves gathering data from humans, transmitting the data to a robot for execution, and providing feedback to the human. To perform complex tasks, such as robotic grasping and manipulation, which require both human intelligence and robotic capabilities, effective interaction modes are required. To address this issue, we use a wearable glove to collect relevant data from a human demonstrator for improved human–robot interaction. Accelerometer, pressure, and flexi sensors were embedded in the wearable glove to measure motion and force information for handling objects of different sizes, materials, and conditions. A machine learning algorithm is proposed to recognize grasp orientation and position, based on the multi-sensor fusion method.

In this work, we propose a novel approach for tomato pollination that utilizes visual servo control. The objective is to meet the growing demand for automated robotic pollinators to overcome the decline in bee populations. Our approach focuses on addressing this challenge by leveraging visual servo control to guide the pollination process. The proposed method leverages deep learning to estimate the orientations and depth of detected flower, incorporating CAD-based synthetic images to ensure dataset diversity. By utilizing a 3D camera, the system accurately estimates flower depth information for visual servoing. The robustness of the approach is validated through experiments conducted in a laboratory environment with a 3D printed tomato flower plant. The results demonstrate a high detection rate, with a mean average precision of 91.2 %. Furthermore, the average depth error for accurately localizing the pollination target is impressively minimal, measuring only 1.1 cm. This research presents a promising solution for tomato pollination, showcasing the effectiveness of visual-guided servo control and its potential to address the challenges posed by diminishing bee populations in greenhouses.

This paper deals with the fault detection, isolation (FDI) and reconfiguration of the locomotion of a bipedal-legged robot. Initially, the planar model of the legged robot in the vertical plane is developed using a bond graph (BG) approach. Then, the planar BG model of the legged robot is extended to the three-dimensional legged robot. Two individual motors are used to actuate the prismatic leg of the robot for locomotion. The BG simulation provides results for straight walking based on an oscillating cylinder mechanism and the turning motion of the legged robot are discussed. The prototype model of the legged robot is also developed and experimentation is done for straight and inclined plane applications. Finally, an FDI technique for the three-dimensional model of a legged robot is developed for the generation of fault indicators (i.e., analytical redundancy relations; ARRs) in the presence of system failure. The ARRs are derived from the BG model of the legged robot during the occurrences of the fault. The experimental results are validated with the simulation results for FDI and reconfiguration when the robot manoeuvres in a U-shaped path. The real-time fault diagnosis and reconfiguration for locomotion of the legged robot is possible with this FDI approach.

The development of lightweight, durable composites for industrial use is constrained by traditional tribological evaluation methods that are costly, time-consuming, and inadequate for capturing nonlinear interactions between material and operational parameters. This study proposes an integrated framework combining Global Sensitivity Analysis (GSA) and Machine Learning (ML) to predict the coefficient of friction (COF) and wear loss in Al7075/B4C nanocomposites. Four GSA techniques - Sobol indices, delta index, PAWN index, and mutual information - were employed to rank the significance of input parameters, including applied load, B4C reinforcement percentage, time, sliding velocity, and sliding distance. Using 10,800 experimental records from pin-on-disc tests, a Deep Residual Regression Network (DRRN) was developed to model tribological behavior. The Al7075 matrix was reinforced with boron carbide (B4C) particles at weight fractions of 0%, 4%, 8%, and 12%. Results show that B4C reinforcement significantly enhances wear resistance, with the 12% B4C composite reducing wear loss by 74% and COF by 20% under severe conditions. The proposed framework achieved high predictive accuracy (R2 = 0.93 for COF, 0.99 for wear loss).