This paper aims at a preliminary research on designing a hybrid mechanism inspired by human leg's musculoskeletal architecture and locomotion capabilities. Complexity of the workspace solution in serial-parallel m...
详细信息
In a fast-changing world we are in today, unmanned vehicles are displacing old, obsolete and frustrating tasks. Since unmanned vehicles are intended to work in an environment without any conduction, finding a collisio...
详细信息
The main objective of a hand-geometry-based user identification model consists in designing a method that could be implemented with a satisfactory trade-off between cost, accuracy, and ease of utilizing. In this paper...
详细信息
ISBN:
(纸本)9781665480871
The main objective of a hand-geometry-based user identification model consists in designing a method that could be implemented with a satisfactory trade-off between cost, accuracy, and ease of utilizing. In this paper, an efficient, peg-free hand geometry-based approach for user identification is presented, which can be easily implemented with low-cost pieces of equipment and results up to a 98.7% accuracy in user identification. Three different feature-based methods, namely, circular representation, FPL, and FPLKW, are designed and examined. In order to extract required landmarks to the end of generating intended features, the so-called Media-Pipe is used. The first method is the circular representation, in which the radius ratio of the circles formed by the three middle finger straps is taken as model features. The second method extracts fingers length and phalanges length of each finger and uses their ratio as features, which is called Finger Phalanges Length (FPL) method. The third method extracts Fingers Length and Phalanges Length of each finger and Knuckles Width, abbreviated as (FPLKW), and uses their ratios for extracting features. Two datasets are collected, one for examining the efficiency of the proposed methods and finding the promising one, and the second one for validating the performance of the selected method. The first dataset contains images of the back-side of both hands in spread fingers mode and picked altogether fingers mode, taken from 21 individuals on a black background. The second dataset contains images of 100 individual hands in the front-side of both hands and the back-side of both hands in spread fingers mode. Different classifications are used to obtain maximum accuracy, including SVM SVC, SVM NuSVC, the decision tree classifier, the extra tree classifier, the random forest classifier, and one hidden layer MLP. Obtained results reveal that the extra tree classifier leads to better performance with respect to the methods mentioned abov
In this paper, a new publicly available 1 web-Scraped Iranian Vehicle Dataset (SIVD) for simultaneous real-time vehicle tracking and recognition is proposed. The datasets provided for Iranian cars in the literature ha...
详细信息
In the pursuit of developing autonomous agents capable of learning optimal behaviors, reinforcement learning has emerged as a foundational frame-work. This paper introduces a novel approach for state definition in dee...
详细信息
In this research, a nonlinear model predictive controller is designed for the trajectory tracking of spatial cable-suspended parallel robots with four cables. The dynamic model of this robot is derived firstly using t...
详细信息
Today, system identification plays a pivotal role in control science and offering a myriad of applications. This paper places its focus on the identification of actuator models within real-world delta robots for infor...
详细信息
In response to the increasing supply and demand in the food industry, automated food packaging has emerged as a substitute for the manual labor that was previously required. The surge in automation is evident, with ro...
详细信息
ISBN:
(数字)9798350376340
ISBN:
(纸本)9798350376357
In response to the increasing supply and demand in the food industry, automated food packaging has emerged as a substitute for the manual labor that was previously required. The surge in automation is evident, with robotics playing a pivotal role in tasks ranging from sorting and packaging to labeling, and notably, pick-and-place operations which is the specific focus of this study. Leveraging robots for these tasks not only improves precision and speed but also takes packaging hygiene into account, resulting in a substantial cost reduction and heightened efficiency. Delta Parallel robots are among the robotic systems well-suited for high-speed pick-and-place tasks. In this experimental study, the focus is on packaging several scattered chocolates, where both the container box and the chocolates will be randomly positioned and oriented. This is achieved by employing a Delta Parallel robot paired with a two-fingered gripper, designed to incur minimal damage to the product when compared to other types of end-effectors. The positioning of the box and products will be determined through a classical image processing approach, specifically Edge Detection and Hough Transform, implemented using the OpenCV Python library. Following identification, the objects are localized through camera-robot calibration, and the resulting coordinates are transmitted to the robot as target points. Extensive testing has demonstrated that this study, employing visual perception for pick-and-place tasks, achieved a success rate of 82%. This study showcases the potential of automating packaging tasks in the food industry through the integration of robotic systems and artificial intelligence. The approach proves effective across diverse tasks, encompassing the packaging of variously shaped food items and drugs, applicable across different industries.
Grasping is a fundamental aspect of humaninteraction with the environment, and replicating this ability in robotic systems is challenging due to the complexity of hand-object interactions. This paper explores the map...
详细信息
ISBN:
(数字)9798350376340
ISBN:
(纸本)9798350376357
Grasping is a fundamental aspect of humaninteraction with the environment, and replicating this ability in robotic systems is challenging due to the complexity of hand-object interactions. This paper explores the mapping of human grasping behavior to a 3-finger robotic gripper using a deep learning approach. The study focuses on the development of a model that predicts grasping points based on RGB images of objects and finger coordinates during the approaching state. The research contributes a novel dataset of human-like grasping with 3 fingers, capturing various object orientations and scenarios. Finger coordinates are extracted in both the approaching and grasping states by using Mediapipe. The proposed model employs a pretrained Faster R-CNN for object detection, coupled with fully connected networks for regression of grasping points. Evaluation metrics include Intersection over Union (IoU), grasp angle, and a proposed metric, Grasp Configurational Accuracy (GCA). The trained network achieves compelling results, demonstrating 88.2% GCA, 96.5% in grasp angle, and a 45.7% IoU in grasp rectangle representation. Results indicate the efficacy of the approach, offering a promising avenue for advancing robotic grasping capabilities.
This paper uses reinforcement learning techniques to introduce a novel approach to the controller design for a wide range of serial chain robots and cable-driven serial chain robots. The method addresses the regulatio...
详细信息
暂无评论