This paper explores the automated detection of surface defects on 3-D printed products and concrete structures. They are the main factors to evaluate their quality in addition to dimension and roughness. Traditional detection by human inspectors is far from satisfactory. Manual inspection is time-consuming, error-prone and often leads to loss of resources. For this purpose, image processing and deep learning-based object detection adopted by Google Cloud Machine Learning (ML) Engine is used to detect surface defects. In the case of image processing, two approaches are presented in this paper. In both cases, pixels are being considered to differentiate a smooth or rough surface from a picture taken by a USB camera. For the deep learning-based solution, MobileNet -a base convolution neural network treated as an image feature extractor in combination with Single Shot MultiBox Detector (SSD) as an object detector hence MobileNet-SSD. The model was successfully trained on the Google Cloud ML Engine with the dataset of 20000+ images. The review of the results confirms that with the help of MobileNet-SSD can automatically detect surface defects more accurately and rapidly than conventional deep learning methods.
Autonomous ground robots autonomously are being used in the places where it is very hazardous for human beings to reach and operate, such as nuclear power plants and chemical industries. The aim of the research presented here is to develop a control system that enables such ground robots navigate autonomously with various sensors as the depth camera, 2D scanning laser, 3D Lidar, GPS, and IMU. The controller uses the current position measured using the sensors on the Husky A200, given the waypoints of the destination. Then it calculates the best possible route based on the recent events provided using IMU data and GPS. The Model Predictive Control (MPC) improves the robot's motion, by using a path planner for the robot's trajectory generation. The use of global reference frame waypoints is planned to create the appropriate path and the actions required to follow the motion planner's direction. The path planner depends on the active sensor data such as locations and size of obstacles. Then, a feasible path is generated based on the sensor data. The desired trajectory consists of a set of waypoints fit in a 3 rd -order polynomial. They determine the path's feasibility for the ground robot's dynamics and a series of points generated with a certain velocity and acceleration profile. The MPC adjusts the robot's lateral, longitudinal, yaw motions and approximates a continuous trajectory with discrete paths to command behaviors. The kinematic model of a robot, Husky is used as the dynamic model for transient and steady-state characteristics. The camera captures the images and other types of data processed through the computational framework used to build machine learning models. TensorFlow is used for deep learning and to identify and classify various objects around the Husky. This research has limitations such as using the linear dynamic model as the LQR method. Also on vehicle models, the vehicle model considered in this research considers a constant value to describe the slope in the most linear region. Detailed discussion on MPC development with a major system design factor has been emphasized with logical steps in MPC.
Researchers around the world are developing autonomous robot systems to assist humans with tedious, hazardous, complex, and timeconsuming tasks. Today, autonomous robots are used for a variety of applications, ranging from delivering food to operating on patients. A robotic system can be operated with or without human assistance, depending on whether it is manned or unmanned. Additionally, the robot may be classified as an Unmanned Aerial Vehicle (UAV), Unmanned Surface Vehicle (USV), or Unmanned Ground Vehicle (UGV), depending on its operational conditions. An autonomous robot may be equipped with manipulators and advanced sensors, such as a LiDAR camera, which can collect 3D spatial data. The present study reviews recent research and development of unmanned ground vehicles, including two nature-inspired designed robots called SPOT and DIGIT and presents its technical data. In addition, this study introduces the challenges and applications of these robots. A novel type of comparison was made in this study between some UGVs, human and animalinspired designed robots, where researchers were able to get an idea of different robot basic specifications that could be used to select these robots for future research. As a result of this study, two more studies will be conducted in the future.
Unmanned Ground Vehicles (UGVs) have, as of late, been utilized in a wide assortment of utilizations because of their flexibility, diminished expense, and quick response, among other benefits. Search and Rescue (SAR) is quite possibly the most conspicuous zones for the work of UGVs instead of a monitored mission, mainly due to its impediments on the expenses, human resources, and view of the human administrators. An ongoing way of arranging to utilize numerous helpful UGVs for the SAR mission is proposed in this study. This study aims to introduce the initial moves towards a Model Predictive Control (MPC) based peril evasion calculation for UGVs representing the vehicle elements through high constancy models and uses just surrounding data about the environment as given by the available onboard sensors. In particular, the paper presents the MPC definition for peril evasion utilizing a Light Detection and Ranging (LiDAR) sensor and applies it to a contextual of the effect of model constancy on the calculation's presentation, where execution is estimated principally when to arrive at the objective point. The Robot Operating System (ROS) is used to drive the sensors and visualize the data in RVIZ. This study presents MPC development for navigating Husky A200 by adjusting the longitudinal, lateral, and yaw motion command behaviors. The proposed algorithm for Husky A200 is tested indoors and compared the results with the simulation results plotted using MATLAB and GAZEBO. A novel simulator package is developed for the Husky using RVIZ and GAZEBO. The efficiency of the proposed MPC design is tested through simulation and compared with real world experiments, the real-time longitudinal movement follows the simulation results closely. For MPC's short-term optimization, an optimized control signal from a linear framework is utilized for a linear quadratic controller. According to the Husky position and orientation, applying a transformation to convert the map coordinate system to the Husky coordinate system. Transforming the map coordinate system helped in computing the errors because the initial vector considers position and orientation as zero.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.