Vision-based Strategies Research Articles

Automated potholes detection using vibration and vision-based techniques

Roads serve as vital parts of our infrastructure, providing as crucial conduits for people's mobility and connectivity. However, the growing number of vehicles on the road has resulted in an increase in pavement strain and degradation, which has a substantial impact on the entire riding experience. To achieve a high-quality surface, roadways must be consistently monitored and maintained. In recent years, transportation infrastructure agencies and governments have shown a rising interest in leveraging new technologies to monitor road pavements. This interest derives from the difficult and time-consuming nature of manual and instrumented techniques. Automated technologies have arisen as a response to these issues, notably in recognizing pavement deterioration, such as the common problem of potholes. The objective of this research is to identify potholes using two low-cost automated techniques: a vibration-based method that uses the G-Sensor Logger application and a vision-based way that uses image processing. On the same roads, both approaches were employed and compared, with manual surveying utilized to validate the results. The results showed that vision-based strategies were more effective than vibration-based methods. Finally, although vibration-based analysis is appropriate for routine monitoring, vision-based analysis provides a more comprehensive and in-depth examination of road conditions. These discoveries will help future efforts to better monitor and maintain road surfaces, ensuring a smooth and safe travel experience for everybody.

Indoor fire detection utilizing computer vision-based strategies

Fires are an ever-increasing risk in the world for both indoor and outdoor environments. Current technologies for detection in indoor environments are smoke and flame detectors. However, these detectors have several limitations during both the ignition phase of a fire and propagation. These systems cannot detect an exact position of the fire nor how the fire is spreading, or its size, all of which is necessary information for fire services when dealing with these incidents. A potential solution is to use artificial intelligence techniques such as computer vision, which has shown the potential to detect and recognize objects and activities in indoor spaces. This study aims to develop a vision-based indoor fire and smoke detection system. Existing models based on the Faster R–CNN Inception V2 and the SSD MobileNet V2 models were explored and adopted in this work. This study utilized small training and testing datasets (for indoor specific fire cases) of varying pixel density images. Initial evaluation of the approach was carried out by testing both models on videos, including a mock-up bedroom and living room and a CCTV video of office space. Both high-density smoke environments and flame density scenarios were recognized. The promising results were achieved from using only 480 training images. Despite the success achieved by the Faster R–CNN Inception V2, the SSD MobileNet V2 model showed low accuracy and missed detection results. Future works can focus on integrating this with our previously developed approach, such as occupancy detection, enhancing training data and models, using more advanced detection models, and integrating the proposed approach with the fire fighting and HVAC control systems.

An image-based decoupling controller of quadrotor for moving target tracking

Different from the traditional controller system for quadrotor tasks, the vision-based strategies are more practical and powerful to execute more complex tasks, becoming more attractive to researchers. In this paper, an image-feature-based controller with states extracted in images directly is proposed for the quadrotor to track a moving underground target, in which suitable image features are defined and a coupling problem between position and attitude loop is solved by a decoupling algorithm. Moreover, the external disturbances caused by visual noise, wind, or other problems are eliminated by a robust observer with low-pass filters. A Lyapunov-based stability method is presented to prove the convergent properties of the system. Finally, a simulation using python3.6 with realistic images is established to verify that the system is stable and with high performance, the results of which show the data of unmanned aerial vehicle in moving target tracking.

A novel vision-based weakly supervised framework for autonomous yield estimation in agricultural applications

Autonomous systems have been established as a ground-breaking technology in agriculture, particularly for resource optimization and labor savings. However, even those solutions that are limited to monitoring activities, such as yield estimation, rely on costly robotic platforms equipped with a series of range devices (e.g., LIDAR and GPS-RTK). Recently, vision-based strategies have gained considerable attention as a less expensive and more efficient alternative, capable to be on par with or even surpass approaches that benefit from range sensors. Nonetheless, they exploit deep learning methodologies, which require burdensome labeling procedures to perform training. To address these shortcomings, we present a novel approach that performs yield estimation requiring only a monocular camera and needs a limited amount of supervision information. It detects, locates and maps fruits and tree canopies to estimate the total yield of a specific crop. To keep the image labeling effort to a minimum, we propose a weakly-supervision paradigm that only requires a simple binary label encoding the presence or the absence of fruits in the training images. Our approach does not make any assumptions on the underlying platform, i.e., it can be used by collecting images either with a hand-held camera or with an autonomous robot. Therefore, we are able to considerably reduce the deployment time, the energy and the cost of the overall yield estimation system. At the same time, we keep the performance comparable to both vision-based fully supervised baselines (which require costly labeling operations) and classical systems that rely on more expensive and power-demanding sensors.

A Novel Artificial Organic Control System for Mobile Robot Navigation in Assisted Living Using Vision- and Neural-Based Strategies.

Robots in assisted living (RAL) are an alternative to support families and professional caregivers with a wide range of possibilities to take care of elderly people. Navigation of mobile robots is a challenging problem due to the uncertainty and dynamics of environments found in the context of places for elderly. To accomplish this goal, the navigation system tries to replicate such a complicated process inspired on the perception and judgment of human beings. In this work, we propose a novel nature-inspired control system for mobile RAL navigation using an artificial organic controller enhanced with vision-based strategies such as Hermite optical flow (OF) and convolutional neural networks (CNNs). Particularly, the Hermite OF is employed for obstacle motion detection while CNNs are occupied for obstacle distance estimation. We train the CNN using OF visual features guided by ultrasonic sensor-based measures in a 3D scenario. Our application is oriented to avoid mobile and fixed obstacles using a monocular camera in a simulated environment. For the experiments, we use the robot simulator V-REP, which is an integrated development environment into a distributed control architecture. Security and smoothness metrics as well as quantitative evaluation are computed and analyzed. Results showed that the proposed method works successfully in simulation conditions.

Finite-Time LPV Analysis of a Vision Based Landing System with Anti-Windup Augmentation

Abstract In contrast to standard ILS-based landing control systems, vision-based strategies rely on specific measures that introduce non-linearities into the control laws. Moreover, during the landing phase high gains are often used to improve performances at the expense of stability which is then degraded. This is not much an issue since both final approach and flare segments are short-time maneuvers. However, high gains induce saturations which in turn lead to performances degradation. For these reasons, vision-based landing systems design and tuning require a specific attention which generally involves a time-consuming trial-and-error process. The latter could certainly be fasten and improved with the help of adapted analysis tools providing a reliable estimate of the finite-time performance level of the aforementioned saturated nonlinear closed-loop system. As observed in Biannic and Burlion (2017), it can be rewritten as a saturated Linear Parameter Varying (LPV) system for which many tools exist or can be extended. In this contribution, original discrete-time extensions are then proposed to characterize finite-time performance indexes for a vision-based landing system including an anti-windup device.

Veering in hemi-Parkinson’s disease: Primacy of visual over motor contributions

Veering while walking is often reported in individuals with Parkinson’s disease (PD), with potential mechanisms being vision-based (asymmetrical perception of the visual environment) or motoric (asymmetry in stride length between relatively affected and non-affected body side). We examined these competing hypotheses by assessing veering in 13 normal control participants (NC) and 20 non-demented individuals with PD: 9 with left-side onset of motor symptoms (LPD) and 11 with right-side onset (RPD). Participants walked in a corridor under three conditions: eyes-open, egocentric reference point (ECRP; walk toward a subjectively perceived center of a target at the end of the corridor), and vision-occluded. The visual hypothesis predicted that LPD participants would veer rightward, in line with their rightward visual-field bias, whereas those with RPD would veer leftward. The motor hypothesis predicted the opposite pattern of results, with veering toward the side with shorter stride length. Results supported the visual hypothesis. Under visual guidance, RPD participants significantly differed from NC, veering leftward despite a shorter right- than left-stride length, whereas LPD veered rightward (not significantly different from NC), despite shorter left- than right-stride length. LPD participants showed significantly reduced rightward veering and stride asymmetry when they walked in the presence of a visual landmark (ECRP) than in the eyes-open condition without a target. There were no group differences in veering in the vision-occluded condition. The findings suggest that interventions to correct walking abnormalities such as veering in PD should incorporate vision-based strategies rather than solely addressing motor asymmetries, and should be tailored to the distinctive navigational profiles of LPD and RPD.

A mini work-cell for handling and assembling microcomponents

Purpose – The purpose of this paper was the design, development, and test of a flexible and reconfigurable experimental setup for the automatic manipulation of microcomponents, enhanced by an accurately developed vision-based control. Design/methodology/approach – To achieve a flexible and reconfigurable system, an experimental setup based on 4 degrees of freedom robot and a two-camera vision system was designed. Vision-based strategies were adopted to suitably support the motion system in easily performing precise manipulation operations. A portable and flexible program, incorporating the machine vision module and the control module of the task operation, was developed. Non-conventional calibration strategies were also conceived for the complete calibration of the work-cell. The developed setup was tested and exploited in the execution of repetitive tests of the grasping and releasing of microcomponents, testing also different grasping and releasing strategies. Findings – The system showed its ability in automatically manipulating microcomponents with two different types of vacuum grippers. The performed tests evaluated the success and precision of the part grasping and release, which is a crucial aspect of micromanipulation. The results confirm reliability in grasping and that the release is precluded by adhesive effects. Thus, different strategies were adopted to improve the efficiency in the release of stuck components without negatively affecting the accuracy nor the repeatability of the positioning. Originality/value – This work provided a flexible and reconfigurable architecture devoted to the automatic manipulation of microcomponents, methodologies for the characterization of different vacuum microgrippers, and quantitative information about their performance, to date missing in literature.

Design and Characterization of a Vision-Assisted System for Surface-Mount Component Placement

The accurate placement of surface-mount components on their respective pads is an important problem in surface-mount technology. As the lead/pad pitch size decreases, the placement problem becomes more difficult and a vision system is required to provide error information necessary to implement an accurate placement. Most vision-based placement strategies described in the open literature use component-to-pad alignment techniques unable to compensate for local errors in the component or the pad. This paper describes the development of a lead-to-pad placement strategy able to compensate for global and local errors. Sharp images with good illumination techniques reduce the amount of image processing, and simulation results show that lead-to-pad matching performs significantly better than component-to-pad matching.