A nondestructive testing method for defects of road and bridge concrete

Abstract

The application discloses the technical field of concrete detection and particularly relates to a road and bridge concrete defect nondestructive detection method, which comprises the following steps: step one, multi-modal synchronous acquisition and data set construction; step two, multi-modal space-time registration; step three, defect feature enhancement and cross-modal fusion; step four, defect intelligent identification, recording and key marking; step five, three-dimensional tomographic quantification, safety classification and report generation; and step six, cyclic reexamination and rectification progress supervision. The application has the beneficial effects that the problems of weak deep defect detection, isolated data and difficult fusion are solved, the identification accuracy is improved, defect visualization and quantitative analysis are realized, cyclic reexamination and rectification progress supervision are supported, closed-loop management is formed, the problem that rectification cannot be continuously monitored is solved, full-automatic operation is achieved, the anti-interference is strong, the misjudgment rate is low, and the application is suitable for long-term and normalized detection and operation of road and bridge concrete.

Description

Technical Field

[0001] This invention relates to the field of concrete testing technology, specifically a non-destructive testing method for defects in road and bridge concrete. Background Technology

[0002] Road and bridge concrete is a type of load-bearing structural concrete specifically used for transportation infrastructure such as roads, bridges, culverts, and ramps. It is made by mixing cement, sand, aggregate, water, and admixtures in a specific ratio and has the characteristics of high strength, high durability, frost resistance, impermeability, and resistance to load fatigue. It mainly bears vehicle loads, structural self-weight, and external environmental effects, and is the core load-bearing material in road and bridge engineering.

[0003] During the long-term use of road and bridge projects, concrete structures are prone to defects such as voids, honeycombing, cracks, interlayer voids, and incomplete grouting of prestressed ducts. The presence of defects such as cracks, voids, honeycombing, and voids in road and bridge concrete will directly affect the safety and service life of the road and bridge, thus requiring professional non-destructive testing and maintenance.

[0004] However, existing technologies for non-destructive testing of concrete defects in roads and bridges have weak capabilities for detecting deep defects, making it impossible to achieve accurate quantification and three-dimensional imaging. The test data is isolated and cannot be effectively integrated, resulting in a high rate of false positives. Furthermore, there is a lack of standardized recording, key marking, and follow-up re-inspection mechanisms after testing, making it impossible to continuously monitor the rectification of defects and to grasp the progress and effectiveness of the rectification.

[0005] Therefore, we propose a non-destructive testing method for defects in road and bridge concrete. Summary of the Invention

[0006] In view of the problems existing in the above and / or existing non-destructive testing methods for defects in road and bridge concrete, the present invention is proposed.

[0007] Therefore, the purpose of this invention is to provide a non-destructive testing method for defects in road and bridge concrete. By improving the existing non-destructive testing methods for road and bridge concrete, this invention can solve the problems mentioned above in the prior art, such as weak detection capability for deep defects, inability to achieve accurate quantification and three-dimensional imaging, isolated and ineffective data fusion, high defect misjudgment rate, lack of standardized recording, key marking and follow-up re-inspection mechanism after detection, inability to continuously monitor the defect rectification status, and difficulty in grasping the rectification progress and repair effect.

[0008] To address the aforementioned technical problems, according to one aspect of the present invention, the present invention provides the following technical solution:

[0009] A non-destructive testing method for defects in concrete used in road and bridge construction includes the following steps:

[0010] Step 1: Multimodal synchronous acquisition and dataset construction. A unified detection grid is divided on the concrete surface of the road and bridge. Impact echo signals, ultrasonic transmission signals and infrared thermal imaging sequences are synchronously acquired at the same grid point to construct a time-space-frequency three-dimensional multimodal dataset.

[0011] Step 2: Multimodal spatiotemporal registration. The three-dimensional multimodal dataset is time-stamp aligned and spatial coordinates are normalized to eliminate positional offsets and time delays between signals.

[0012] Step 3: Defect feature enhancement and cross-modal fusion. Defect features are enhanced and extracted for the three types of signals respectively. Then, the waveform features, propagation features and temperature features are correlated and encoded through the Transformer fusion network to generate a unified defect feature vector.

[0013] Step 4: Intelligent Defect Identification, Recording, and Highlighting. Based on the defect feature vector, a deep learning model is used to complete defect identification and classification, outputting the defect type, confidence level, and boundary range, and recording and highlighting the defect location and condition in detail.

[0014] Step 5: 3D tomography quantification, safety classification and report generation. Based on the fusion feature inversion, a 3D grayscale cloud map of the defect is constructed. The defect depth, area, volume and equivalent diameter are automatically calculated. The safety classification is performed according to the defect size and location and an inspection report is generated.

[0015] Step Six: Cyclic Re-inspection and Rectification Progress Monitoring. Cyclic re-inspection is carried out on the tested road and bridge concrete areas. During the re-inspection, priority is given to checking the defective parts that were marked in the early stage. The rectification status is compared and analyzed to achieve real-time monitoring of the rectification construction progress.

[0016] As a preferred embodiment of the non-destructive testing method for road and bridge concrete defects described in this invention, in step four, when performing intelligent defect identification and classification based on defect feature vectors using a deep learning model, five types of defects inside the concrete—voids, honeycombs, cracks, interlayer voids, and incomplete grouting of prestressed ducts—are accurately determined in terms of type, confidence score, and boundary contour location. The spatial coordinates of the defects, their grid number, defect type, confidence score, boundary range, and initial depth value are simultaneously written into the defect database. At the same time, the defect areas are highlighted with color grading in the detection grid planar diagram, generating a distribution map of key defect markers with spatial coordinates.

[0017] In step six, when performing cyclical re-inspection on the inspected road and bridge concrete areas, the system automatically retrieves the defect database and key defect marker distribution map from the initial inspection stage. Based on the marker priority, it performs directional re-measurement and data collection on the defect grid points. The quantitative parameters of defect depth, area, volume, and equivalent diameter obtained from the re-measurement are compared with the initial inspection data, and trend analysis is performed. This automatically determines whether the defect rectification is qualified, whether the repair is in place, and whether the defect has re-developed. Furthermore, it generates a rectification progress monitoring report in real time, including the rectification qualification rate, repair effect, and development trend, thus achieving closed-loop monitoring of the entire road and bridge concrete defect repair construction process.

[0018] As a preferred embodiment of the non-destructive testing method for defects in road and bridge concrete described in this invention, the impact echo signal, ultrasonic transmission signal, and infrared thermal imaging sequence in step one are synchronously triggered and acquired using the same timestamp to ensure that each modal signal remains consistent in time and space.

[0019] As a preferred embodiment of the non-destructive testing method for defects in road and bridge concrete described in this invention, the spatiotemporal registration in step two includes normalizing the coordinates of the detection grid and aligning the signal sampling points in time to eliminate positional deviations and delay errors between multimodal data.

[0020] As a preferred embodiment of the non-destructive testing method for defects in road and bridge concrete described in this invention, the defect feature enhancement and extraction of the three types of signals in step three includes: wavelet denoising and resonance peak extraction of the impact echo signal, travel time correction and attenuation compensation of the ultrasonic transmission signal, and gradient enhancement and abnormal region segmentation of the infrared thermal imaging sequence.

[0021] As a preferred embodiment of the non-destructive testing method for defects in road and bridge concrete described in this invention, in step three, the Transformer fusion network performs global correlation encoding on waveform features, propagation features, and temperature features through a self-attention mechanism, thereby suppressing environmental noise and steel reinforcement interference and enhancing weak defect features.

[0022] As a preferred embodiment of the non-destructive testing method for defects in road and bridge concrete described in this invention, the three-dimensional grayscale cloud map in step five is generated based on multi-modal fusion feature inversion, which can intuitively reflect the spatial morphology, distribution location and development trend of defects inside the concrete.

[0023] As a preferred embodiment of the non-destructive testing method for defects in road and bridge concrete described in this invention, the safety classification in step five is automatically divided into four levels: I, II, III, and IV, based on the depth, area, volume, and stress region of the defect, and corresponding handling suggestions are given.

[0024] As a preferred embodiment of the non-destructive testing method for defects in road and bridge concrete described in this invention, the cycle of cyclic re-inspection in step six is ​​automatically set according to the defect safety level, and the higher the safety level of the defect area, the higher the re-inspection frequency.

[0025] As a preferred embodiment of the non-destructive testing method for defects in road and bridge concrete described in this invention, the testing method is fully automated and can complete the entire process of data acquisition, defect identification, quantitative analysis, safety rating, re-inspection supervision and report generation without human intervention.

[0026] Compared with the prior art, the beneficial effects achieved by the present invention are:

[0027] 1. This invention solves the problems of weak detection of deep defects and difficulty in data fusion by using three-modal fusion detection of impact echo, ultrasonic transmission, and infrared thermal imaging, thereby improving the accuracy of identification. It adopts three-dimensional tomographic imaging and precise quantification to realize defect visualization and quantitative analysis, making up for the shortcomings of traditional methods in that they cannot perform quantitative imaging. Through automatic identification, marking, database construction and tracking, it improves the defect recording and tracking mechanism, supports cyclic re-inspection and rectification progress monitoring, forms a closed-loop management, solves the problem of continuous monitoring of rectification, and is fully automated. It has strong anti-interference and low misjudgment rate, and is suitable for long-term routine inspection and maintenance of road and bridge concrete.

[0028] 2. In this invention, by achieving three-dimensional tomographic imaging and quantitative calculation based on fusion features, the depth, area, volume, and equivalent diameter of defects can be accurately obtained, and safety classification can be automatically completed. The detection results are intuitive, quantitative, and traceable, supporting cyclic re-inspection and rectification progress monitoring. It can automatically compare initial data with retest data to determine the repair effect and defect development trend, generate rectification progress reports, and realize detection-marking-rectification-re-inspection-closed-loop monitoring. The entire process is automated and requires no manual intervention. It has high detection efficiency, low misjudgment rate, and strong stability, and is suitable for large-area, long-term, and routine non-destructive testing and operation and maintenance monitoring of road and bridge concrete.

[0029] 3. In this invention, spatiotemporal registration is used to align timestamps and normalize coordinates of multimodal data, eliminating offsets and delays, and ensuring detection accuracy and consistency. Feature enhancement processing such as wavelet denoising, time-of-flight correction, and gradient enhancement is performed on each modal signal. Global correlation encoding is performed through a Transformer fusion network to suppress noise and steel reinforcement interference, significantly improving the accuracy of defect identification. It can automatically classify, locate, score confidence, and record in detail five types of defects: voids, honeycombs, cracks, interlayer voids, and incomplete grouting. It also generates a distribution map of key defects with coordinates, which is convenient for subsequent tracking and re-inspection. Attached Figure Description

[0030] Figure 1This is a schematic diagram of the detection method of the present invention;

[0031] Figure 2 This is a schematic diagram of the multimodal synchronous acquisition of the present invention;

[0032] Figure 3 This is a schematic diagram of the multimodal spatiotemporal registration of the present invention;

[0033] Figure 4 This is a schematic diagram of defect identification according to the present invention;

[0034] Figure 5 This is a schematic diagram of the three-dimensional chromatographic quantitative method of the present invention;

[0035] Figure 6 This is a schematic diagram illustrating the rectification progress monitoring of the present invention. Detailed Implementation

[0036] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0037] This invention provides a non-destructive testing method for defects in road and bridge concrete. It features three-modal fusion detection using impact echo, ultrasonic transmission, and infrared thermal imaging, solving the problems of weak detection of deep defects and difficulty in data fusion, thus improving the accuracy of identification. It adopts three-dimensional tomographic imaging and precise quantification to achieve defect visualization and quantitative analysis, making up for the shortcomings of traditional methods in terms of quantitative imaging. Through automatic identification, marking, database construction, and tracking, it improves the defect recording and tracking mechanism, supports cyclic re-inspection and rectification progress monitoring, forming a closed-loop management system, solving the problem of continuous monitoring of rectification. The entire process is automated, with strong anti-interference and low misjudgment rate, making it suitable for long-term routine inspection and maintenance of road and bridge concrete.

[0038] Please see Figure 1-6 A non-destructive testing method for defects in road and bridge concrete, comprising the following steps:

[0039] Step 1: Multimodal synchronous acquisition and dataset construction. A unified detection grid is divided on the concrete surface of the road and bridge. Impact echo signals, ultrasonic transmission signals and infrared thermal imaging sequences are synchronously acquired at the same grid point to construct a time-space-frequency three-dimensional multimodal dataset. Step 2: Multimodal spatiotemporal registration. The three-dimensional multimodal dataset is time-stamp aligned and spatial coordinates are normalized to eliminate positional offsets and time delays between signals.

[0040] Step 3: Defect Feature Enhancement and Cross-Modal Fusion. Defect features are enhanced and extracted for the three types of signals respectively. Then, the waveform features, propagation features, and temperature features are correlated and encoded through a Transformer fusion network to generate a unified defect feature vector. Step 4: Intelligent Defect Identification, Recording, and Key Marking. Based on the defect feature vector, a deep learning model is used to complete defect identification and classification, outputting the defect type, confidence level, and boundary range. The defect location and condition are recorded in detail and key markings are performed.

[0041] Step 5: 3D tomographic quantification, safety grading, and report generation. Based on fusion feature inversion, a 3D grayscale cloud map of defects is constructed. The depth, area, volume, and equivalent diameter of defects are automatically calculated. Safety grading is performed according to the scale and location of defects, and a test report is generated. Step 6: Cyclic re-inspection and rectification progress monitoring. Cyclic re-inspection is performed on the inspected road and bridge concrete areas. During re-inspection, priority is given to checking the previously marked defect areas. The rectification status is compared and analyzed to achieve real-time monitoring of the rectification construction progress. Through the fusion detection of three modes of impact echo, ultrasonic transmission, and infrared thermal imaging, the problems of weak detection of deep defects and difficulty in data fusion are solved, improving the recognition accuracy. 3D tomographic imaging and precise quantification are used to achieve defect visualization and quantitative analysis, making up for the shortcomings of traditional methods in quantitative imaging. Through automatic identification, marking, database construction, and tracking, the defect recording and tracking mechanism is improved, supporting cyclic re-inspection and rectification progress monitoring, forming a closed-loop management, solving the problem of non-continuous monitoring of rectification. The whole process is automated, with strong anti-interference and low misjudgment rate, and is suitable for long-term routine inspection and maintenance of road and bridge concrete.

[0042] In step four, when using a deep learning model to intelligently identify and classify defects based on defect feature vectors, five types of defects—voids, honeycombs, cracks, interlayer voids, and incomplete grouting of prestressed ducts—are accurately identified, scored for confidence, and located for boundary contours. The spatial coordinates, grid number, defect type, confidence level, boundary range, and initial depth of the defects are simultaneously written into the defect database. Meanwhile, defect areas are highlighted with color grading on the detection grid planar map, generating a distribution map of key defect markers with spatial coordinates. In step six, when performing cyclical re-inspections on the inspected road and bridge concrete areas, the defect database and key defect marker distribution map from the initial inspection stage are automatically retrieved. Defect grid points are re-measured and data collected according to marker priority. The quantitative parameters of defect depth, area, volume, and equivalent diameter obtained from the re-measurements are compared with the initial inspection data, and trend analysis is performed. This automatically determines whether the defect rectification is qualified, whether the repair is in place, and whether the defect has re-developed. A rectification progress monitoring report containing the rectification qualification rate, repair effect, and development trend is generated in real time, achieving closed-loop monitoring of the entire road and bridge concrete defect repair construction process.

[0043] In step one, the impact echo signal, ultrasonic transmission signal, and infrared thermal imaging sequence are synchronously triggered and acquired using the same timestamp to ensure that the signals of each modality are consistent in time and space. In step two, spatiotemporal registration includes normalizing the coordinates of the detection grid and aligning the time of the signal sampling points to eliminate positional deviations and delay errors between multimodal data. In step three, defect feature enhancement and extraction for the three types of signals includes: wavelet denoising and resonance peak extraction for the impact echo signal, time-lapse correction and attenuation compensation for the ultrasonic transmission signal, and gradient enhancement and abnormal region segmentation for the infrared thermal imaging sequence. In step three, the Transformer fusion network uses a self-attention mechanism to globally correlate and encode waveform features, propagation features, and temperature features, suppressing environmental noise and steel reinforcement interference and enhancing weak defect features.

[0044] In step five, the 3D grayscale cloud map is generated based on multimodal fusion feature inversion, which can intuitively reflect the spatial morphology, distribution location, and development trend of defects inside the concrete. In step five, the safety classification is automatically divided into four levels: I, II, III, and IV, according to the depth, area, volume, and stress area of ​​the defect, and corresponding treatment suggestions are given. In step six, the cycle of cyclic re-inspection is automatically set according to the defect safety level, and the higher the safety level of the defect area, the higher the re-inspection frequency. The detection method is fully automated and can complete the entire process of data collection, defect identification, quantitative analysis, safety rating, re-inspection supervision, and report generation without human intervention.

[0045] The workflow of this invention is as follows: First, a uniform and regular detection grid is divided on the concrete surface of the road and bridge. At the same grid point, impact echo signals, ultrasonic transmission signals and infrared thermal imaging sequences are simultaneously triggered and collected. The three types of signals are recorded using the same timestamp to construct a three-dimensional multimodal dataset containing time, space and frequency information, providing comprehensive original information for subsequent detection.

[0046] Subsequently, spatiotemporal registration processing was performed on the acquired multimodal dataset. The spatial coordinates of all detection points were normalized, and the signal sampling points were time-aligned to eliminate positional offsets and time delays between different modal data, ensuring accurate data correspondence. After registration, defect feature enhancement and extraction were performed on three types of signals: wavelet denoising and formant extraction were performed on the impact echo signal, travel time correction and attenuation compensation were performed on the ultrasonic transmission signal, and gradient enhancement and abnormal region segmentation were performed on the infrared thermal imaging sequence. The three enhanced features were then input into the Transformer fusion network, and global association encoding was completed through the self-attention mechanism to generate a unified defect feature vector.

[0047] Next, based on the defect feature vectors, a deep learning model is used for intelligent identification and classification. Five types of defects, namely voids, honeycombs, cracks, interlayer voids, and loose grouting of prestressed ducts, are identified, their confidence scores are scored, and their boundaries are located. The spatial coordinates, grid number, type, confidence score, boundary range, and initial depth of the defects are uniformly written into the defect database. At the same time, color-coded highlighting is performed on the grid map to form a distribution map of key defects. Then, a three-dimensional grayscale cloud map is constructed based on the fused features. The depth, area, volume, and equivalent diameter of the defects are automatically calculated. The safety classification is completed by combining the defect scale and the stress area, and a standardized inspection report is generated.

[0048] Finally, the system enters the cyclical re-inspection and rectification progress monitoring stage. The system automatically sets the re-inspection cycle according to the safety level. The higher the level, the higher the re-inspection frequency. During the re-inspection, the system automatically retrieves the initial database and the marked distribution map, and prioritizes the targeted re-inspection of key defect grids. The system compares the difference and trend analysis between the re-inspection measurement data and the initial data, and automatically determines whether the rectification is qualified, whether the repair is in place, and whether the defect has developed. The system also generates a rectification progress monitoring report in real time, which includes the pass rate, repair effect, and development trend. This achieves closed-loop management of road and bridge concrete defects from detection, identification, quantification to repair, re-inspection, and monitoring.

[0049] Although the present invention has been described above with reference to embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of the invention. In particular, as long as there is no structural conflict, the features in the disclosed embodiments can be combined with each other in any manner. The lack of an exhaustive description of these combinations in this specification is merely for the sake of brevity and resource conservation. Therefore, the present invention is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A non-destructive testing method for defects in road and bridge concrete, characterized in that, The detection method includes the following steps: Step 1: Multimodal synchronous acquisition and dataset construction. A unified detection grid is divided on the concrete surface of the road and bridge. Impact echo signals, ultrasonic transmission signals and infrared thermal imaging sequences are synchronously acquired at the same grid point to construct a time-space-frequency three-dimensional multimodal dataset. Step 2: Multimodal spatiotemporal registration. The three-dimensional multimodal dataset is time-stamp aligned and spatial coordinates are normalized to eliminate positional offsets and time delays between signals. Step 3: Defect feature enhancement and cross-modal fusion. Defect features are enhanced and extracted for the three types of signals respectively. Then, the waveform features, propagation features and temperature features are correlated and encoded through the Transformer fusion network to generate a unified defect feature vector. Step 4: Intelligent Defect Identification, Recording, and Highlighting. Based on the defect feature vector, a deep learning model is used to complete defect identification and classification, outputting the defect type, confidence level, and boundary range, and recording and highlighting the defect location and condition in detail. Step 5: 3D tomography quantification, safety classification and report generation. Based on the fusion feature inversion, a 3D grayscale cloud map of the defect is constructed. The defect depth, area, volume and equivalent diameter are automatically calculated. The safety classification is performed according to the defect size and location and an inspection report is generated. Step Six: Cyclic Re-inspection and Rectification Progress Monitoring. Cyclic re-inspection is carried out on the tested road and bridge concrete areas. During the re-inspection, priority is given to checking the defective parts that were marked in the early stage. The rectification status is compared and analyzed to achieve real-time monitoring of the rectification construction progress.

2. The non-destructive testing method for defects in road and bridge concrete according to claim 1, characterized in that, In step four, when performing intelligent defect identification and classification based on defect feature vectors using a deep learning model, five types of defects in concrete—voids, honeycombs, cracks, interlayer voids, and incomplete grouting of prestressed ducts—are accurately identified, scored for confidence, and located for boundary contours. The spatial coordinates of the defects, their grid number, defect type, confidence level, boundary range, and initial depth are simultaneously written into the defect database. At the same time, the defect areas are highlighted with color grading in the detection grid planar map, generating a distribution map of key defect markers with spatial coordinates. In step six, when performing cyclical re-inspection on the inspected road and bridge concrete areas, the system automatically retrieves the defect database and key defect marker distribution map from the initial inspection stage. Based on the marker priority, it performs directional re-measurement and data collection on the defect grid points. The quantitative parameters of defect depth, area, volume, and equivalent diameter obtained from the re-measurement are compared with the initial inspection data, and trend analysis is performed. This automatically determines whether the defect rectification is qualified, whether the repair is in place, and whether the defect has re-developed. Furthermore, it generates a rectification progress monitoring report in real time, including the rectification qualification rate, repair effect, and development trend, thus achieving closed-loop monitoring of the entire road and bridge concrete defect repair construction process.

3. The method for non-destructive testing of defects in road and bridge concrete according to claim 2, characterized in that, The shock echo signal, ultrasonic transmission signal, and infrared thermal imaging sequence mentioned in step one are synchronously triggered and acquired using the same timestamp to ensure that each modal signal is consistent in time and space.

4. The non-destructive testing method for defects in road and bridge concrete according to claim 3, characterized in that, The spatiotemporal registration described in step two includes normalizing the coordinates of the detection grid and aligning the signal sampling points in time to eliminate positional deviations and delay errors between multimodal data.

5. The non-destructive testing method for defects in road and bridge concrete according to claim 4, characterized in that, Step 3, which involves enhancing and extracting defect features from the three types of signals, includes: performing wavelet denoising and resonant peak extraction on the impact echo signal, performing travel time correction and attenuation compensation on the ultrasonic transmission signal, and performing gradient enhancement and abnormal region segmentation on the infrared thermal imaging sequence.

6. The non-destructive testing method for defects in road and bridge concrete according to claim 5, characterized in that, The Transformer fusion network described in step three uses a self-attention mechanism to globally correlate and encode waveform features, propagation features, and temperature features, thereby suppressing environmental noise and steel reinforcement interference and enhancing weak defect features.

7. The non-destructive testing method for defects in road and bridge concrete according to claim 6, characterized in that, The three-dimensional grayscale cloud map described in step five is generated based on multimodal fusion feature inversion, which can intuitively reflect the spatial morphology, distribution location and development trend of defects inside concrete.

8. The non-destructive testing method for defects in road and bridge concrete according to claim 7, characterized in that, The safety classification described in step five is automatically divided into four levels: I, II, III, and IV, based on the depth, area, volume, and stress zone of the defect, and corresponding handling suggestions are given.

9. The non-destructive testing method for defects in road and bridge concrete according to claim 8, characterized in that, The cycle of cyclic re-inspection described in step six is ​​automatically set according to the defect safety level. The higher the safety level of the defect area, the higher the re-inspection frequency.

10. The non-destructive testing method for defects in road and bridge concrete according to claim 9, characterized in that, The detection method is fully automated and can complete the entire process of data collection, defect identification, quantitative analysis, safety rating, re-inspection supervision and report generation without human intervention.

Images