Online garbage detecting and sorting system and method coupled with Raman spectrum and near infrared spectrum

Abstract

The invention belongs to the technical field of garbage identification and sorting, and particularly relates to a Raman spectrum and near infrared spectrum coupled garbage online detection and sorting system and method. The system comprises a feeding and conveying module, a target detection and positioning module, a near infrared spectrum detection module, a Raman spectrum detection module, a data processing and database module, a classification recognition and fusion decision module, a multi-sensor space-time synchronization module and a sorting execution module. According to the invention, online coupling detection and classification decision of Raman spectrum and near infrared spectrum are adopted, and the two complements each other, so that wider garbage categories can be covered, and the classification accuracy and robustness are remarkably enhanced; through the multi-sensor space-time synchronization module, precise space-time synchronization of multi-modal detection and sorting execution is achieved, it is ensured that a detection object and an execution object are strictly consistent, and the running reliability of the online garbage sorting system is improved.

Description

Technical Field

[0001] This invention belongs to the field of waste identification and sorting technology, specifically relating to an online waste detection and sorting system and method that couples Raman spectroscopy and near-infrared spectroscopy. Background Technology

[0002] Faced with the ever-increasing volume of waste and the deteriorating environmental situation, how to maximize the utilization of waste resources, reduce waste disposal volume, and improve the quality of the living environment through waste sorting and management is one of the most pressing issues of common concern to countries around the world. Currently, most areas in China still rely primarily on manual sorting, while a few areas use magnetic separation and air separation technologies for preliminary waste sorting. In recent years, spectral analysis technology, due to its high sensitivity to material composition and non-contact, non-destructive detection characteristics, has gradually become one of the core technologies in modern waste sorting. Chinese patent CN115753735B discloses a waste identification and sorting method and system based on online Raman spectroscopy detection, using Raman spectroscopy for waste identification and rejection, achieving fine classification of all components of household waste. Chinese patent CN119368308A discloses a multi-functional intelligent pre-screening system and its usage method, including a bag-breaking device, a large stone removal mechanism, a magnetic separation device, a particle size separation mechanism, an air separation mechanism, and a near-infrared spectroscopy plastic sorting device connected in sequence. This system can more efficiently sort urban household waste, especially improving the quality of plastic recycling. The aforementioned existing technologies utilize the rich waste composition information provided by spectral analysis technology to enhance waste sorting effectiveness, but some shortcomings still exist in practical applications: First, existing technologies employ a single mode for detection and decision-making, namely based on a single Raman spectrum or a single near-infrared spectrum. Near-infrared spectroscopy offers fast response and low cost for common plastics (PET, PP, PE, etc.), paper, and fabrics, but it cannot identify black materials. While Raman spectroscopy can identify black plastics and distinguish structurally similar polymers, it is susceptible to fluorescence interference and has a slow sampling speed. Therefore, single-mode applications have their respective drawbacks, limiting the improvement of recognition coverage and classification accuracy. Second, in the online waste sorting process, existing technologies do not consider how to maintain precise spatiotemporal synchronization between waste detection and sorting execution under high-speed conveying conditions. Fluctuations in conveyor belt speed may lead to sorting misalignment due to inconsistencies between the detected and executed objects, reducing the reliability of the online sorting system. Summary of the Invention

[0003] To address the shortcomings of the existing technologies, the technical problem to be solved by this invention is to provide an online waste detection and sorting system and method that couples Raman spectroscopy and near-infrared spectroscopy, establishes a spectral multimodal fusion complementary perception and decision-making system, and improves the overall waste identification coverage and accuracy; at the same time, it achieves spatiotemporal synchronization of multiple sensors, avoids sorting misalignment caused by conveyor belt speed fluctuations, and improves the reliability of the online sorting system.

[0004] The technical solution adopted in this invention is as follows:

[0005] An online waste detection and sorting system coupled with Raman spectroscopy and near-infrared spectroscopy includes a feeding and conveying module, a target detection and positioning module, a near-infrared spectroscopy detection module, a Raman spectroscopy detection module, a data processing and database module, a classification recognition and fusion decision module, a multi-sensor spatiotemporal synchronization module, and a sorting execution module.

[0006] The feeding and conveying module includes a feeding device, a conveyor belt, a speed-regulating motor, and an encoder or speed measuring component;

[0007] Preferably, the encoder or speed measuring component is used to acquire belt speed and displacement, providing a real-time conveying speed and displacement reference.

[0008] The target detection and positioning module includes an industrial camera, used to detect the position, length, and center point information of the waste entering the detection area, and output the ID of the waste for trajectory tracking. The ID includes a timestamp, belt area number, and frame number.

[0009] The near-infrared spectroscopy detection module includes a near-infrared spectrometer, a light source, and a acquisition probe or line scan assembly, used to acquire the near-infrared spectrum of the waste and output raw near-infrared spectral data;

[0010] The Raman spectroscopy detection module includes a Raman spectrometer, a laser, and a acquisition probe or focusing component, used to acquire the Raman spectrum of the waste and output raw Raman spectral data;

[0011] The data processing and database module is used to establish and store a database of waste samples, including a Raman spectroscopy database, a near-infrared spectroscopy database, and a fused spectroscopy database. The database contains waste category label information, collection conditions, preprocessing parameters, characteristic bands or characteristic vectors.

[0012] The classification, identification, and fusion decision module is used to output the final waste category and classification confidence level.

[0013] Preferably, the classification, identification, and fusion decision module includes at least three classifiers: a Raman spectroscopy-based classifier, a near-infrared spectroscopy-based classifier, and a fusion spectroscopy or fusion feature-based classifier.

[0014] The multi-sensor spatiotemporal synchronization module is used to map the spatiotemporal information of the target waste in the detection zone to the valve triggering time and position window in the execution zone; the module includes three parts: time synchronization, spatial calibration and tracking compensation.

[0015] Preferably, the time synchronization of the multi-sensor spatiotemporal synchronization module unifies the timestamps of Raman and near-infrared spectral acquisition; the spatial calibration establishes a mapping from the detection area coordinate system to the sorting execution area coordinate system; and the tracking compensation updates the time and position of garbage arriving at the execution mechanism in real time based on encoder displacement and belt speed fluctuations to avoid misaligned sorting.

[0016] Specifically, in terms of time synchronization, the industrial camera, Raman spectroscopy detection module, and near-infrared spectroscopy detection module all use a unified time reference to timestamp the acquired data, enabling image frames, Raman spectra, and near-infrared spectra under the same target ID to be aligned in time. In terms of spatial calibration, a geometric mapping relationship is established between the coordinate system of the detection area and the coordinate system of the sorting execution area, mapping the target center point coordinates and length direction range measured in the detection area to the corresponding spray valve array number and effective range of the execution area. In terms of tracking compensation, the displacement increment and belt speed output by the encoder or speed measuring component are used as a reference to update the arrival time drift caused by the fluctuation of the conveyor belt speed online.

[0017] The sorting execution module includes a pneumatic spray valve array and a controller for performing waste sorting.

[0018] This invention also provides an online waste detection and sorting method coupled with Raman and near-infrared spectroscopy, implemented based on the aforementioned online waste detection and sorting system coupled with Raman and near-infrared spectroscopy, comprising the following steps:

[0019] Step S1: Establish the database and perform spatial calibration;

[0020] Preferably, step S1 includes:

[0021] S1.1 Collect multiple types of waste samples and obtain the Raman and near-infrared spectra of the waste samples under preset collection conditions;

[0022] S1.2, preprocess the obtained Raman spectra of the garbage samples, including smoothing, baseline subtraction, normalization, and extraction of characteristic bands / peaks;

[0023] S1.3, preprocess the obtained near-infrared spectra of the garbage samples, including smoothing, SNV, MSC, derivative, and feature band extraction;

[0024] S1.4 Based on the spectral features extracted in S1.2 and S1.3, a Raman database, a near-infrared database, and a fusion database are constructed; the fusion database consists of feature splicing of Raman and near-infrared spectra, and fusion vectors or fusion spectra after feature selection.

[0025] S1.5, complete the spatial calibration of the detection area-execution area, and establish a mapping model between encoder belt speed and displacement and target arrival time and position;

[0026] S1.5.1 Establish the mapping relationship between the detection area coordinate system and the execution area coordinate system to obtain the correspondence table between the detection area coordinates and the valve array channel number;

[0027] S1.5.2 Establish an encoder displacement-target position update model, and calculate the predicted time and position window of the target arriving in the execution area based on belt speed v(t) and displacement increment Δs;

[0028] S1.5.3 Write the above mapping relationship and model parameters into the synchronization module configuration table for online tracking and compensation calls.

[0029] Step S2, classifier training;

[0030] Preferably, in step S2, the Raman classifier M is trained using the Raman database, near-infrared database, and fused database obtained in step S1, respectively. R Near-infrared classifier M N and fusion classifier M F And set the output confidence or reliability metric for each classifier;

[0031] Step S3 involves online detection, synchronous tracking, and classification decision-making for waste to obtain waste category, arrival time, and location window;

[0032] Preferably, step S3, targeting the waste to be detected, includes:

[0033] S3.1 Detect and locate the target entering the detection area, obtain the target center point and attitude information, record the initial position x0 and timestamp t0 and assign a target ID; the ID includes the timestamp, belt area number and frame number;

[0034] S3.2, Calculate the predicted time t for the target to arrive at the execution zone based on the encoder's real-time belt speed and displacement. arr and location window;

[0035] Preferably, in step S3.2, based on the initial parameters obtained in step S3.1, the system integrates the encoder displacement to obtain the predicted positioning of the target over time, and calculates the predicted arrival time t of the target entering the execution area accordingly. arr Simultaneously, the valve's operating position window is calculated by combining target length and attitude information; multiple frame updates are performed as the belt speed fluctuates during the target's movement to correct prediction errors introduced by changes in the target's center point and attitude, and the arrival time t is updated in real time. arr and location window.

[0036] S3.3, trigger acquisition, acquire the target's near-infrared and Raman spectra in parallel or serially;

[0037] S3.4, After processing the collected near-infrared and Raman spectra of the target, input them into the classifier M respectively. R M N M F Output the target category and classification confidence score determined by each classifier;

[0038] S3.5 determines and outputs the waste category based on the preset fusion strategy.

[0039] Step S4: Based on the waste category, arrival time, and location window obtained in step S3, perform waste sorting.

[0040] The beneficial effects obtained by adopting the above technical solution are as follows:

[0041] (1) The present invention provides an online detection and sorting system and method for waste that couples Raman spectroscopy and near-infrared spectroscopy. The system uses online coupled detection and classification decision-making of Raman spectroscopy and near-infrared spectroscopy. The two complement each other and can cover a wider range of waste categories, significantly improving classification accuracy and robustness.

[0042] (2) The present invention achieves precise spatiotemporal synchronization of multimodal detection and sorting execution through a multi-sensor spatiotemporal synchronization module, ensuring that the detection object and the execution object are strictly consistent, thereby improving the reliability of the online waste sorting system. Attached Figure Description

[0043] Figure 1 This is a schematic diagram of the overall structure of the online waste detection and sorting system that couples Raman spectroscopy and near-infrared spectroscopy according to the present invention.

[0044] Figure 2 This is a flowchart of the online waste detection and sorting method that couples Raman spectroscopy and near-infrared spectroscopy according to the present invention.

[0045] Figure 3 This is a schematic diagram illustrating the principle of spatiotemporal synchronization and target tracking compensation between the detection zone and the execution zone.

[0046] Figure 4 This is a schematic diagram of the Raman spectrum and spectral preprocessing of a waste sample.

[0047] Figure 5 This is a schematic diagram of Raman spectral characteristic band extraction from garbage samples.

[0048] Figure 6 This is a schematic diagram of the near-infrared spectrum and spectral preprocessing of a waste sample.

[0049] Figure 7 This is a schematic diagram of the extraction of near-infrared spectral feature bands from garbage samples.

[0050] Figure 8 This is a schematic diagram of principal component analysis (PCA) for the fusion of Raman and near-infrared eigenvectors.

[0051] Figure 9 A schematic diagram of waste sorting results using Raman spectroscopy as input.

[0052] Figure 10 A schematic diagram of the dual-modal joint classification decision results.

[0053] Figure 11 A schematic diagram of the classification results when Raman and near-infrared spectral features are used as model input. Detailed Implementation

[0054] The technical solution of the present invention will now be described more clearly and completely with reference to the accompanying drawings.

[0055] This embodiment presents an online waste detection and sorting system that couples Raman and near-infrared spectroscopy, such as... Figure 1 As shown, it includes a feeding and conveying module, a target detection and positioning module, a near-infrared spectroscopy detection module, a Raman spectroscopy detection module, a data processing and database module, a classification, recognition and fusion decision-making module, a multi-sensor spatiotemporal synchronization module, and a sorting execution module.

[0056] Based on the above-mentioned online waste detection and sorting system coupled with Raman and near-infrared spectroscopy, this embodiment also provides an online waste detection and sorting method coupled with Raman and near-infrared spectroscopy, such as... Figure 2 As shown, it includes the following steps:

[0057] S1, Establish the database and perform spatial calibration;

[0058] Raman and near-infrared spectra of waste samples were acquired under preset acquisition conditions, and the obtained spectra were preprocessed to construct Raman database, near-infrared database and fusion database.

[0059] Specifically, select waste categories to be sorted (e.g., 9 different types of plastics such as PET, PE, PP, PS, PVC, and ABS), with a total of 6 samples for each category. The number of samples for each category meets the needs of model training and testing (sampling can be done in several batches for each type of waste to ensure coverage of differences in source, color, and surface condition). Figure 4 As shown, the raw Raman spectrum is acquired at the Raman acquisition end, and then smoothed, baseline subtracted, and normalized; as... Figure 5 As shown, the characteristic peak intensity, peak area, or characteristic band vector is extracted and written into the "Garbage Raman Feature Database". Figure 6As shown, the raw near-infrared spectrum is acquired at the near-infrared acquisition end and processed using SNV / MSC and first / second derivatives, etc.; Figure 7 As shown, the feature band vectors are extracted and written into the "Garbage Near-Infrared Spectral Feature Database". For example... Figure 8 As shown, a "fused spectral feature database" is established by performing feature-level fusion using "Raman feature vector + near-infrared feature vector" (after splicing, PCA can be used for dimensionality reduction).

[0060] On the other hand, the spatial calibration of the detection zone and execution zone is completed, and a mapping model of encoder belt speed and displacement with target arrival time and position is established.

[0061] S2, Classifier training;

[0062] Raman classifier M is trained using the Raman database, near-infrared database, and fused database obtained in step S1. R Near-infrared classifier M N and fusion classifier M F :

[0063] M R Input Raman features;

[0064] M N Input near-infrared features;

[0065] M F Input fusion features;

[0066] Save the model version and preprocessing parameters and write them to the database metadata table for easy online access and traceability later.

[0067] The Raman feature vector of the test sample is input into the Raman classifier M. R The classification results and confidence scores for each category are obtained. The classification results are illustrated as follows: Figure 9 As shown in the figure. The results show that when Raman spectroscopy is used as the input alone, all 6 samples in the PMMA category were misidentified as PVC, indicating that under single Raman mode conditions, PMMA and PVC are confounded under the acquisition conditions / interference background of this embodiment, which leads to the classifier making systematic misclassifications of this type of sample.

[0068] S3 enables online detection, synchronous tracking, and classification decisions for waste.

[0069] To reduce the misclassification rate of the aforementioned easily confused categories, this embodiment employs a dual-modal joint classification strategy:

[0070] First, let's take the Raman classifier M... R Output category C R With confidence level P R As an initial judgment result; when the review trigger condition is met (e.g., P),R Below the threshold P th , or C R (Belonging to a pre-defined easily confused set, such as PMMA / PVC, etc.), the target's near-infrared spectrum is further acquired and processed and input into the near-infrared classifier M. N Get category C N With confidence level P N Finally, the target category C is output based on the preset joint decision-making strategy.

[0071] A serial strategy of "Raman fast master judgment + near-infrared verification" or a parallel strategy of "feature-level fusion + confidence-weighted decision" can be adopted. Multiple model consistency verification can be selected. The specific details are as follows:

[0072] 1. Online sorting – a serial strategy of “Raman rapid primary judgment + near-infrared verification”

[0073] (1) After the system starts, the encoder outputs belt speed and displacement in real time; the target detection module establishes target ID and initial position for the garbage entering the detection area;

[0074] (2) First, acquire the Raman spectrum: Acquire the Raman spectrum of the target, preprocess it, and then input it into the classifier M. R Category C is obtained. R With confidence level P R ;

[0075] (3) Near-infrared spectral verification is triggered when any of the following conditions are met:

[0076] (a)P R <P th (Low confidence level), where P th The confidence threshold;

[0077] (b)C R Belongs to easily confused groups (such as among various hydrocarbon plastics);

[0078] Near-infrared data acquisition is triggered and input into classifier M. N Category C is obtained. N and confidence level P N ;

[0079] (4) Example of final decision rule:

[0080] If a verification is triggered, the near-infrared spectroscopy results will prevail: C=C N ;

[0081] Otherwise C=C R .

[0082] The results of the dual-modal joint classification decision are illustrated below. Figure 10As shown in the figure. The results show that the classification accuracy is improved after using dual-modal joint classification. Only 1 of the 6 samples in the PMMA category was still misidentified as PVC. Compared with the case of using only Raman spectroscopy input, the misclassification is significantly reduced.

[0083] 2. Online sorting – a parallel strategy of “feature-level fusion + confidence-weighted decision”

[0084] Furthermore, the fusion features from the fusion database are input into the fusion classifier MF to obtain the classification results, as shown in the diagram. Figure 11 As shown in the figure. The results indicate that when fused spectra or fused features are used as model input, the classification accuracy of the nine plastics is 100%. This result demonstrates that by fusing Raman and near-infrared spectral information at the feature layer, the complementary information of the two can be fully utilized to enhance inter-class separability, thereby significantly improving the overall classification accuracy and robustness, especially for easily confused categories such as PMMA and PVC.

[0085] (1) After the target enters the detection area, the Raman spectroscopy detection module and the near-infrared spectroscopy detection module acquire data in parallel under the same target ID, and output two spectra and a unified timestamp.

[0086] (2) Perform preprocessing and feature extraction respectively to obtain Raman feature vector and near-infrared feature vector;

[0087] (3) Construction fusion features:

[0088] splicing: F=F[F R ,F N ];or

[0089] Weighted fusion: F = αF R ⊕(1-α)F N (α can be adaptively determined by signal-to-noise ratio / fluorescence interference level / NIR reflection intensity).

[0090] (4) Input F into the fusion model M F The final category C is obtained. F With confidence level P F .

[0091] 3. Optional multi-model consistency check: Simultaneously outputs C R C N and C F If the three conflict, the one with the highest confidence level shall be adopted or a priority shall be set (e.g., Raman spectroscopy shall be given priority to black materials).

[0092] S4, perform waste sorting;

[0093] The synchronization module uses the target ID as an index and the encoder displacement to perform online correction of the "detection-execution" time difference, such as... Figure 3 As shown, when the target arrives at the execution zone, the corresponding spray valve / actuator is controlled to perform the sorting.

Claims

1. A waste on-line detection and sorting system coupling Raman spectroscopy and near infrared spectroscopy, characterized in that, The system comprises a feeding and conveying module, a target detection and positioning module, a near-infrared spectrum detection module, a Raman spectrum detection module, a data processing and database module, a classification identification and fusion decision module, a multi-sensor time-space synchronization module and a sorting execution module. The feeding and conveying module comprises a feeding device, a conveyor belt, a speed-regulating motor and an encoder or a speed-measuring component. The target detection and positioning module comprises an industrial camera, which is used to detect the position, length and center point information of the garbage entering the detection area and output the ID of the garbage for trajectory tracking, the ID comprising a time stamp, a belt area number and a frame serial number. The near-infrared spectrum detection module comprises a near-infrared spectrometer, a light source and a collection probe or a line-scan component, which are used to collect the near-infrared spectrum of the garbage and output the near-infrared original spectrum data. The Raman spectrum detection module comprises a Raman spectrometer, a laser, a collection probe or a focusing component, which are used to collect the Raman spectrum of the garbage and output the Raman original spectrum data. The data processing and database module is used to establish and store the database of the garbage samples, including a Raman spectrum database, a near-infrared spectrum database and a fusion spectrum database, the database containing the garbage category label information, the collection condition, the pretreatment parameter, the characteristic waveband or the characteristic vector. The classification identification and fusion decision module is used to output the final category of the garbage and the classification confidence. The multi-sensor time-space synchronization module is used to map the time-space information of the target garbage in the detection area into the trigger time and position window of the spray valve in the execution area; the module comprises three parts of time synchronization, space calibration and tracking compensation. The sorting execution module comprises a pneumatic spray valve array and a controller, which are used to execute the garbage diversion.

2. The on-line trash detection and sorting system of claim 1, wherein, The encoder or speed-measuring component is used to obtain the belt speed and displacement and provide the real-time conveying speed and displacement reference.

3. The on-line trash detection and sorting system of claim 2, wherein, The classification identification and fusion decision module comprises at least three classifiers: a Raman spectrum-based classifier, a near-infrared spectrum-based classifier and a fusion spectrum or fusion feature-based classifier.

4. The on-line trash detection and sorting system of claim 3, wherein, The time synchronization of the multi-sensor time-space synchronization module is to unify the Raman and near-infrared spectrum collection time stamps. The space calibration is to establish the mapping from the detection area coordinate system to the sorting execution area coordinate system; the tracking compensation is to update the time and position of the garbage reaching the execution mechanism in real time according to the encoder displacement and belt speed fluctuation.

5. A method for on-line waste sorting and detection by coupling Raman spectroscopy and near infrared spectroscopy, implemented based on the on-line waste sorting and detection system according to claim 4, characterized in that, The method comprises the following steps: Step S1, establishing a database and performing space calibration; Step S2, classifier training; Step S3, performing online detection, synchronous tracking and classification decision on the garbage to obtain the garbage category, arrival time and position window; Step S4, performing garbage sorting according to the garbage category, arrival time and position window obtained in step S3.

6. The method of claim 5, wherein, In step S1, the following steps are included: S1.1, collecting multi-category garbage samples to obtain the Raman spectrum and near-infrared spectrum of the garbage samples under preset collection conditions; S1.2, performing pretreatment on the obtained garbage sample Raman spectrum, including smoothing, baseline deduction, normalization and characteristic waveband / characteristic peak extraction; S1.3, performing pretreatment on the obtained garbage sample near-infrared spectrum, including smoothing, SNV, MSC, derivative and characteristic waveband extraction; S1.4, constructing Raman database, near-infrared database and fusion database based on the spectral features extracted in S1.2 and S1.3; the fusion database is composed of the fusion vectors or fusion spectra after feature splicing and feature selection of Raman and near-infrared spectra; S1.5, completing spatial calibration of the detection area and the execution area, and establishing a mapping model of the encoder belt speed and displacement and the target arrival time and position; S1.5.1, establishing a mapping relationship between the detection area coordinate system and the execution area coordinate system, and obtaining a corresponding table of the detection area coordinates to the nozzle array channel number; S1.5.2, establishing an encoder displacement-target position update model, and calculating the predicted time and position window of the target arriving at the execution area based on the belt speed v(t) and displacement increment Δs; S1.5.3, writing the mapping relationship and model parameters into the synchronization module configuration table for online tracking compensation call.

7. The method of claim 6, wherein the step of detecting the presence of the waste material comprises the step of: In the step S2, the Raman database, the near-infrared database and the fusion database obtained in the step S1 are respectively used to train a Raman classifier M R , a near-infrared classifier M N and a fusion classifier M F , and an output confidence or reliability index is set for each classifier. ​ 8. The method of claim 7, wherein the step of detecting the presence of the waste material comprises the step of: In the step S3, the garbage to be detected is taken as the target, comprising: S3.1, detecting and positioning the target entering the detection area, obtaining the target center point and attitude information, recording the initial position x0 and time stamp t0 and assigning the target ID; the ID includes the time stamp, the belt area number and the frame serial number; S3.2, according to the encoder real-time speed and displacement, calculate the predicted time t of the target reaching the execution area arr and the position window; S3.3, triggering acquisition, and acquiring the near-infrared and Raman spectra of the target in parallel or in series; S3.4, the collected target near-infrared and Raman spectra are processed and input into a classifier M R N F , and the target category and classification confidence determined by each classifier are output.​​ S3.5, determining and outputting the garbage category according to the preset fusion strategy.

9. The method of claim 8, wherein, In the step S3.2, based on the initial parameters obtained in the step S3.1, the system integrates the encoder displacement to obtain the predicted positioning of the target over time, and calculates the predicted arrival time t of the target into the execution area according to the predicted positioning arr Meanwhile, the spray valve action position window is calculated in combination with the target length and the attitude information; during the target travel, multiple frame updates are performed according to the speed fluctuation, the prediction error introduced by the change of the target center point and the attitude is corrected, and the arrival time t is updated in real time arr and the position window.

Images