Method for detecting remaining capacity of vehicle-mounted intelligent garbage can of sanitation vehicle

Abstract

This application discloses a method for detecting the remaining capacity of an intelligent garbage bin on a sanitation vehicle. The method involves acquiring garbage bin weight data, multi-point distance data on the garbage surface, and vehicle motion state data; dynamically compensating the weight data based on the motion state to obtain an estimated load-bearing weight and calculating the mass filling rate; performing validity processing on the multi-point distance data and attitude compensation according to the motion state; and estimating the volume filling rate by combining the garbage bin's geometric parameters; inputting the mass filling rate and volume filling rate into a rule-based state machine for fusion judgment to obtain capacity status information, which indicates a full-load state or an abnormal state, and outputting it to the vehicle terminal, alarm terminal, and / or remote management platform. This method incorporates vehicle dynamic conditions into weighing and distance measurement correction and performs dual-dimensional fusion judgment, improving the stability and reliability of judgment under vehicle vibration and tilt conditions, reducing false positives and false negatives, and facilitating real-time monitoring and coordinated management of sanitation operations.

Description

Technical Field

[0001] This application relates to the field of intelligent sanitation vehicles and vehicle-mounted sensor fusion technology, and in particular to a method for detecting the remaining capacity of an intelligent vehicle-mounted trash can for sanitation vehicles. Background Technology

[0002] With the increasing demand for smart cities and refined sanitation management, sanitation vehicles such as sweepers and washing trucks are rapidly developing towards intelligence and networking. As an important functional component of sanitation vehicles, the ability to accurately and in real time sense the remaining capacity of onboard garbage collection bins directly affects the optimization of operating routes and cycles, the avoidance of overloading risks, and the improvement of fleet dispatching and operational management efficiency.

[0003] Existing methods for detecting the capacity of trash cans are mostly based on static or civilian scenarios, commonly including single-range measurement solutions (such as ultrasonic ranging and infrared beam measurement) and weight-based detection methods. However, when these solutions are directly applied to the scenario of onboard trash cans in sanitation vehicles, they often reveal significant shortcomings: Firstly, sanitation vehicles experience dynamic interference such as continuous vibration, acceleration, deceleration, and attitude changes during driving and operation, which can easily cause drastic fluctuations or even jumps in sensor measurements; secondly, relying solely on volume or weight as a detection dimension is rather one-sided and cannot simultaneously consider both load safety and the risk of overflow; at the same time, many solutions lack dynamic compensation for the vehicle's motion state, resulting in low data reliability under driving conditions; furthermore, system functions are often relatively isolated, lacking intelligence and linkage capabilities, making it difficult to meet the needs of refined management and self-optimization.

[0004] Therefore, in the detection of the remaining capacity of garbage recycling bins on sanitation vehicles, the obvious disturbances in the vehicle's dynamic operating conditions, the limited detection dimensions and accuracy of sensors, and the lack of motion state compensation lead to poor real-time reliability, which have become urgent problems to be solved. Summary of the Invention

[0005] This application provides a method for detecting the remaining capacity of an onboard intelligent trash can in a sanitation vehicle, aiming to solve the problems of poor real-time reliability in the existing technology for detecting the remaining capacity of onboard trash cans in sanitation vehicles, such as significant disturbances in the vehicle's dynamic operating conditions, limited sensor detection dimensions and accuracy, and lack of motion state compensation.

[0006] To address the aforementioned problems, a method for detecting the remaining capacity of an intelligent vehicle-mounted trash can in sanitation vehicles is provided, the method comprising: Acquire the weight data of the trash can, the distance data between multiple points on the surface of the trash inside the trash can, and the motion status data of the vehicle; Based on the motion state data, the weight data is dynamically compensated to obtain an estimated value of the current weight that the trash can is carrying; Based on the estimated value, the mass filling rate of the trash can is calculated; The multi-point distance data is subjected to validity processing to obtain processed multi-point distance data; Based on the motion state data, attitude compensation is performed on the processed multi-point distance data, and the volume filling rate of the trash can is estimated by combining the geometric parameters of the trash can. The mass filling rate and the volume filling rate are input into a rule-based state machine for fusion and determination to obtain the capacity status information of the trash can, which is either a full-load state or an abnormal state. The capacity status information is output to the target device, which is a vehicle-mounted terminal, an alarm terminal, and / or a remote management platform.

[0007] Optionally, in the above solution, obtaining the weight data of the trash can includes: The weight data of the trash can is collected by multiple weighing sensors installed on the load-bearing tray of the trash can. The weighing data from the multiple weighing sensors are processed to obtain the weight data of the trash can.

[0008] Optionally, in the above scheme, the step of dynamically compensating the weight data based on the motion state data to obtain an estimated value of the current load capacity of the trash can includes: The vehicle motion intensity is determined based on the aforementioned motion state data; When the vehicle motion intensity meets the preset conditions, the weight data is subjected to low-pass filtering to obtain a first estimated value of the current load weight of the trash can; When the vehicle motion intensity does not meet the preset condition, Kalman filtering dynamic estimation is performed on the weight data to obtain a second estimated value of the current load weight of the trash can, and the acceleration component in the motion state data is introduced into the state update process of the Kalman filtering. Based on the first estimate or the second estimate, an estimate of the current load capacity of the trash can is obtained.

[0009] Optionally, in the above scheme, when performing low-pass filtering on the weight data, Butterworth low-pass filtering is used to filter the weight data; The cutoff frequency of the Butterworth low-pass filter is 0.2Hz.

[0010] In the above scheme, optionally, when performing Kalman filtering dynamic estimation on the weight data, a filtering model is constructed with the estimated load weight and the rate of change of weight as state variables, and the weight data is used as the observation. During state prediction and / or state update, the acceleration component in the motion state data is introduced as an input to output the second estimate.

[0011] Optionally, in the above scheme, multi-point distance data of the surface of the trash inside the trash can can be obtained, including: The multi-point distance data is formed by collecting distance data on the surface of the trash inside the trash can using at least three distance sensors; wherein the at least three distance sensors are evenly distributed along the length of the trash can.

[0012] Optionally, in the above scheme, the step of performing validity processing on the multi-point distance data to obtain processed multi-point distance data includes: Perform range out-of-bounds detection on the multi-point distance data, and remove or replace out-of-bounds data; Missing data detection is performed on the multi-point distance data, and missing data is imputed or marked as invalid. Outlier detection is performed on the multi-point distance data, and outliers are suppressed or removed to obtain the processed multi-point distance data.

[0013] Optionally, in the above scheme, the step of performing attitude compensation on the processed multi-point distance data based on the motion state data, and estimating the volume filling rate of the trash can in conjunction with the geometric parameters of the trash can, includes: Based on the pitch angle and / or roll angle in the motion state data, tilt compensation is performed on the processed multi-point distance data to obtain attitude-compensated multi-point distance data. Based on the multi-point distance data after attitude compensation, the filling height distribution along the length direction of the trash can is determined; Numerical integration is performed on the filling height distribution to obtain the estimated volume; wherein, the numerical integration includes summing the filling height distribution using the trapezoidal rule; The volume filling rate is calculated based on the estimated volume and the geometric parameters of the trash can.

[0014] Optionally, in the above scheme, the step of inputting the mass filling rate and the volume filling rate into a rule-based state machine for fusion determination to obtain the capacity status information of the trash can includes: When the mass fill rate is not less than the first threshold, or the volume fill rate is not less than the second threshold, the capacity status information is determined to be in a full-load state. When the volume fill rate is greater than the third threshold and the mass fill rate is less than the fourth threshold, the capacity status information is determined to be an abnormal state, and the abnormal state is marked as a bridging abnormality. When the multi-point distance data after attitude compensation meets the preset dispersion condition, the capacity status information is determined to be an abnormal state, and the abnormal state is marked as an off-center load abnormality. Wherein, the first threshold is greater than the fourth threshold, and the second threshold is not less than the third threshold.

[0015] Optionally, in the above scheme, the method further includes: Obtain confirmation information for the dumping operation, and determine the threshold update trigger time based on the confirmation information; After determining the threshold update trigger time, record the historical quality fill rate and historical volume fill rate corresponding to the confirmation information, and construct a historical sample set; Regression modeling is performed on the historical quality fill rate and historical volume fill rate in the historical sample set to obtain the density coefficient; The second threshold and / or the third threshold in the rule state machine are updated based on the density coefficient.

[0016] Compared with the prior art, this application has at least the following beneficial effects: Based on further analysis and research of existing technical problems, this application recognizes that existing technologies for detecting the remaining capacity of garbage bins on sanitation vehicles suffer from significant dynamic disturbances during vehicle operation, limited sensor detection dimensions and accuracy, and poor real-time reliability due to a lack of motion state compensation. This application addresses these issues by simultaneously acquiring the weight data of the garbage bin, multi-point distance data, and vehicle motion state data, thus covering two types of observations directly related to remaining capacity under vehicle operation conditions: "load change" and "space occupancy change." Furthermore, this solution utilizes motion state data to dynamically compensate for the weight data to obtain an estimate of the current weight carried by the garbage bin, and calculates the mass filling rate accordingly. Therefore, it incorporates the impact of dynamic disturbances such as acceleration, deceleration, and vibration during vehicle movement and operation into the compensation processing chain, thereby avoiding significant fluctuations in the weight dimension under dynamic conditions that could lead to unstable judgments. Simultaneously, the multi-point distance data is first processed... The system effectively processes data, then combines motion state data for attitude compensation and incorporates the geometric parameters of the trash can to estimate the volume filling rate. This suppresses and corrects outliers and missing values ​​in the ranging dimension, as well as measurement deviations introduced by changes in vehicle attitude, before entering the volume estimation stage. This ensures that the volume dimension maintains a stable input usable for judgment in vehicle tilt and bumpy environments. Furthermore, the mass filling rate and volume filling rate are input into a rule-based state machine for fusion judgment, outputting capacity status information (full load or abnormal). This is equivalent to incorporating both "weight constraints" and "volume constraints" into the same decision-making process at the judgment level, avoiding one-sided judgments such as "weight is close to the upper limit but volume is not full" or "volume is close to the upper limit but weight is not high" when relying on only a single dimension. This reduces the probability of misjudgment and missed judgment caused by a single sensing dimension. Finally, the capacity status information is output to the vehicle terminal, alarm terminal, and / or remote management platform, enabling the detection results to be presented and processed in a timely manner during operation.

[0017] Therefore, it can be deduced that by combining the technical means of "weight dynamic compensation involving motion state, multi-point distance effectiveness processing and attitude compensation, and mass and volume dual-dimensional fusion judgment", the problems of obvious vehicle dynamic working condition disturbance, insufficient reliability of single detection dimension, and poor real-time reliability due to lack of motion state compensation in the background technology can be addressed. This achieves a more stable and reliable state judgment of the remaining capacity of garbage cans in the actual driving and working environment of sanitation vehicles, thereby solving the problems that urgently need to be solved in the background technology. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating a method for detecting the remaining capacity of an onboard intelligent trash can in a sanitation vehicle, provided in one embodiment of this application. Figure 2 This is a schematic diagram of the hardware composition and installation of a system provided in one embodiment of this application; Figure 3 A general flowchart of a method provided in one embodiment of this application; Figure 4 The following is a logic diagram of a dual-mode information fusion decision state machine provided in one embodiment of this application. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0020] With the increasing demands for smart city management, the refinement and intelligence of sanitation operations have become an inevitable trend. As a core functional component of sanitation vehicles, the real-time and accurate sensing of the remaining capacity of onboard garbage collection bins is crucial for optimizing the operation cycle of individual vehicles, preventing equipment overload damage, and realizing intelligent dispatching and management of the fleet.

[0021] Currently, when garbage bin capacity detection technologies (such as single ultrasonic ranging and infrared photodetectors) for static or civilian scenarios are directly applied to sanitation vehicles, a series of serious defects are exposed: First, they have poor environmental tolerance, and the continuous high-frequency vibration of the vehicle can easily cause drastic changes in sensor measurements; second, the detection dimensions are one-sided and unreliable, and relying solely on volume or weight detection cannot fully guarantee operational safety; third, existing solutions generally lack the ability to dynamically compensate for the vehicle's motion state, resulting in low reliability of data during operation; and finally, they are functionally isolated and have a low level of intelligence, making it impossible to achieve data linkage and self-optimization.

[0022] To address the above problems, in one embodiment, such as Figure 1 As shown, a method for detecting the remaining capacity of an onboard intelligent trash can in a sanitation vehicle is provided, including the following steps: Acquire the weight data of the trash can, the distance data between multiple points on the surface of the trash inside the trash can, and the motion status data of the vehicle; Based on the motion state data, the weight data is dynamically compensated to obtain an estimated value of the current weight that the trash can is carrying; Based on the estimated value, the mass filling rate of the trash can is calculated; The multi-point distance data is subjected to validity processing to obtain processed multi-point distance data; Based on the motion state data, attitude compensation is performed on the processed multi-point distance data, and the volume filling rate of the trash can is estimated by combining the geometric parameters of the trash can. The mass filling rate and the volume filling rate are input into a rule-based state machine for fusion and determination to obtain the capacity status information of the trash can, which is either a full-load state or an abnormal state. The capacity status information is output to the target device, which is a vehicle-mounted terminal, an alarm terminal, and / or a remote management platform.

[0023] In one embodiment, the vehicle-mounted intelligent controller acquires the weight data of the trash can, the multi-point distance data of the trash surface inside the trash can, and the motion status data of the vehicle in the main loop. The weight data is collected by a weighing sensor installed at the bottom of the trash can's load-bearing structure, the multi-point distance data is collected by a multi-node ranging array installed inside or on top of the trash can, and the motion status data is collected by an IMU installed near the trash can.

[0024] In one embodiment, the on-board intelligent controller performs dynamic compensation processing on the weight data based on motion state data to obtain an estimated value of the current load-bearing weight of the trash can; and converts the estimated value into a mass filling rate using the rated maximum load-bearing weight of the trash can or the maximum allowable load-bearing weight configured in the system as a normalization benchmark. The dynamic compensation processing can be implemented using an "adaptive selection of filtering strategy based on motion state recognition".

[0025] In one embodiment, the onboard intelligent controller performs validity processing on the multi-point distance data (including checks and corrections for out-of-bounds, missing, and outlier conditions) to obtain processed multi-point distance data. Based on the tilt information in the motion state data, it performs attitude compensation on the processed multi-point distance data and estimates the volume fill rate by combining the garbage bin's geometric model (such as cross-sectional shape, length, and effective volume). The volume fill rate can be estimated by numerically integrating the cross-sectional height distribution corresponding to each distance measurement point, for example, by using a trapezoidal rule-based cross-sectional surface accumulation algorithm to calculate the garbage's occupied volume and convert it into a volume fill rate.

[0026] In one embodiment, the vehicle-mounted intelligent controller inputs the mass fill rate and volume fill rate into a rule-based state machine for fusion determination and outputs capacity status information. The capacity status information is output as either a full-load state or an abnormal state, and an abnormal code can be further generated to distinguish the abnormal type. The capacity status information is then output to the vehicle-mounted terminal, the alarm terminal, and / or uploaded to the remote management platform via the vehicle network communication module.

[0027] This embodiment synchronously collects weighing, ranging, and IMU motion status data, performs dynamic compensation and dual-mode fusion judgment, so that the capacity status output remains available under vehicle vibration and attitude change environment, and supports local display alarm and remote reporting, which facilitates real-time control and linkage management of sanitation operation process.

[0028] In this embodiment, obtaining the weight data of the trash can includes: The weight data of the trash can is collected by multiple weighing sensors installed on the load-bearing tray of the trash can. The weighing data from the multiple weighing sensors are processed to obtain the weight data of the trash can.

[0029] In one embodiment, the trash can is placed on a special load-bearing tray, and weighing sensors are installed at multiple locations on the tray (e.g., one weighing sensor at each of the four corners) to collect multiple weighing data.

[0030] In one embodiment, "processing the weighing data of multiple weighing sensors" may specifically include: zero-point calibration (e.g., recording a reference value and offsetting the bias under empty bin or no-load conditions) and range calibration (e.g., unifying the proportional coefficient according to calibration weights / calibration curves) of each weighing data channel; consistency detection of multiple weighing data channels (e.g., comparing pairwise differences or relative deviations) and removal, amplitude limiting, or replacement of abnormal channel data; after completing calibration and anomaly handling, summarizing and fusing multiple weighing data channels (e.g., summing or weighted summing) to obtain the weight data of the trash can.

[0031] In one embodiment, the above-mentioned "processing" can be performed by the software module of the vehicle-mounted intelligent controller; the sampling frequency of the weighing sensor can be configured according to the vibration characteristics of the vehicle operation, and time-aligned with the IMU and ranging data for subsequent dynamic compensation.

[0032] This embodiment obtains stable weight data input that can be used for subsequent dynamic compensation by calibrating, eliminating anomalies, and fusing and summarizing multi-sensor weighing data, thereby reducing the impact of single sensor drift or transient interference on the weight data.

[0033] In this embodiment, the step of dynamically compensating the weight data based on the motion state data to obtain an estimated value of the current load of the trash can includes: The vehicle motion intensity is determined based on the aforementioned motion state data; When the vehicle motion intensity meets the preset conditions, the weight data is subjected to low-pass filtering to obtain a first estimated value of the current load weight of the trash can; When the vehicle motion intensity does not meet the preset condition, Kalman filtering dynamic estimation is performed on the weight data to obtain a second estimated value of the current load weight of the trash can, and the acceleration component in the motion state data is introduced into the state update process of the Kalman filtering. Based on the first estimate or the second estimate, an estimate of the current load capacity of the trash can is obtained.

[0034] In one embodiment, the vehicle-mounted intelligent controller determines the vehicle's motion intensity based on motion state data; the vehicle's motion intensity can be characterized by the continuous exceeding of threshold values, peak amplitudes, or statistics of the longitudinal or lateral acceleration output by the IMU within a preset time window.

[0035] In one embodiment, when the vehicle motion intensity meets a preset condition, a first estimate is obtained by performing low-pass filtering on the weight data; when the vehicle motion intensity does not meet the preset condition, a second estimate is obtained by performing Kalman filtering dynamic estimation on the weight data, and the acceleration component in the motion state data is introduced during the state update process of Kalman filtering; finally, an estimate of the current load weight of the trash can is output based on the first estimate or the second estimate.

[0036] In one embodiment, the "preset condition" can be set as "the IMU detects that the longitudinal or lateral acceleration of the vehicle continuously exceeds a set threshold" or its equivalent criterion; and frequent strategy switching can be avoided by performing noise reduction, sliding window judgment or hysteresis threshold logic on the acceleration signal.

[0037] This embodiment adaptively selects a weight estimation strategy based on vehicle motion intensity and introduces IMU acceleration input into dynamic estimation. This allows for a data processing path that is more suitable for the vehicle environment under different operating conditions, thereby improving the stability and usability of load weight estimation.

[0038] In this embodiment, when performing low-pass filtering on the weight data, Butterworth low-pass filtering is used to filter the weight data; The cutoff frequency of the Butterworth low-pass filter is 0.2Hz.

[0039] In one embodiment, the low-pass filtering process uses a Butterworth low-pass filter to filter the weight data. The filter order can be configured according to the controller's computing power and suppression requirements (e.g., second-order or fourth-order) to ensure a balance between real-time performance and filtering effect.

[0040] In one embodiment, the cutoff frequency of the Butterworth low-pass filter is set to 0.2 Hz; when the vehicle experiences continuous vibration or acceleration disturbance, the low cutoff frequency is used to perform strong suppression filtering on the weighing data to output a first estimate for subsequent mass fill rate calculation.

[0041] This embodiment uses a Butterworth low-pass filter with a cutoff frequency of 0.2Hz to process the weight data, which can significantly suppress weighing fluctuations caused by high-frequency vibrations in the vehicle, and enable the output of weight estimation results that can be used for judgment even under severe motion conditions.

[0042] In this embodiment, when performing Kalman filtering dynamic estimation on the weight data, a filtering model is constructed with the estimated load weight and the rate of change of weight as state variables, and the weight data is used as the observation. During state prediction and / or state update, the acceleration component in the motion state data is introduced as an input to output the second estimate.

[0043] In one embodiment, the Kalman filter dynamic estimation is implemented using a state-space model, where the state variables include the load weight estimate and the weight change rate; the observations are weight data; and the acceleration component in the motion state data is introduced as an input during state prediction and / or state update, thereby modeling and compensating for vehicle dynamic disturbances.

[0044] In one embodiment, the process noise and measurement noise of the Kalman filter can be calibrated based on the vehicle vibration level, sensor range, and noise level; when a small change in acceleration is detected and the vehicle body is relatively stable, the dynamic weighing model is activated to obtain a second estimate.

[0045] This embodiment constructs a dynamic weighing model and introduces IMU acceleration input into the filtered state equation. Under stable or low-disturbance conditions, it can make full use of model constraints and observation information to achieve continuous and smooth estimation of the load-bearing weight, and reduce the impact of occasional disturbances on the weighing results.

[0046] In this embodiment, acquiring multi-point distance data on the surface of trash inside the trash can includes: The multi-point distance data is formed by collecting distance data on the surface of the trash inside the trash can using at least three distance sensors; wherein the at least three distance sensors are evenly distributed along the length of the trash can.

[0047] In one embodiment, multi-point distance data is acquired by at least three ranging sensors, which can be industrial-grade laser ranging sensors; multiple ranging sensors are arranged equidistantly along the length of the trash can to form a linear ranging array to obtain height / distance information of multiple points on the trash surface.

[0048] In one embodiment, the ranging sensor is installed below the top of the bin and facing the inner cavity of the trash can. The installation angle and height can be configured according to the depth of the trash can and the field of view. The ranging sampling period and the weighing sampling period can be unified or synchronized by frequency doubling to facilitate subsequent volume filling rate estimation and fusion determination.

[0049] This embodiment uses at least three ranging sensors evenly distributed along the length of the trash can to obtain multi-point contour information of the trash surface along the length direction, providing a more sufficient data basis for subsequent volume estimation and reducing the sensitivity of single-point ranging to local accumulation patterns.

[0050] In this embodiment, the step of performing validity processing on the multi-point distance data to obtain processed multi-point distance data includes: Perform range out-of-bounds detection on the multi-point distance data, and remove or replace out-of-bounds data; Missing data detection is performed on the multi-point distance data, and missing data is imputed or marked as invalid. Outlier detection is performed on the multi-point distance data, and outliers are suppressed or removed to obtain the processed multi-point distance data.

[0051] In one embodiment, when performing range overrun detection on multi-point distance data, the distance value is compared with the effective range of the ranging sensor and the geometric boundary of the trash can (such as the maximum distance from the mouth of the can to the bottom of the can); the data that exceeds the boundary can be removed, limited, or replaced by interpolation with the nearest time / nearby ranging point.

[0052] In one embodiment, when performing missing data detection on multi-point distance data, the determination can be based on communication timeout, data frame verification failure, or fixed anomaly code; missing data can be processed by interpolation or marking as invalid, and invalid points can be reduced in weight or skipped in subsequent volume estimation.

[0053] In one embodiment, when performing outlier detection on multi-point distance data, sliding window midpoint filtering, threshold detection based on mean and variance, or spatial constraint detection based on the consistency of adjacent ranging points can be used; outliers can be suppressed or eliminated, and the processed multi-point distance data can be output for attitude compensation and volume estimation.

[0054] This embodiment improves the reliability of multi-point distance data under vehicle operating conditions such as dust obstruction, abnormal reflection, or communication jitter by effectively handling out-of-bounds, missing, and out-of-group data, thus providing stable input for attitude compensation and volume estimation.

[0055] In this embodiment, the step of performing attitude compensation on the processed multi-point distance data based on the motion state data, and estimating the volume filling rate of the trash can in conjunction with the geometric parameters of the trash can, includes: Based on the pitch angle and / or roll angle in the motion state data, tilt compensation is performed on the processed multi-point distance data to obtain attitude-compensated multi-point distance data. Based on the multi-point distance data after attitude compensation, the filling height distribution along the length direction of the trash can is determined; Numerical integration is performed on the filling height distribution to obtain the estimated volume; wherein, the numerical integration includes summing the filling height distribution using the trapezoidal rule; The volume filling rate is calculated based on the estimated volume and the geometric parameters of the trash can.

[0056] In one embodiment, attitude compensation is achieved based on the pitch angle and / or roll angle in the motion state data; the on-board intelligent controller can use pitch and roll to calculate the influence of the angle between the ranging direction and the gravity direction, and perform tilt compensation on the processed multi-point distance data to obtain attitude-compensated multi-point distance data.

[0057] In one embodiment, the vehicle-mounted intelligent controller determines the filling height distribution along the length of the trash can based on multi-point distance data after attitude compensation; for example, the distance values ​​of each measuring point are converted into the "trash height" or "remaining space height" of the corresponding position, and arranged according to the position of the measuring points in the length direction to form a height distribution sequence.

[0058] In one embodiment, the volume estimation is achieved using numerical integration. Specifically, the cross-sectional area can be constructed based on the distance values ​​of each measuring point and the cross-sectional shape of the trash can (rectangular, approximately trapezoidal, or other known cross-sections). The cross-sectional area accumulation algorithm based on the trapezoidal rule is then used to accumulate the area along the length direction to obtain the volume estimation result. The volume filling rate is then calculated using the effective volume of the trash can as a normalization benchmark.

[0059] This embodiment reduces the impact of vehicle attitude changes on ranging results by using attitude compensation based on IMU tilt angle and numerical integration volume estimation based on trapezoidal law, and transforms multi-point contour information into volume fill rate that can be used for judgment, thereby improving the stability of volume dimension monitoring.

[0060] In this embodiment, the step of inputting the mass filling rate and the volume filling rate into a rule-based state machine for fusion determination to obtain the capacity status information of the trash can includes: When the mass fill rate is not less than the first threshold, or the volume fill rate is not less than the second threshold, the capacity status information is determined to be in a full-load state. When the volume fill rate is greater than the third threshold and the mass fill rate is less than the fourth threshold, the capacity status information is determined to be an abnormal state, and the abnormal state is marked as a bridging abnormality. When the multi-point distance data after attitude compensation meets the preset dispersion condition, the capacity status information is determined to be an abnormal state, and the abnormal state is marked as an off-center load abnormality. Wherein, the first threshold is greater than the fourth threshold, and the second threshold is not less than the third threshold.

[0061] In one embodiment, the rule state machine takes mass fill rate and volume fill rate as inputs; when the mass fill rate is not less than a first threshold or the volume fill rate is not less than a second threshold, the output capacity status information is full load; when the volume fill rate is greater than a third threshold and the mass fill rate is less than a fourth threshold, the output capacity status information is abnormal and marked as bridging abnormal, wherein the first threshold, the second threshold, the third threshold, and the fourth threshold are preset values.

[0062] In one embodiment, the off-center load anomaly is achieved through the dispersion condition of multi-point distance data; for example, the standard deviation of multiple distance values ​​after attitude compensation is calculated and compared with a set range. When the standard deviation exceeds the set range, the capacity status information is output as an abnormal state and marked as an off-center load anomaly.

[0063] In one embodiment, to satisfy the threshold size relationship, a first threshold can be set to be greater than a fourth threshold and a second threshold not less than a third threshold during the threshold configuration stage; and in engineering implementation, threshold management can be achieved by using a fixed threshold or by having the threshold issued by the background parameters.

[0064] This embodiment uses "OR" logic-based full load determination and "combined conditions / discretion"-based anomaly determination to simultaneously cover the triggering conditions of both weight and volume dimensions within the same rule state machine. It also classifies and marks abnormal operating conditions such as bridging and off-center loading, facilitating subsequent alarms and handling.

[0065] In this embodiment, the method further includes: Obtain confirmation information for the dumping operation, and determine the threshold update trigger time based on the confirmation information; After determining the threshold update trigger time, record the historical quality fill rate and historical volume fill rate corresponding to the confirmation information, and construct a historical sample set; Regression modeling is performed on the historical quality fill rate and historical volume fill rate in the historical sample set to obtain the density coefficient; The second threshold and / or the third threshold in the rule state machine are updated based on the density coefficient.

[0066] In one embodiment, the confirmation information for the dumping operation can be generated by the limit switch of the vehicle dumping mechanism, the hydraulic cylinder position sensor, the garbage bin tilting angle sensor, or the "dumping completed" flag in the vehicle operation control signal; after receiving the confirmation information, the on-board intelligent controller uses it as the basis for triggering the threshold update and determines the threshold update trigger time.

[0067] In one embodiment, after determining the threshold update trigger time, the vehicle-mounted intelligent controller records the historical quality fill rate and historical volume fill rate corresponding to the confirmation information to construct a historical sample set. The historical sample set may contain data pairs from multiple dumping cycles and may include corresponding extended fields such as operation time, vehicle identification, and operation route for offline analysis.

[0068] In one embodiment, a density coefficient is obtained by performing regression modeling on the historical mass filling rate and historical volume filling rate in the historical sample set; one implementation is to use a linear regression model to learn a conversion coefficient K for characterizing the apparent density of waste, where K can be used as the regression slope of the volume filling rate with respect to the mass filling rate; and based on the density coefficient, the second threshold and / or the third threshold in the rule state machine are updated, for example, when K increases beyond a preset range, the volume threshold is increased to adapt to the characteristics of fluffy waste.

[0069] This embodiment uses dumping confirmation information as a correction basis and performs regression modeling on historical mass filling rate and volume filling rate to obtain a density coefficient that characterizes the change in apparent density of waste. Based on this, the second threshold and / or the third threshold are updated, so that the judgment threshold can be adaptively adjusted with changes in waste composition and accumulation characteristics, thereby reducing the risk of misjudgment caused by changes in waste density during long-term operation.

[0070] In one embodiment, a method and system for detecting the remaining capacity of an onboard intelligent trash can for sanitation vehicles are provided to solve the core problems of inaccurate, unreliable, and unintelligent capacity detection under complex dynamic environments on vehicles.

[0071] To achieve the above objectives, this embodiment adopts the following technical solution: a method for detecting the remaining capacity of an intelligent vehicle-mounted trash can, the core of which lies in constructing a collaborative processing chain of "data acquisition - dynamic compensation - fusion decision-making," specifically including: S1: Synchronous acquisition of multi-source heterogeneous data. Deploy load cells, a multi-node linear laser ranging array, and a six-axis inertial measurement unit (IMU) to synchronously acquire mass, distance profile, and motion status signals.

[0072] S2: Intelligent signal processing for dynamic environments. Based on IMU data, the system identifies the vehicle's motion state and adaptively switches filtering strategies: a strong suppression low-pass filter (such as a 0.2Hz Butterworth filter) is used during rapid motion, while a Kalman filter incorporating IMU acceleration input is used during stable or stationary conditions to calculate a high-precision mass fill rate (W%). Simultaneously, tilt compensation is performed on the ranging values ​​using IMU tilt data, and the volume fill rate (V%) is estimated based on numerical integration methods such as the trapezoidal rule.

[0073] S3: Rule-based dual-mode fusion intelligent decision-making. A state machine model with weight safety and space safety as dual objectives is established. When W% ≥ mass threshold (W_th) or V% ≥ volume threshold (V_th), it is determined to be "fully loaded". It can also diagnose physical anomalies such as "bridging" and "off-center loading" based on specific logic (such as extremely high V% and extremely low W%).

[0074] S4: Multi-layered information interaction and cloud-based empowerment. Decision results are displayed as alarms through the in-vehicle human-machine interface, and data such as capacity, anomalies, and location are packaged and uploaded to the cloud platform through the vehicle network terminal.

[0075] Furthermore, the method also includes a system self-learning step, which dynamically updates the coefficient K representing garbage density by analyzing historical W%-V% data pairs through linear regression, and adaptively fine-tunes the decision threshold (such as V_th) accordingly, so that the system has environmental adaptability.

[0076] This embodiment constructs a robust monitoring system specifically designed for extreme vehicle environments by deeply fusing and compensating for information from three sources: mass, volume, and IMU. This fundamentally solves the problem of inaccurate detection under dynamic conditions.

[0077] The system innovatively employs an "OR" logic dual-threshold judgment rule, simultaneously ensuring both weight and space safety, resulting in a more scientific and comprehensive definition. It possesses intelligent diagnostic capabilities for abnormal operating conditions such as "bridging" and "off-center loading," achieving a leap from "monitoring" to "diagnosis." The introduction of an online self-learning mechanism enables the system to adapt to different waste compositions and operational characteristics, exhibiting environmentally adaptive and intelligent features. It provides a highly reliable, multi-dimensional data source for the smart sanitation big data platform, empowering precise management and efficiency optimization.

[0078] In this embodiment, combined with Figure 2 The hardware system deployment shown is as follows: The system hardware is deployed inside the garbage storage compartment of the sanitation sweeper truck. A steel garbage recycling bin 1 is placed on a dedicated load-bearing pallet 2, with a weighing sensor 3 at each of the four corners of the pallet. Below the compartment roof 4, three industrial-grade laser rangefinders (5a, 5b, 5c) are installed at equal intervals along the length of the garbage bin, forming a linear rangefinder array. A six-axis IMU module 6 is fixed to the side of the pallet's steel frame. All sensors are connected to the onboard intelligent controller 7. The controller 7 is also connected to a display and alarm terminal 8 in the driver's cab and a remote information terminal (T-Box) 9 for the vehicle.

[0079] In this embodiment, combined with Figure 3The overall process of the method shown is as follows: After the system powers on and completes the initialization self-test (S201), it enters the main loop. First, all sensor data are collected synchronously (S202). Based on the IMU data, the vehicle motion state is analyzed (S203): If the acceleration continuously exceeds the threshold, it is determined to be "violent motion", and a Butterworth low-pass filter with a cutoff frequency of 0.2Hz is activated for the weighing data; otherwise, a Kalman filter that introduces the IMU acceleration input is activated, and then the mass fill rate W% is calculated (S204). Simultaneously, the ranging data is verified, and tilt compensation is performed using the IMU tilt angle, using the formula H_avg=(d1+2 The average height is calculated using the compound trapezoidal rule (d2+d3) / 4, which is then used to estimate the volumetric fill rate V% (S205). Subsequently, the following steps are performed: Figure 3 The defined dual-mode fusion decision (S206) outputs the status. The results are used for local display and audible and visual alarms, and are uploaded to the cloud via T-Box (9) (S207). The background self-learning task (S208) continues to run. For example, when the system analysis finds that multiple "full load alarms" are triggered by the second threshold (i.e., the volume threshold V_th) and the average W% is low, the V_th value is automatically fine-tuned to increase to adapt to the characteristics of fluffy garbage.

[0080] In this embodiment, combined with Figure 4 The state machine logic shown is as follows: The system initially enters the "idle / empty bucket" state. When W% or V% exceeds the start threshold (e.g., 5%), it transitions to the "normal monitoring" state. In this state, continuous monitoring continues: if (W% ≥ W_th) or (V% ≥ V_th) (condition C1), it jumps to the "full load alarm" state; if (V% ≥ V_high) and (W% ≤ W_low) (condition C2), it jumps to the "abnormal alarm (bridging)" state; if the standard deviation σ of the ranging value exceeds the limit, it jumps to the "abnormal alarm (off-center load)" state. Any alarm state is only reset to the "idle / empty bucket" state after the dumping is complete and W% and V% return to low.

[0081] Taking an 18-ton sweeper truck as an example, let W_th = 98% and V_th = 92%. When sweeping loose fallen leaves in autumn, V% may reach 92% first, triggering an alarm to indicate the space is nearing full. In winter, when sweeping easily compacted dust, W% may reach 98% first, triggering an alarm to ensure load safety. Through self-learning, if the system recognizes frequent volume alarms during the leaf-fall season when the actual weight is low, V_th can be adaptively fine-tuned to 94%, reducing invalid alarms. All data is uploaded to the cloud, assisting in intelligent fleet scheduling and operational analysis.

[0082] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. A method for detecting the remaining capacity of an intelligent vehicle-mounted trash can in a sanitation vehicle, characterized in that, The method includes: Acquire the weight data of the trash can, the distance data between multiple points on the surface of the trash inside the trash can, and the motion status data of the vehicle; Based on the motion state data, the weight data is dynamically compensated to obtain an estimated value of the current weight that the trash can is carrying; Based on the estimated value, the mass filling rate of the trash can is calculated; The multi-point distance data is subjected to validity processing to obtain processed multi-point distance data; Based on the motion state data, attitude compensation is performed on the processed multi-point distance data, and the volume filling rate of the trash can is estimated by combining the geometric parameters of the trash can. The mass filling rate and the volume filling rate are input into a rule-based state machine for fusion and determination to obtain the capacity status information of the trash can, which is either a full-load state or an abnormal state. The capacity status information is output to the target device, which is a vehicle-mounted terminal, an alarm terminal, and / or a remote management platform.

2. The method according to claim 1, characterized in that, Obtaining the weight data of the trash can includes: The weight data of the trash can is collected by multiple weighing sensors installed on the load-bearing tray of the trash can. The weighing data from the multiple weighing sensors are processed to obtain the weight data of the trash can.

3. The method according to claim 1, characterized in that, The step of dynamically compensating the weight data based on the motion state data to obtain an estimated value of the current weight carried by the trash can includes: The vehicle motion intensity is determined based on the aforementioned motion state data; When the vehicle motion intensity meets the preset conditions, the weight data is subjected to low-pass filtering to obtain a first estimated value of the current load weight of the trash can; When the vehicle motion intensity does not meet the preset condition, Kalman filtering dynamic estimation is performed on the weight data to obtain a second estimated value of the current load weight of the trash can, and the acceleration component in the motion state data is introduced into the state update process of the Kalman filtering. Based on the first estimate or the second estimate, an estimate of the current load capacity of the trash can is obtained.

4. The method according to claim 3, characterized in that, When performing low-pass filtering on the weight data, Butterworth low-pass filtering is used to filter the weight data; The cutoff frequency of the Butterworth low-pass filter is 0.2Hz.

5. The method according to claim 3, characterized in that, When performing Kalman filtering dynamic estimation on the weight data, a filtering model is constructed with the estimated load weight and the rate of change of weight as state variables, and the weight data is used as the observation. During state prediction and / or state update, the acceleration component in the motion state data is introduced as an input to output the second estimate.

6. The method according to claim 1, characterized in that, Obtain multi-point distance data on the surface of trash inside the trash can, including: The multi-point distance data is formed by collecting distance data on the surface of the trash inside the trash can using at least three distance sensors; wherein the at least three distance sensors are evenly distributed along the length of the trash can.

7. The method according to claim 1, characterized in that, The step of performing validity processing on the multi-point distance data to obtain processed multi-point distance data includes: Perform range out-of-bounds detection on the multi-point distance data, and remove or replace out-of-bounds data; Missing data detection is performed on the multi-point distance data, and missing data is imputed or marked as invalid. Outlier detection is performed on the multi-point distance data, and outliers are suppressed or removed to obtain the processed multi-point distance data.

8. The method according to claim 1, characterized in that, The step of performing attitude compensation on the processed multi-point distance data based on the motion state data, and estimating the volume filling rate of the trash can in conjunction with the geometric parameters of the trash can, includes: Based on the pitch angle and / or roll angle in the motion state data, tilt compensation is performed on the processed multi-point distance data to obtain attitude-compensated multi-point distance data. Based on the multi-point distance data after attitude compensation, the filling height distribution along the length direction of the trash can is determined; Numerical integration is performed on the filling height distribution to obtain the estimated volume; wherein, the numerical integration includes summing the filling height distribution using the trapezoidal rule; The volume filling rate is calculated based on the estimated volume and the geometric parameters of the trash can.

9. The method according to claim 1, characterized in that, The step of inputting the mass filling rate and the volume filling rate into a rule-based state machine for fusion and determination to obtain the capacity status information of the trash can includes: When the mass fill rate is not less than the first threshold, or the volume fill rate is not less than the second threshold, the capacity status information is determined to be in a full-load state. When the volume fill rate is greater than the third threshold and the mass fill rate is less than the fourth threshold, the capacity status information is determined to be an abnormal state, and the abnormal state is marked as a bridging abnormality. When the multi-point distance data after attitude compensation meets the preset dispersion condition, the capacity status information is determined to be an abnormal state, and the abnormal state is marked as an off-center load abnormality. Wherein, the first threshold is greater than the fourth threshold, and the second threshold is not less than the third threshold.

10. The method according to claim 9, characterized in that, The method further includes: Obtain confirmation information for the dumping operation, and determine the threshold update trigger time based on the confirmation information; After determining the threshold update trigger time, record the historical quality fill rate and historical volume fill rate corresponding to the confirmation information, and construct a historical sample set; Regression modeling is performed on the historical quality fill rate and historical volume fill rate in the historical sample set to obtain the density coefficient; The second threshold and / or the third threshold in the rule state machine are updated based on the density coefficient.

Images