A system state updating method, system and storage medium
By using multi-sensor data acquisition and the MSTSF algorithm to process out-of-order data and outliers, and combining Kalman gain and covariance matrix for system state updates, the problem that classical filtering algorithms cannot handle out-of-order data and outliers is solved, and more accurate system state prediction is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HUIZHOU DESAY SV AUTOMOTIVE
- Filing Date
- 2024-11-30
- Publication Date
- 2026-06-02
AI Technical Summary
Existing classical filtering algorithms cannot handle out-of-order data and are affected by outliers from a single sensor, leading to data loss and increased state prediction errors.
By collecting environmental data from multiple sensors and combining the time values of the measurements, the system state is predicted. The MSTSF algorithm is used to process out-of-order data and outliers. Weighted operations are used to suppress the influence of abnormal measurements. The system state is updated by combining Kalman gain and covariance matrix.
It improves the accuracy and reliability of system state updates, avoids data loss and performance degradation, and enhances the adaptability and robustness of system state prediction.
Smart Images
Figure CN122132054A_ABST
Abstract
Description
Technical Field
[0001] This application belongs to the field of autonomous driving technology, specifically relating to a system state update method, system, and storage medium. Background Technology
[0002] In autonomous driving technology, to provide more accurate and stable perception of autonomous driving scenarios, it is necessary to fuse data from multiple sensors. The classic algorithm for multi-sensor data fusion is the classic Bayesian filtering algorithm, represented by the Kalman filter. However, classic filtering algorithms have two limitations: first, they cannot handle out-of-order data; second, they are greatly affected by outlier measurements from a single sensor.
[0003] Classical filtering algorithms require input data to satisfy a Markov process, meaning the current system state depends only on the previous state and is independent of all previous states. This implies that future states are only related to the current state and have no relation to past history. Based on the current system state and received sensor data, the updated system state is calculated through filtering. During this process, the time of the sensor data must not be earlier than the current system time. However, in practical applications of autonomous driving, different sensors have different processing times, often resulting in out-of-order data—that is, sensor data whose time is earlier than the current system time. Classical filtering algorithms cannot handle out-of-order data and must discard it, leading to information loss and performance degradation.
[0004] Classical filtering algorithms employ sequential filtering for multi-sensor data. This involves sorting the data from multiple sensors by time and filtering them sequentially, using only one sensor's data point per filtering iteration. However, when a single sensor produces an abnormal measurement, the classical filtering algorithm becomes affected by this outlier and cannot use data from other sensors to constrain and balance this abnormal data, leading to increased state estimation errors. Summary of the Invention
[0005] To address the aforementioned technical problems, this application proposes a system state update method, system, and storage medium, aiming to solve the problems of data loss caused by the inability of classic multi-sensor data fusion algorithms to handle out-of-order data, and the increased state prediction error caused by abnormal measurement values from a single sensor.
[0006] Specifically, this application proposes a system state update method, including:
[0007] The system acquires environmental data collected by multiple sensors within a preset period; the environmental data includes at least the measured values and corresponding time values.
[0008] State prediction is performed based on multiple measurements to obtain the predicted system state value.
[0009] Furthermore, the current system state is updated based on the predicted system state value.
[0010] In the above technical solution, the acquisition of environmental data by multiple sensors enables the acquisition of more accurate and stable environmental data. By combining the time values corresponding to the multiple measurements for state prediction, data loss and performance degradation caused by out-of-order data are avoided. Errors in system state prediction caused by abnormal measurements from a single sensor are also avoided, thereby improving the accuracy and reliability of system state updates.
[0011] The current system state value is calculated and obtained based on the first system prediction data and the system state value at time t.
[0012] The covariance matrix of the current system is calculated and obtained based on the prediction data of the first system and the prediction data of the second system.
[0013] The first system prediction data includes at least a system transformation matrix;
[0014] The second system prediction data includes at least the process noise matrix and the covariance matrix at time t, where t is an integer greater than or equal to 1.
[0015] By using the first and second system prediction data to predict the system state, the current system state can be predicted more accurately, reducing prediction errors and improving the accuracy and reliability of system state prediction. The system transformation matrix describes the evolution of the system state over time, improving the adaptability of system state prediction. The process noise matrix considers unforeseen disturbances and external noise in the system, accurately compensating for uncertainties in the system state prediction process.
[0016] Furthermore, obtaining the predicted system state value includes:
[0017] The Kalman gain corresponding to the measured value is calculated based on the first computational data corresponding to the measured value and the covariance matrix at time t; the first computational data includes at least the measurement matrix and measurement noise.
[0018] The predicted covariance matrix is calculated based on the current system covariance matrix, Kalman gain, and the first computational data.
[0019] The predicted system state value is calculated based on the Kalman gain, the current system state value, the measured value, and the first computational data.
[0020] The Kalman gain is calculated using the measurement matrix corresponding to the measured value, the measurement noise, and the covariance matrix at time t. This ensures accurate Kalman gain under varying measurement noise environments, improving the reliability of the system state prediction process. Calculating the predicted system state value reduces error accumulation and improves its accuracy. The covariance matrix at different times reflects the uncertainty of the system state prediction. Calculating the Kalman gain and the predicted covariance matrix allows for real-time acquisition of the predicted system state value, enabling it to better cope with external noise interference and improving the adaptability of system state value prediction.
[0021] Furthermore, at least one of the measured values in the environmental data corresponds to a time value that is earlier than the current system state value.
[0022] By allowing the time value corresponding to the measured value to be earlier than the current system state value, data loss caused by data disorder during system state updates is avoided, ensuring data integrity. Using measured values earlier than the current system state value allows for effective adjustment of the current system state prediction, improving the predictive capability of the current system state value.
[0023] Furthermore, among the multiple environmental data collected by the same sensor within the preset period, only the measurement values within a preset time range from the current time are retained.
[0024] By retaining only the measured values within a preset time range from multiple environmental data collected by the same sensor within a preset period, the timeliness and real-time nature of the measurement data during system status updates are ensured. This avoids the calculation and interference of multiple historical data, thereby reducing the corresponding latency and improving the efficiency of system status updates. Retaining the measured values within the preset time range at the current moment reduces the amount of data stored, thus reducing the consumption of computing and storage resources. It simplifies the complexity of data storage and management, reduces reliance on complex algorithms, and improves the maintainability and stability of the system status update algorithm.
[0025] Furthermore, a preset period is determined in advance, and target environment data collected by multiple sensors within the preset period is acquired, including:
[0026] The system continuously acquires raw environmental data collected by various sensors, and extracts the target environmental data from the raw environmental data based on the preset period.
[0027] By acquiring the target environment data from the original environment data at preset intervals, continuous collection of the target environment data ensures data continuity and avoids system state prediction errors caused by data interruptions. This ensures the reliability of the collected target environment data.
[0028] Furthermore, the process of updating the current system state also includes handling outliers.
[0029] The outliers include at least out-of-order measurements and abnormal measurements; the out-of-order measurements are those in the environmental data whose time values are earlier than the current system state values; when the deviation of any predicted system state value from other predicted system state values exceeds a preset threshold, the measurement value corresponding to the predicted system state value is determined to be an abnormal measurement value.
[0030] The outlier handling described above can effectively eliminate inaccurate data caused by measurement errors, equipment malfunctions, communication delays, and other reasons. When sensors produce abnormal measurement values, predictions of the system state based on these abnormal values will lead to errors in the predicted system state, affecting the accuracy of the prediction results. By promptly detecting and handling outliers, the effectiveness and reliability of system state updates can be ensured.
[0031] Furthermore, the outlier handling includes:
[0032] The predicted system state value is calculated simultaneously using multiple measurements and out-of-order measurements.
[0033] Weighted operations are performed based on the predicted covariance matrix to suppress the out-of-order and abnormal measurements.
[0034] By performing weighted operations based on the predicted covariance matrix, the weights of out-of-order and abnormal measurements are reduced or suppressed, thereby reducing the impact of out-of-order and abnormal measurements on system state prediction, improving the robustness and reliability of system state updates, and ensuring the stability of system state updates.
[0035] Based on the same inventive concept, this application also proposes a system for a system state update method, comprising:
[0036] The data acquisition module is used to acquire environmental data collected by multiple sensors within a preset period; the environmental data includes at least the measured value and the corresponding time value.
[0037] The state prediction module is used to predict the state based on multiple measurements in order to obtain the predicted system state value.
[0038] And a state update module, used to update the current system state based on the predicted system state value.
[0039] Based on the same inventive concept, this application also proposes a computer-readable storage medium storing computer-executable instructions, which, when executed by a control processor, implement the system state update method.
[0040] Compared with the prior art, this application has at least the following beneficial effects:
[0041] This application proposes a system state update method, system, and storage medium, aiming to solve the problems of data loss caused by the inability of classic multi-sensor data fusion algorithms to handle out-of-order data, and the increased state prediction error caused by abnormal measurements from a single sensor. By collecting environmental data from multiple sensors, more accurate and stable environmental data can be obtained. State prediction using multiple measurements combined with their corresponding time values avoids data loss and performance degradation due to out-of-order data, and avoids errors in system state prediction caused by abnormal measurements from a single sensor, thus improving the accuracy and reliability of system state updates. Attached Figure Description
[0042] Figure 1 This is a flowchart illustrating a system state update method according to an embodiment of this application.
[0043] Figure 2 This is a schematic diagram of a system status update system as shown in an embodiment of this application. Detailed Implementation
[0044] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.
[0045] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or server that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.
[0046] Example 1:
[0047] Please refer to Figure 1 The system state update method mainly includes steps S1 to S3:
[0048] Step S1 includes: acquiring environmental data collected by multiple sensors within a preset period; the environmental data includes at least measured values and corresponding time values. The multiple sensors can primarily be lidar, millimeter-wave radar, ultrasonic radar, and visual cameras. Those skilled in the art can select other sensors according to actual conditions, and are not limited to these. The environmental data includes, but is not limited to, point cloud data, image data, and obstacle target-level data. The environmental data includes at least measured values and corresponding time values, allowing multiple measured values to be selected based on the time values during subsequent system state prediction. For example, five measured values can be selected based on the time values corresponding to the measured values for system state prediction. In practical applications, those skilled in the art can select different numbers of measured values based on the time values corresponding to the measured values for prediction according to actual conditions.
[0049] Step S2 includes: performing state prediction based on multiple measurements to obtain the predicted system state value.
[0050] Step S3 includes: updating the current system state based on the predicted system state value.
[0051] Among these methods, the MSTSF (Multi-sensor Time Series Fusion) algorithm can be primarily used for state prediction and updating of the current system state based on multiple measurements. The MSTSF algorithm takes the received sensor time series as input, allowing sensor times to be earlier than the current system time. In a single system state update, it uses all data from the time series of multiple sensors for state prediction and updating, thus overcoming the limitations of classic filtering algorithms in handling out-of-order sensor data and abnormal measurements from single sensors. The MSTSF algorithm is also applicable to other nonlinear filters such as EKF (Extended Kalman Filter) and UKF (Unscented Kalman Filter).
[0052] Taking a system state update as an example, multiple millimeter-wave radars and visual cameras are used as sensors to collect data. Environmental data is collected by these multiple millimeter-wave radars and visual cameras. The time series of the measured values and their corresponding time values in the environmental data are used as input, and the MSTSF algorithm is used for state prediction and state update. The predicted system state value is obtained by simultaneously performing state prediction calculations using six measured values, and the current system state is updated based on the predicted system state value.
[0053] Optionally, the state prediction based on multiple measurements includes:
[0054] The current system state value is calculated and obtained based on the first system prediction data and the system state value at time t.
[0055] The covariance matrix of the current system is calculated and obtained based on the prediction data of the first system and the prediction data of the second system.
[0056] The first system prediction data includes at least a system transformation matrix;
[0057] The second system prediction data includes at least the process noise matrix and the covariance matrix at time t, where t is an integer greater than or equal to 1.
[0058] The formula for calculating the current system state value can be:
[0059] X(t+ΔT i )=F(ΔT i X(t), i∈[1,N];
[0060] Where X(t+ΔT) i Let F be the current system state value, F be the system state transition matrix, and X(t) be the system state value at time t, where t+ΔT is the current system state value. i The system state transition matrix represents the time corresponding to the measured values generated at different times, where N is the number of measured values. The matrix primarily represents the transition from time t to t+ΔT. i The relationship between the state transitions of the system at any given moment.
[0061] The formula for calculating the covariance matrix of the current system can be:
[0062] P(t+ΔT i )=F(ΔT i )P(t)F(ΔT i ) T +Q(ΔT i ), i∈[1,N];
[0063] Wherein, P(t+ΔT) i Let P(t) be the current system covariance matrix, and Q(ΔT) be the covariance matrix at time t. i The process noise matrix of the system is denoted as . The current system covariance matrix is used to quantify the uncertainty of state prediction. The process noise matrix represents the current noise characteristics of the system, such as those caused by external disturbances, model errors, or other uncertainties.
[0064] Optionally, obtaining the predicted system state value includes:
[0065] The Kalman gain corresponding to the measured value is calculated based on the first computational data corresponding to the measured value and the covariance matrix at time t; the first computational data includes at least the measurement matrix and measurement noise.
[0066] The predicted covariance matrix is calculated based on the current system covariance matrix, Kalman gain, and the first computational data.
[0067] The predicted system state value is calculated based on the Kalman gain, the current system state value, the measured value, and the first computational data.
[0068] The formula for calculating the Kalman gain can be:
[0069]
[0070] Where K is the Kalman gain corresponding to the measured value z, H is the measurement matrix corresponding to the measured value z, and R is the measurement noise corresponding to the measured value z. The Kalman gain mainly adjusts the predicted system state value by balancing the confidence level of the measured value of the system state prediction, thereby making the predicted system state value more accurate.
[0071] The formula for calculating the predicted system state value can be:
[0072]
[0073] in, Let z be the predicted system state value, z be the measured value, and N be the number of measured values. The number of measured values can be selected according to actual needs; for example, the number of measured values can be 10.
[0074] Optionally, at least one of the measured values in the environmental data corresponds to a time value earlier than the current system state value. Since the MSTSF algorithm allows sensor time to be earlier than the current system time, at least one of the measured values in the environmental data corresponds to a time value earlier than the current system state value. In classic Kalman filtering, when a measured value has a time value earlier than the current system state value, the Kalman filter cannot perform state prediction and needs to extract the measured value with the earlier time value, which can lead to data loss. Using the MSTSF algorithm avoids this data loss.
[0075] Optionally, among the multiple environmental data collected by the same sensor within the preset period, only the measurement values within a preset time range from the current time are retained.
[0076] The preset period can be 1 second or 3 seconds. Those skilled in the art can adjust the preset period according to actual conditions, and are not limited to this. The environmental data is continuously collected data. Once the current time is determined, the measured values within a preset time range from the environmental data collected by the sensor are selected and retained. For example, the preset time range can be 1 second or other time ranges. Those skilled in the art can adjust the preset time range according to actual conditions, and are not limited to this.
[0077] Optionally, a preset period is determined in advance, and target environment data collected by multiple sensors within the preset period is acquired, including:
[0078] The system continuously acquires raw environmental data collected by various sensors, and extracts the target environmental data from the raw environmental data based on the preset period.
[0079] The length of the preset period depends on the real-time requirements and the frequency of data changes. For example, the preset period can be a few seconds, a few minutes, or a longer period of time. By continuously using the raw environmental data collected simultaneously by various sensors, the consistency and synchronization of the data are ensured. The target environmental data obtained from the raw environmental data can be stored in a local database, cloud storage, or distributed storage system for subsequent querying and system status prediction.
[0080] Optionally, the process of updating the current system state also includes handling outliers.
[0081] The outliers include at least disordered measurements and abnormal measurements.
[0082] The measurement value in the environmental data whose time value is earlier than the current system state value is used as the out-of-order measurement value.
[0083] When the deviation of any predicted system state value from other predicted system state values exceeds a preset threshold, the measurement value corresponding to that predicted system state value is determined to be an abnormal measurement value.
[0084] Out-of-order measurements refer to measurements whose time values are earlier than the current system state. These out-of-order measurements can be due to network latency, disordered data transmission order, or other reasons causing time inconsistencies. For example, in the classic Bayesian filtering algorithm, a measurement generated by the system at time t1 may be received by the system at time t3 due to data processing and data transmission delays, thus becoming an out-of-order measurement. The classic Kalman filter discards these out-of-order measurements, leading to information loss and system performance degradation. For instance, assuming the measurement time t + ΔT ≥ t, i.e., ΔT ≥ 0, when the measurement is out-of-order, ΔT < 0. The Kalman filter cannot perform state prediction based on out-of-order measurements, therefore, these out-of-order measurements are discarded. In the system state prediction and update of this application, the out-of-order measurement values are not discarded. The system state is predicted by multiple measurement values, including out-of-order measurement values. By performing system prediction by multiple measurement values at the same time, data loss and performance degradation caused by discarding out-of-order measurement values are avoided.
[0085] When a sensor malfunction causes system state prediction based on generated measurements, and the deviation of the predicted system state value from other predicted system state values exceeds a preset threshold, the measured value is determined to be an abnormal measurement value. The preset threshold can be set to 1; for example, if the deviation of any predicted system state value from other predicted system state values exceeds 1, the measurement value corresponding to that predicted system state value is determined to be an abnormal measurement value. Those skilled in the art can adjust the preset threshold according to actual circumstances, and are not limited to it.
[0086] Optionally, the outlier handling includes:
[0087] The predicted system state value is calculated simultaneously using multiple measurements and out-of-order measurements.
[0088] Weighted operations are performed based on the predicted covariance matrix to suppress the out-of-order and abnormal measurements.
[0089] Using outlier measurements for system state prediction in classic Kalman filters can lead to prediction errors. This application addresses this by using multiple measurements simultaneously, including out-of-order measurements, for system state prediction. During system state updates, a weighted calculation is performed based on the predicted covariance matrix, thereby reducing the impact of outlier and out-of-order measurements on the system state. The number of measurements can be selected based on actual conditions; the more measurements used for system state prediction, the smaller the impact of outlier measurements on prediction errors.
[0090] Example 2:
[0091] Please refer to Figure 2 This application also proposes a system employing the system state update method described in Embodiment 1, which mainly includes:
[0092] A data acquisition module is used to acquire environmental data collected by multiple sensors within a preset period. The environmental data includes at least measured values and corresponding time values. The preset period is set based on the real-time requirements during system state updates; for example, the preset period can be a few seconds or minutes. The multiple sensors can be lidar, millimeter-wave radar, ultrasonic radar, or a vision camera. Those skilled in the art can select other sensors according to the actual situation, and are not limited to these. The environmental data includes at least measured values and corresponding time values to facilitate subsequent system state prediction and updates based on multiple measured values selected from the time values.
[0093] The state prediction module is used to predict the system state based on multiple measurements to obtain the predicted system state value. By simultaneously performing system state prediction using multiple measurements, errors caused by abnormal measurements can be effectively avoided. These multiple measurements may include out-of-order and abnormal measurements; however, state prediction based on these multiple measurements effectively suppresses state errors caused by predictions based on abnormal measurements. The number of measurements used for system prediction can be selected according to actual conditions; for example, six measurements can be selected for state prediction to obtain the predicted system state value. The more measurements selected, the stronger the suppression effect on abnormal measurements.
[0094] In addition, a state update module is provided to update the current system state based on the predicted system state value. The state prediction module and the state update module can employ the MSTSF algorithm, thereby avoiding the data loss problem caused by discarding out-of-order measurements in the classic Kalman filter, as well as the system prediction error problem caused by abnormal measurements from a single sensor.
[0095] Example 3:
[0096] This application also proposes a computer-readable storage medium, the computer-readable storage medium comprising:
[0097] The computer-readable storage medium stores computer-executable instructions.
[0098] When the computer-executable instructions are executed by the control processor, the system state update method described in Embodiment 1 is implemented.
[0099] The computer-readable storage medium can be implemented entirely or partially by software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented entirely or partially as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in the computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)).
[0100] In summary, this application proposes a system state update method, system, and storage medium, aiming to solve the problems of data loss caused by the inability of classic multi-sensor data fusion algorithms to handle out-of-order data, and the increased state prediction error due to abnormal measurements from a single sensor. By collecting environmental data from multiple sensors, more accurate and stable environmental data can be obtained. State prediction using multiple measurements combined with their corresponding time values avoids data loss and performance degradation caused by out-of-order data, and avoids errors in system state prediction caused by abnormal measurements from a single sensor, thus improving the accuracy and reliability of system state updates.
[0101] In the several embodiments provided in this application, it will be understood that each block in the flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those shown in the figures. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved.
[0102] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause an electronic device to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0103] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of this application. It should be understood that the above descriptions are merely specific embodiments of this application and are not intended to limit the scope of protection of this application. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application for those skilled in the art.
Claims
1. A system state update method, characterized in that, include: Acquire environmental data collected by multiple sensors within a preset period; The environmental data includes at least the measured values and their corresponding time values; State prediction is performed based on multiple measurements to obtain the predicted system state value; The current system state is updated based on the predicted system state value.
2. The system state update method according to claim 1, characterized in that, The state prediction based on multiple measurements includes: The current system state value is calculated and obtained based on the first system prediction data and the system state value at time t. The covariance matrix of the current system is calculated and obtained based on the prediction data of the first system and the prediction data of the second system. The first system prediction data includes at least a system transformation matrix; The second system prediction data includes at least the process noise matrix and the covariance matrix at time t, where t is an integer greater than or equal to 1.
3. The system state update method according to claim 2, characterized in that, The process of obtaining the predicted system state value includes: The Kalman gain corresponding to the measured value is calculated based on the first computational data corresponding to the measured value and the covariance matrix at time t; the first computational data includes at least the measurement matrix and measurement noise; The predicted covariance matrix is calculated based on the current system covariance matrix, Kalman gain, and the first computational data; the predicted system state value is calculated based on the Kalman gain, the current system state value, the measured value, and the first computational data.
4. The system state update method according to claim 1, characterized in that, The environmental data measurements include at least one measurement whose time value is earlier than the current system state value.
5. The system state update method according to claim 1, characterized in that, Of the multiple environmental data collected by the same sensor within the preset period, only the measurement values within the preset time range from the current time are retained.
6. The system state update method according to claim 1, characterized in that, A preset period is determined in advance, and target environment data collected by multiple sensors within the preset period are acquired, including: The system continuously acquires raw environmental data collected by various sensors, and extracts the target environmental data from the raw environmental data based on the preset period.
7. The system state update method according to claim 1, characterized in that, The process of updating the current system state also includes handling outliers; The outliers include at least disordered measurements and abnormal measurements; The environmental data whose time value is earlier than the current system state value is used as the out-of-order measurement value; When the deviation of any predicted system state value from other predicted system state values exceeds a preset threshold, the measurement value corresponding to that predicted system state value is determined to be an abnormal measurement value.
8. The system state update method according to claim 7, characterized in that, The outlier handling includes: The predicted system state value is calculated simultaneously using multiple measurements and out-of-order measurements. Weighted operations are performed based on the predicted covariance matrix to suppress the out-of-order and abnormal measurements.
9. A system based on the system state update method according to any one of claims 1-8, characterized in that, The system includes: a data acquisition module, used to acquire environmental data collected by multiple sensors within a preset period; the environmental data includes at least a measured value and a corresponding time value; The state prediction module is used to predict the state based on multiple measurements in order to obtain the predicted system state value. And a state update module, used to update the current system state based on the predicted system state value.
10. A computer-readable storage medium storing computer-executable instructions, characterized in that, When the computer-executable instructions are executed by the control processor, they implement the system state update method as described in any one of claims 1-8.