Unmanned sweeper indoor positioning method based on UWB and IMU

Abstract

The invention provides an unmanned sweeper indoor positioning method based on UWB and IMU. The method comprises the following steps: acquiring UWB base station deployment information of an indoor environment, UWB ranging data of an unmanned sweeper and IMU motion parameters; based on the UWB base station deployment information, performing fusion positioning operation on the UWB ranging data and IMU motion parameters to generate a positioning result; according to the method, the coupling error of the UWB multipath effect and IMU drift is effectively solved, the motion state is accurately modeled through federated filtering, high-confidence centimeter-level positioning is realized in combination with map periodic correction, compared with a traditional scheme, the error is greatly reduced, the problems of path offset, missing sweeping or repeated sweeping are thoroughly avoided, and a reliable position reference is provided for path planning.

Description

Technical Field

[0001] This invention relates to the field of automation control technology, and in particular to an indoor positioning method for unmanned cleaning vehicles based on UWB and IMU. Background Technology

[0002] In the cross-disciplinary field of automation control and sensor fusion, the autonomous operation capability of unmanned sweeping vehicles is highly dependent on indoor high-precision positioning technology. Its positioning performance directly determines the accuracy of sweeping path planning, the completeness of operation coverage, and the safety of operation.

[0003] Current indoor positioning technologies for unmanned cleaning vehicles mainly rely on UWB positioning technology, IMU inertial navigation technology, or traditional multi-sensor fusion solutions, but all of these have significant limitations and together constitute the core bottleneck for technological development. Firstly, UWB positioning technology suffers from multipath interference limitations. While UWB technology can achieve high-precision ranging with its nanosecond-level narrow pulses, in complex indoor environments, signals are easily reflected and diffracted by obstacles such as walls and furniture, creating multipath interference. This leads to significant fluctuations in ranging errors, ranging from 0.3 to 1.2 meters in practical applications. This dynamically changing error cannot provide a stable position reference for the sweeper, easily causing path deviation, missed sweeps, or repeated sweeping, severely affecting operational accuracy.

[0004] Secondly, there is the drift accumulation defect of IMU inertial navigation. IMU sensors can capture motion parameters such as speed, attitude, and acceleration of the sweeper in real time, enabling short-term motion state estimation. However, due to the characteristics of the device, its measurement error accumulates rapidly over time. Experimental data shows that when relying solely on IMU for positioning, the cumulative error can reach 1.5 meters within 10 seconds. After long-term operation, the positioning results will become completely invalid, failing to meet the positioning requirements of continuous sweeper operation. Summary of the Invention

[0005] In view of this, the embodiments of the present invention aim to provide an indoor positioning method for unmanned cleaning vehicles based on UWB and IMU, so as to solve or alleviate the technical problems existing in the prior art, and at least provide a beneficial option.

[0006] To address the aforementioned technical problems, this application adopts the following technical solution: providing an indoor positioning method for unmanned cleaning vehicles based on UWB and IMU, comprising the following steps: Acquire UWB base station deployment information for indoor environments, UWB ranging data of unmanned cleaning vehicles, and IMU motion parameters; Based on the UWB base station deployment information, the UWB ranging data and IMU motion parameters are fused for positioning to generate a positioning result; The UWB base station deployment information includes base station coordinates and positioning coverage area; the UWB ranging data includes signal strength (RSSI) and distance measurement value; and the IMU motion parameters include velocity, attitude and acceleration data. Based on the UWB base station deployment information, the fusion positioning operation is performed on the UWB ranging data and IMU motion parameters to generate the positioning result, including: Based on the time protocol, data timestamp alignment between UWB and IMU is achieved, and a unified spatiotemporal benchmark is constructed. Based on the UWB signal strength, the fusion weights of UWB and IMU are dynamically allocated, and the initial fusion of multi-source data is completed through a federated filtering architecture to obtain the initial localization result. A pre-built indoor environmental semantic map is introduced as a constraint condition, and the initial positioning result is periodically corrected and optimized to obtain the final positioning result; The dynamic allocation of fusion weights based on the UWB signal strength includes: In response to the UWB signal strength meeting the first signal threshold, UWB is assigned a dominant weight, and IMU is assigned an auxiliary weight; In response to the UWB signal strength being lower than the second signal threshold, the system automatically switches to IMU-dominated mode to increase the weighting of the IMU. In response to the interruption of the UWB signal, the positioning validity is maintained for a preset duration.

[0007] As a further preferred embodiment of this technical solution: the time protocol adopts the IEEE 1588PTP protocol, and the timestamp alignment error is determined by the formula... Calculation, where For UWB data timestamps, For IMU data timestamps, the requirements are as follows: ; The construction of the unified spatiotemporal benchmark also includes spatial coordinate transformation, which is achieved through a homogeneous transformation matrix:

[0008] in It is a 3×3 rotation matrix. The translation vector is 3×1, representing the motion parameters of the body coordinate system output by the IMU. Transform to UWB global coordinate system The conversion relationship is as follows Conversion accuracy meets , These are the true global coordinates.

[0009] As a further preferred embodiment of this technical solution: the fusion weights are dynamically allocated through a piecewise function, as shown in the formula:

[0010] in, For UWB fusion weights, The fusion weights of the IMU, and satisfying .

[0011] As a further preferred embodiment of this technical solution: the federated filtering architecture adopts a 15-dimensional state vector, defined as follows: The first three dimensions represent position, the middle three dimensions represent velocity, the next three dimensions represent attitude angle, and the last six dimensions represent sensor deviation. The main filter of the federated filtering uses information allocation coefficients. To achieve optimal global state estimation, the formula is: ,in This is the local estimation result for the UWB sub-filter. This represents the local estimation results for the IMU sub-filters; Process noise covariance matrix Based on the speed of the unmanned cleaning vehicle Dynamic adjustment, the formula is:

[0012] It is the noise covariance matrix of the basic process.

[0013] As a further preferred embodiment of this technical solution: the indoor environment semantic map includes geometric features such as walls and obstacles, as well as spatial constraint information; the correction algorithm adopts an improved iterative nearest-point method; and the matching error is calculated using the formula...

[0014] in, For the number of point clouds in the lidar, The current point cloud coordinates, Given the coordinates of feature points to be matched in the map, the error of a single match is required. ; The triggering condition for the periodic correction optimization is determined by the formula. Define, where This is the initial localization result. This is the theoretical location under map constraints. When this condition is met, the correction process is initiated to suppress IMU cumulative drift.

[0015] As a further preferred embodiment of this technical solution, the UWB signal interruption handling strategy includes: In response to a short interruption of the UWB signal (duration < 2s), the IMU weight is adaptively increased to 0.9, and the zero-speed update algorithm is activated to suppress inertial device drift, keeping the positioning error within the preset short-term error range; In response to a long-term interruption in the UWB signal, the map matching module is activated, and the spatial constraints of the indoor environmental semantic map are used to correct the trajectory calculated by the IMU. The positioning error is stably controlled within the preset long-term error range, wherein the duration of the long-term interruption in the UWB signal is ≤10s.

[0016] As a further preferred embodiment of this technical solution, a low-power optimization step is also included: The sampling frequency of UWB and IMU is dynamically adjusted based on the motion state of the unmanned sweeper. The sampling frequency is reduced when stationary or moving at low speed, and high frequency sampling is maintained when moving at high speed. The federated filtering algorithm is optimized for lightweighting to reduce computational load and keep the total power consumption of the positioning module below 2W, thus meeting the continuous operation and battery life requirements of the unmanned sweeper.

[0017] As a further preferred embodiment of this technical solution: the zero-speed update algorithm is triggered when the speed of the unmanned sweeper is lower than a preset speed threshold. By detecting the IMU data deviation in a stationary state, the cumulative error of the accelerometer and gyroscope is corrected in real time.

[0018] As a further preferred embodiment of this technical solution, the deployment of the UWB base station must meet the following requirements: Deploy at least four UWB base stations at key locations in the positioning area to form a rectangular or polygonal positioning network; The coordinates of the base stations are calibrated using a laser tracker to ensure that the three-dimensional coordinate errors of each base station are controlled within the preset calibration error range; The baseline distance between adjacent base stations is set reasonably according to the size of the positioning area to ensure that there are no blind spots in positioning coverage.

[0019] As a further preferred embodiment of this technical solution, the positioning result satisfies the following performance indicators: Its positioning accuracy is suitable for both open indoor spaces and complex non-line-of-sight environments, effectively suppressing errors caused by multipath effects and inertial drift, and achieving a high-confidence centimeter-level positioning effect.

[0020] The embodiments of the present invention have the following advantages due to the adoption of the above technical solutions: I. This invention effectively resolves the coupling error between UWB multipath effect and IMU drift. By accurately modeling motion state through federated filtering and combining it with periodic map correction, it achieves high-confidence centimeter-level positioning, significantly reducing errors compared to traditional solutions and completely avoiding path deviation, missed scans, or repeated scanning problems, thus providing a reliable position benchmark for path planning.

[0021] Second, this invention dynamically allocates fusion weights based on UWB signal status to achieve adaptive switching between line-of-sight and non-line-of-sight scenarios, maintaining stable positioning in complex interference environments. For UWB signal interruptions, a tiered response strategy ensures positioning effectiveness, coupled with module-level fault isolation capabilities of federated filtering to avoid the impact of single sensor anomalies, maintaining high positioning continuity in scenarios such as dynamic occlusion.

[0022] Third, this invention achieves low-power operation of the positioning module by dynamically adjusting the sensor sampling frequency and combining it with lightweight algorithm optimization. This perfectly matches the battery capacity of the unmanned sweeper and supports long-term continuous operation. It does not affect the power supply of the main cleaning function and solves the core contradiction between high-precision positioning and long battery life.

[0023] Fourth, this invention significantly improves cleaning coverage and operational safety, reduces missed areas and equipment collisions, and lowers maintenance costs and failure rates. The base station is flexibly deployable, adaptable to various indoor scenarios, and can be implemented without complex modifications. It also features an automatic calibration and adjustment mechanism to mitigate performance degradation issues over long-term use, facilitating the large-scale application of unmanned cleaning vehicles.

[0024] The above overview is for illustrative purposes only and is not intended to be limiting in any way. In addition to the illustrative aspects, embodiments, and features described above, further aspects, embodiments, and features of the invention will become readily apparent from the accompanying drawings and the following detailed description. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a schematic diagram of UWB and IMU weight allocation and positioning error under different RSSI values ​​in an embodiment of the present invention; Figure 3 This is a bar chart comparing the positioning errors in two scenarios according to an embodiment of the present invention. Figure 4 This is a schematic diagram illustrating the proportion of the state vector dimension in an embodiment of the present invention. Detailed Implementation

[0027] The embodiments of this disclosure will now be described in detail with reference to the accompanying drawings.

[0028] It should be understood that the following specific examples illustrate the implementation of this disclosure, and those skilled in the art can easily understand other advantages and effects of this disclosure from the content disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of this disclosure, and not all of them. This disclosure can also be implemented or applied through other different specific implementation methods, and the details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of this disclosure. It should be noted that, in the absence of conflict, the following embodiments and features in the embodiments can be combined with each other. Based on the embodiments in this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.

[0029] like Figure 1 As shown, this invention provides an indoor positioning method for unmanned cleaning vehicles based on UWB and IMU, including the following steps: Acquire UWB base station deployment information for indoor environments, UWB ranging data of unmanned cleaning vehicles, and IMU motion parameters; Based on UWB base station deployment information, a fusion positioning operation is performed on UWB ranging data and IMU motion parameters to generate positioning results; Among them, the UWB base station deployment information includes the base station coordinates and positioning coverage area, the UWB ranging data includes the signal strength (RSSI) and distance measurement value, and the IMU motion parameters include speed, attitude and acceleration data. Based on UWB base station deployment information, a fusion positioning operation is performed on UWB ranging data and IMU motion parameters to generate positioning results, including: Based on the time protocol, data timestamp alignment between UWB and IMU is achieved, and a unified spatiotemporal benchmark is constructed. The time protocol adopts the IEEE 1588PTP protocol, and the timestamp alignment error is determined by the formula. Calculation, where For UWB data timestamps, For IMU data timestamps, the requirements are as follows: ; The construction of a unified spatiotemporal reference also includes spatial coordinate transformation, which is achieved through a homogeneous transformation matrix:

[0030] in It is a 3×3 rotation matrix. The translation vector is 3×1, representing the motion parameters of the body coordinate system output by the IMU. Transform to UWB global coordinate system The conversion relationship is as follows Conversion accuracy meets , Use the true global coordinates; Based on the dynamic allocation of fusion weights between UWB and IMU according to UWB signal strength, the initial fusion of multi-source data is completed through a federated filtering architecture to obtain the initial localization result; The fusion weights are dynamically allocated using a piecewise function, as shown in the formula:

[0031] in, For UWB fusion weights, The fusion weights of the IMU, and satisfying .

[0032] The federated filtering architecture uses a 15-dimensional state vector, defined as follows: The first three dimensions represent position, the middle three dimensions represent velocity, the next three dimensions represent attitude angle, and the last six dimensions represent sensor deviation. The main filter of the federated filtering uses information allocation coefficients. To achieve optimal global state estimation, the formula is: ,in This is the local estimation result for the UWB sub-filter. This represents the local estimation results for the IMU sub-filters; Process noise covariance matrix Based on the speed of the unmanned cleaning vehicle Dynamic adjustment, the formula is:

[0033] It is the noise covariance matrix of the basic process.

[0034] A pre-built semantic map of the indoor environment is introduced as a constraint condition, and the initial positioning results are periodically corrected and optimized to obtain the final positioning results; The indoor environmental semantic map includes geometric features such as walls and obstacles, as well as spatial constraint information. The correction algorithm employs an improved iterative nearest-point method, and the matching error is calculated using the formula...

[0035] in, For the number of point clouds in the lidar, The current point cloud coordinates, Given the coordinates of feature points to be matched in the map, the error of a single match is required. ; The triggering condition for periodic correction optimization is expressed by the formula. Define, where This is the initial localization result. This is the theoretical location under map constraints. When this condition is met, the correction process is initiated to suppress IMU cumulative drift.

[0036] The dynamic allocation of fusion weights based on UWB signal strength includes: In response to the UWB signal strength meeting the first signal threshold, UWB is assigned a dominant weight, and IMU is assigned an auxiliary weight. In response to the UWB signal strength falling below the second signal threshold, it automatically switches to IMU-dominated mode, increasing the weight of the IMU. In response to UWB signal interruption, maintain positioning validity for a preset duration; The handling strategies for UWB signal interruptions include: In response to a short interruption of the UWB signal (duration < 2s), the IMU weight is adaptively increased to 0.9, and the zero-speed update algorithm is activated to suppress inertial device drift, keeping the positioning error within the preset short-term error range; The zero-speed update algorithm is triggered when the speed of the unmanned sweeper is lower than a preset speed threshold. By detecting the deviation of IMU data in a stationary state, the cumulative error of the accelerometer and gyroscope is corrected in real time.

[0037] In response to a long-term interruption in the UWB signal, the map matching module is activated, and the spatial constraints of the indoor environmental semantic map are used to correct the trajectory calculated by the IMU. The positioning error is stably controlled within the preset long-term error range, wherein the duration of the long-term interruption in the UWB signal is ≤10s.

[0038] Furthermore, the method of the present invention also includes a low-power optimization step: The sampling frequency of UWB and IMU is dynamically adjusted based on the motion state of the unmanned sweeper. The sampling frequency is reduced when stationary or moving at low speed, and high frequency sampling is maintained when moving at high speed. The federated filtering algorithm is optimized for lightweighting to reduce computational load and keep the total power consumption of the positioning module below 2W, thus meeting the continuous operation and battery life requirements of the unmanned sweeper.

[0039] Specifically, UWB base station deployment must meet the following requirements: Deploy at least four UWB base stations at key locations in the positioning area to form a rectangular or polygonal positioning network; The coordinates of the base stations are calibrated using a laser tracker to ensure that the three-dimensional coordinate errors of each base station are controlled within the preset calibration error range; The baseline distance between adjacent base stations is set reasonably according to the size of the positioning area to ensure that there are no blind spots in positioning coverage.

[0040] The positioning results meet the following performance indicators: Its positioning accuracy is suitable for both open indoor spaces and complex non-line-of-sight environments, effectively suppressing errors caused by multipath effects and inertial drift, and achieving a high-confidence centimeter-level positioning effect.

[0041] The present invention also provides an application example of implementing the method steps according to the present invention: Example of an open area in a shopping mall: (I) System Deployment and Equipment Configuration Within a 15m x 20m open positioning area in the shopping mall, four UWB base stations were deployed according to the principle of "rectangular full coverage," with an 8m horizontal baseline distance and a 6m vertical baseline distance between adjacent base stations, forming a blind-spot-free positioning network. The three-dimensional coordinates of the base stations were calibrated using a laser tracker, with the error controlled to ≤0.1m, providing a high-precision spatial reference for positioning.

[0042] The unmanned cleaning vehicle is equipped with a combination unit of UWB tags and IMU sensors: the UWB tags are sampled at a frequency of 10Hz to collect signal strength (RSSI) and distance measurements; the IMU uses 100Hz high-frequency sampling to capture speed, attitude, and acceleration data. A semantic map of the mall's indoor environment, including geometric features such as walls and columns, is pre-stored to provide constraints for subsequent calibration.

[0043] (II) Spatiotemporal registration and data preprocessing UWB and IMU timestamp alignment is achieved using the IEEE 1588PTP protocol, through the formula... To control the alignment error to ≤5ms, a unified spatiotemporal benchmark is established. Spatial coordinate transformation is achieved through a homogeneous transformation matrix:

[0044] IMU body coordinate system parameters Transform to UWB global coordinate system Conversion accuracy meets .

[0045] Data preprocessing stage: Kalman filtering is used on the raw IMU data to suppress drift and noise; the 3σ rule is used to remove outliers from the UWB data, and the RSSI value of each frame of data is recorded synchronously. Figure 2 The dynamic weight allocation shown provides input.

[0046] (iii) Dynamic fusion of federated filtering, such as Figure 2 , Figure 4 As shown: 1. State Vector Definition: A 15-dimensional state vector is used. ,like Figure 4As shown, the dimensions include 3D position, 3D velocity, 3D attitude, and 6D sensor bias. The sensor bias dimension accounts for the highest proportion (6 / 15), while position, velocity, and attitude each account for 20% (3 / 15), comprehensively covering kinematic and sensor characteristics.

[0047] 2. State Prediction and Observation Update: State prediction is performed based on high-frequency IMU data, and the output is... The prediction results are corrected using UWB ranging values ​​and RSSI data to obtain the local estimate of the UWB sub-filter. Process noise covariance matrix Dynamically adjusts with speed. hour It is suitable for high-speed motion scenarios.

[0048] 3. Dynamic weight allocation: Weights are allocated based on UWB signal strength using a piecewise function, ensuring perfect matching. Figure 2 The schematic data shown is as follows: When RSSI = -75dB (line-of-sight environment, good signal), , ,correspond Figure 2 The positioning error is 0.22m. When personnel movement causes multipath interference and the RSSI drops to -90dB, the weights are adjusted to... , ,correspond Figure 2 The positioning error is 0.28m. The main filter assigns coefficients through information allocation. Calculate global estimate

[0049] The initial positioning results are obtained.

[0050] (iv) Map correction and optimization When the initial positioning result and the map feature deviation satisfy When this occurs, the correction process is triggered. An improved ICP algorithm is used to calculate the matching error.

[0051] Controlling single-match error This effectively suppresses IMU cumulative drift.

[0052] (v) Performance verification, such as Figure 3 As shown: In a 30-minute continuous test, the cumulative error of this method was 0.45m. Combined with... Figure 3It can be seen that in open shopping mall scenarios, the average positioning accuracy of this method is 0.25m, which is significantly better than the 0.58m of the traditional method; the 95% confidence interval error is 0.32m, which is 62.4% higher than the traditional method (0.85m). The average power consumption of the system is 1.8W, and the peak power consumption is ≤2.0W, which meets the low power consumption requirements.

[0053] This invention employs a three-layer architecture of spatiotemporal synchronization, dynamic fusion, and map correction to effectively resolve the coupling error between UWB multipath effects and IMU drift. Through federated filtering of 15-dimensional state vectors, it accurately models the motion state, combined with periodic map correction using an improved ICP algorithm, achieving high-confidence centimeter-level positioning accuracy. This reduces errors by more than 50% compared to traditional single-sensor or fixed-weight fusion schemes, completely avoiding path offset, missed scans, or redundant scanning issues, and providing a reliable positional reference for path planning.

[0054] This invention dynamically allocates fusion weights based on UWB signal strength to achieve adaptive switching between line-of-sight and non-line-of-sight scenarios, maintaining positioning stability even in multipath interference environments. For UWB signal interruption scenarios, a tiered response strategy (zero-rate update algorithm for short interruptions, map constraints for medium- to long-term interruptions) extends the effective positioning duration to 10 seconds, with controllable cumulative errors during interruptions. The module-level fault isolation capability of federated filtering avoids the impact of single sensor anomalies on positioning results, achieving over 99.5% positioning continuity in complex scenarios such as dynamic occlusion and device movement.

[0055] This invention dynamically adjusts the sensor sampling frequency based on motion status (reducing the frequency when stationary / low speed and maintaining the frequency when high speed), combined with lightweight optimization of the federated filtering algorithm, to control the total power consumption of the positioning module to within 2W, which is 43% lower than the traditional solution. It perfectly matches the battery capacity of the unmanned sweeper, supports 8 hours of continuous operation, avoids affecting the power supply of the main cleaning function, and solves the core contradiction between high precision and long battery life.

[0056] The improved positioning accuracy of this invention increases cleaning coverage by 15% and reduces missed areas; enhanced environmental robustness reduces equipment collision rate by 80%, significantly reducing maintenance costs and equipment failure rate. Meanwhile, the flexible base station deployment (supporting rectangular / polygonal networks) adapts to different indoor scenarios such as shopping malls and offices, requiring no complex modifications for deployment. It also features automatic calibration and adjustment mechanisms, reducing the impact of performance degradation over long-term use and promoting the large-scale application of unmanned cleaning vehicles in complex indoor environments.

[0057] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this disclosure to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. An indoor positioning method for unmanned cleaning vehicles based on UWB and IMU, characterized in that, Includes the following steps: Acquire UWB base station deployment information for indoor environments, UWB ranging data of unmanned cleaning vehicles, and IMU motion parameters; Based on the UWB base station deployment information, the UWB ranging data and IMU motion parameters are fused for positioning to generate a positioning result; The UWB base station deployment information includes base station coordinates and positioning coverage area; the UWB ranging data includes signal strength (RSSI) and distance measurement value; and the IMU motion parameters include velocity, attitude and acceleration data. Based on the UWB base station deployment information, the fusion positioning operation is performed on the UWB ranging data and IMU motion parameters to generate the positioning result, including: Based on the time protocol, data timestamp alignment between UWB and IMU is achieved, and a unified spatiotemporal benchmark is constructed. Based on the UWB signal strength, the fusion weights of UWB and IMU are dynamically allocated, and the initial fusion of multi-source data is completed through a federated filtering architecture to obtain the initial localization result. A pre-built indoor environmental semantic map is introduced as a constraint condition, and the initial positioning result is periodically corrected and optimized to obtain the final positioning result; The dynamic allocation of fusion weights based on the UWB signal strength includes: In response to the UWB signal strength meeting the first signal threshold, UWB is assigned a dominant weight, and IMU is assigned an auxiliary weight; In response to the UWB signal strength being lower than the second signal threshold, the system automatically switches to IMU-dominated mode to increase the weighting of the IMU. In response to the interruption of the UWB signal, the positioning validity is maintained for a preset duration.

2. The indoor positioning method for unmanned cleaning vehicles based on UWB and IMU according to claim 1, characterized in that: The time protocol adopts the IEEE 1588PTP protocol, and the timestamp alignment error is determined by the formula. Calculation, where For UWB data timestamps, For IMU data timestamps, the requirements are as follows: ; The construction of the unified spatiotemporal benchmark also includes spatial coordinate transformation, which is achieved through a homogeneous transformation matrix: ； in It is a 3×3 rotation matrix. The translation vector is 3×1, representing the motion parameters of the body coordinate system output by the IMU. Transform to UWB global coordinate system The relationship is transformed into The conversion accuracy meets the requirements. , These are the true global coordinates.

3. The indoor positioning method for unmanned cleaning vehicles based on UWB and IMU according to claim 2, characterized in that: The fusion weights are dynamically allocated using a piecewise function, as shown in the formula: ； in, For UWB fusion weights, The fusion weights of the IMU, and satisfying .

4. The indoor positioning method for unmanned cleaning vehicles based on UWB and IMU according to claim 1, characterized in that: The federated filtering architecture uses a 15-dimensional state vector, defined as follows: The first three dimensions represent position, the middle three dimensions represent velocity, the next three dimensions represent attitude angle, and the last six dimensions represent sensor deviation. The main filter of the federated filtering uses information allocation coefficients. To achieve optimal global state estimation, the formula is: ,in This is the local estimation result for the UWB sub-filter. This represents the local estimation results for the IMU sub-filters; Process noise covariance matrix Based on the speed of the unmanned cleaning vehicle Dynamic adjustment, the formula is: ； It is the noise covariance matrix of the basic process.

5. The indoor positioning method for unmanned cleaning vehicles based on UWB and IMU according to claim 1, characterized in that: The indoor environment semantic map includes geometric features such as walls and obstacles, as well as spatial constraint information. The correction algorithm employs an improved iterative nearest-point method, and the matching error is calculated using the formula... ； in, For the number of point clouds in the lidar, The current point cloud coordinates, Given the coordinates of feature points to be matched in the map, the error of a single match is required. ; The triggering condition for the periodic correction optimization is determined by the formula. Define, where This is the initial localization result. This is the theoretical location under map constraints. When this condition is met, the correction process is initiated to suppress IMU cumulative drift.

6. The indoor positioning method for unmanned cleaning vehicles based on UWB and IMU according to claim 1, characterized in that: The handling strategy for UWB signal interruption includes: In response to a short interruption of the UWB signal, the IMU weight is adaptively increased to 0.9, and a zero-rate update algorithm is activated to suppress inertial device drift, keeping the positioning error within a preset short-term error range; In response to a long-term interruption in the UWB signal, the map matching module is activated, and the spatial constraints of the indoor environmental semantic map are used to correct the trajectory calculated by the IMU, so that the positioning error is stably controlled within the preset long-term error range.

7. The indoor positioning method for unmanned cleaning vehicles based on UWB and IMU according to claim 1, characterized in that: It also includes low-power optimization steps: The sampling frequency of UWB and IMU is dynamically adjusted based on the motion state of the unmanned sweeper. The sampling frequency is reduced when stationary or moving at low speed, and high frequency sampling is maintained when moving at high speed. The federated filtering algorithm is optimized for lightweighting to reduce computational load and keep the total power consumption of the positioning module below 2W, thus meeting the continuous operation and battery life requirements of the unmanned sweeper.

8. The indoor positioning method for unmanned cleaning vehicles based on UWB and IMU according to claim 1, characterized in that: The zero-speed update algorithm is triggered when the speed of the unmanned sweeper is lower than a preset speed threshold. By detecting the deviation of IMU data in a stationary state, the cumulative error of the accelerometer and gyroscope is corrected in real time.

9. The indoor positioning method for unmanned cleaning vehicles based on UWB and IMU according to claim 1, characterized in that: The deployment of the UWB base station must meet the following requirements: Deploy at least four UWB base stations at key locations in the positioning area to form a rectangular or polygonal positioning network; The coordinates of the base stations are calibrated using a laser tracker to ensure that the three-dimensional coordinate errors of each base station are controlled within the preset calibration error range; The baseline distance between adjacent base stations is set reasonably according to the size of the positioning area to ensure that there are no blind spots in positioning coverage.

10. The indoor positioning method for unmanned cleaning vehicles based on UWB and IMU according to claim 1, characterized in that: The positioning results meet the following performance indicators: Its positioning accuracy is suitable for both open indoor spaces and complex non-line-of-sight environments, effectively suppressing errors caused by multipath effects and inertial drift, and achieving a high-confidence centimeter-level positioning effect.

Images