Acoustic synchronous localization and mapping method based on robust potential probability hypothesis density

Abstract

The invention discloses an acoustic synchronous localization and mapping method based on robust potential probability hypothesis density. The acoustic synchronous localization and mapping method specifically comprises the following steps: performing recursive propagation on a sound source state and the potential probability hypothesis density of DoA observation data; uniformly distributing the distance minimum value and the distance maximum value from the sound source to the robot; the DoA observation data and the sound source state missing distance information are synthesized into a mean value of sound source state birth potential probability hypothesis density; updating the mean value and the covariance of the sound source state potential probability hypothesis density by adopting an unscented Kalman filter; calculating an approximate potential probability hypothesis density likelihood probability by using a variational Bayesian method; a single feature strategy is adopted to generate a robot pose particle weight, and resampling operation is performed on the robot pose particle weight; and trimming and combining the updated mean value and covariance of the potential probability hypothesis density of the sound source state, and when the weight of the potential probability hypothesis density is greater than a set threshold value, determining that the mean value of the corresponding potential probability hypothesis density is the position of the sound source, and determining that the weight number of the potential probability hypothesis density meeting the condition is the number of the sound sources.

Description

Technical Field

[0001] The present invention relates to the field of acoustic synchronous positioning, and in particular to an acoustic synchronous positioning and map construction method based on robust potential probability hypothesis density. Background Art

[0002] In the past few decades, with the development of technology and the increase in demand, Simultaneous Localization and Mapping (SLAM) technology has become a research hotspot and has been widely used in many fields. These application scenarios include mobile robots, navigation, drones, augmented reality (AR), virtual reality (VR) and self-driving cars. The core goal of SLAM is to achieve self-localization of robots or vehicles in unknown environments and build environmental maps through observation data obtained by sensors. SLAM observation data is obtained through on-board sensors for environmental measurement and vehicle positioning, so the positioning performance is closely related to sensor characteristics. Visual sensors often have difficulty identifying targets in sudden changes in light, darkness or occlusion conditions, and lidar data collection is also inaccurate under dusty and foggy conditions, resulting in a decrease in SLAM performance. It is worth noting that traditional SLAM algorithms usually rely on Gaussian noise models, that is, assuming that the noise distribution of observation data is Gaussian. However, in actual scenarios, observation noise is often affected by a variety of complex factors, which makes the measurement noise often deviate from the Gaussian distribution, showing non-Gaussianity or even long-tail distribution characteristics. Traditional algorithms have difficulty in handling data anomalies under these non-Gaussian noises, resulting in a decrease in the accuracy and robustness of the SLAM system.

[0003] In the prior art, such as in the paper "Acoustic simultaneous localization and mapping (a-SLAM) of a moving microphone array and its surrounding speakers", Christine et al. proposed a novel acoustic simultaneous localization and mapping (a-SLAM) method, which uses a moving microphone array to accurately locate the position of the sound source and sensor in a real environment. The proposed method uses a single cluster probability hypothesis density (SC-PHD) filter, which exhibits good robustness in the face of challenges such as reverberation and measurement interference. By utilizing the spatial diversity of the moving array, the system is able to infer the distance between the sound source and the sensor, thereby improving the accuracy of the sound source mapping and the positioning accuracy of the microphone array. This innovative approach effectively solves the difficulties faced by traditional methods in complex environments. Experimental results show that this method has significant improvements in tracking performance and estimation accuracy, fully demonstrating the potential and advantages of acoustic SLAM in practical applications.

[0004] In the paper “A Graph Optimization-Based Acoustic SLAM Edge Computing System Offering Centimeter-Level Mapping Services with Reflector Recognition Capability”, Zhou et al. proposed an acoustic SLAM system based on graph optimization. First, the robot moves in an indoor environment while emitting pulsed sound waves, and a four-channel microphone array records the room impulse response (RIR). By extracting the cepstrum features of the RIR, the system can identify echoes from different reflective materials and achieve echo tagging. Then, using the graph optimization method, the acoustic data is combined with the inertial measurement unit (IMU) data to construct a graph model with pose constraints, thereby eliminating the accumulated error during the robot's movement. Finally, by optimizing the nodes and edges in the graph, the system can accurately locate the robot's trajectory and generate a detailed indoor map containing the locations of walls, doors and windows. The advantage of this acoustic SLAM system is that it can work effectively in low-light intensity environments, avoiding the reliance on expensive laser or visual sensors and reducing the cost of the SLAM system. However, the system also has some shortcomings. For example, it may be limited by the propagation characteristics of sound waves in complex indoor environments, resulting in reduced accuracy of map recognition and position estimation. Summary of the invention

[0005] In view of the problems existing in the prior art, the present invention discloses an acoustic synchronous positioning and map construction method based on robust potential probability hypothesis density, which specifically includes the following steps:

[0006] The sound source state and DoA observation data are modeled as random finite sets, where the noise of the DoA observation data obeys the Student T distribution, and the potential probability hypothesis density of the sound source state and DoA observation data is recursively propagated, where the potential probability hypothesis density includes potential distribution, weight, mean and covariance;

[0007] The minimum and maximum distances from the sound source to the robot are evenly distributed, and the random numbers randomly generated by the even distribution are used as the missing distance information of the sound source state;

[0008] The DoA observation data and the missing distance information of the sound source state are combined into the mean value of the potential probability hypothesis density of the sound source state, and the mean value of the potential probability hypothesis density of the sound source state is used as the input information of the potential probability hypothesis density prediction process;

[0009] The unscented Kalman filter based on Student-T distribution is used to update the mean and covariance of the probability hypothesis density of the sound source state potential;

[0010] The variational Bayes method is used to calculate the approximate potential probability hypothesis density likelihood probability, which is used to update the weight of the potential probability hypothesis density;

[0011] A single feature strategy is used to generate the robot pose particle weights, and the robot pose particle weights are weighted averaged with the Rao-Blackwellized filter particles to obtain the robot pose estimation, and the robot pose particle weights are resampled.

[0012] The mean and covariance of the updated potential probability hypothesis density of the sound source state are trimmed and merged. When the weight of the potential probability hypothesis density is greater than the set threshold, the corresponding mean of the potential probability hypothesis density is the sound source position, and the number of weights of the potential probability hypothesis density that meets the conditions is the number of sound sources.

[0013] When the sound source state and DoA observation data are modeled as random finite sets: the number of constructions is N t The source state is a random finite set s t,n Indicates the status of each sound source

[0014]

[0015] Among them, n t,n is the process noise with covariance Q;

[0016]

[0017] The DoA estimation algorithm is used to infer the instantaneous direction of each sound source relative to the robot at time t. The multi-source measurement process is assumed to be the union of the measurement value generated by the target and the clutter caused by the reverberation. The measurement data Including quantity M t DoA observation data and robot measurement data The robot measurement data includes the robot running speed y t,v and direction, y t,γ The DoA observation data noise follows the Student T distribution. The DoA observation data is shown below:

[0018] ω t,m =g(s t,n )+e t,m ,e t,m ~St(0 2×1 ,R t,m ,ν)

[0019] where St(·) is the Student T distribution, g(·) is the Cartesian to spherical transformation, ν is the degree of freedom parameter, and e t,m is the covariance R t,m DoA observation data error;

[0020]

[0021] Among them, K t is a random set of false measurements generated by clutter, and each DoA observation data is Azimuth φ t,m ∈[0,2π) and the elevation angle θ t,m ∈[0,π).

[0022] Furthermore, the random numbers generated by uniform distribution are used as the sound source state. When the distance information is missing: the distance of each unmeasured sound source relative to the robot is estimated by using J b The sound source state is missing distance Initialize the mean value of the birth probability hypothesis density of the i-th particle

[0023]

[0024] in Represents the minimum distance r from the sound source to the robot min and the maximum value r max The mean of the birth probability density is assumed to be Built

[0025]

[0026] Among them, g -1(·) is the transformation from spherical coordinates to Cartesian coordinates.

[0027] Furthermore, in the potential probability hypothesis density prediction step: the Gaussian mixture potential probability hypothesis density is used to estimate the sound source position. The mean of the Gaussian term represents the possible location of the sound source. The predicted potential probability hypothesis density potential distribution is the sum of the birth and survival target potential distributions, that is, the convolution of the birth and survival target potential distributions, Expressed as

[0028]

[0029] in, is the binomial coefficient, P S is the probability of target existence, is the birth potential distribution at time step t, the strength of the prediction It is expressed as follows:

[0030]

[0031] in, Indicates birth intensity, Indicates the strength of the survival goal.

[0032] Furthermore, when the mean and covariance of the sound source state potential probability hypothesis density are updated: when the microphone array receives the sound source signal, each sound source potential probability hypothesis density is updated separately, where the updated potential distribution and update intensity It is expressed as follows:

[0033]

[0034]

[0035] Among them, P D is the target detection probability, Given n targets, multi-target observations Z t The likelihood of DoA observation data noise follows Student T distribution, for z∈Ω t Detection item|Ω t | can be expressed as

[0036]

[0037] Among them, Gamma(·) represents the gamma distribution, α t ~Gamma(ν / 2,ν / 2);

[0038] Given the estimated robot pose The mean of the jth Gaussian component and covariance The unscented Kalman filter is used in the update process of formula (15), and the predicted state is given by the following formula Covariance matrix and the cross-covariance matrix Make an update

[0039]

[0040]

[0041]

[0042] The updated probability hypothesis density mean and covariance are obtained using the following formula;

[0043]

[0044] in, is the Kalman gain.

[0045] Furthermore, when using the variational Bayes method to calculate the approximate potential probability hypothesis density likelihood probability: obtain an approximation of the target likelihood

[0046]

[0047] The variational Bayes method is calculated by minimizing the Kullback-Leibler divergence and

[0048]

[0049]

[0050] Variational Bayesian methods approximate nonlinear probabilities through an iterative process. represents the expectation of the auxiliary random variable, A measure of the difference between the observed and predicted values.

[0051] Furthermore, the robot pose estimation is obtained by weighted averaging the robot pose particle weights and the Rao-Blackwellized filter particles: The robot pose estimation is obtained using the Rao-Blackwell particle filter The robot's posture is represented by particles with different weights in

[0052]

[0053] The single feature strategy is used to generate the robot pose particle weights. The single feature strategy formula is as follows:

[0054]

[0055] in, and Resampling get

[0056] Due to the adoption of the above technical scheme, the present invention provides an acoustic synchronous positioning and map construction method based on robust potential probability hypothesis density, which models the sound source state and azimuth observation data as a random finite set, the DoA observation data noise obeys the Student T distribution, and recursively propagates its probability hypothesis density and potential distribution, and uses the uniform distribution of the minimum and maximum distances from the sound source to the robot as the missing distance information to construct the range of the sound source state; uses Gaussian mixture potential probability hypothesis density (CPHD) to estimate the sound source state, wherein the state update of nonlinear observations is completed by the unscented Kalman filter (UKF); uses the variational Bayes method to approximate the potential probability hypothesis density posterior likelihood; the single feature strategy is used for the particle weighting process of the Rao-Blackwellized filter; and finally, the number and position of the sound source and the motion trajectory of the robot are jointly estimated. This method does not require a complex data association process, can adapt to changes in the number of sound sources, and has good positioning performance and robustness to non-Gaussian outliers. . BRIEF DESCRIPTION OF THE DRAWINGS

[0057] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0058] Figure 1 is a flow chart of the method of the present invention;

[0059] Figure 2 Schematic diagram of acoustic SLAM in the present invention;

[0060] Figure 3 This is a sample diagram of the acoustic SLAM results of the present invention. DETAILED DESCRIPTION

[0061] In order to make the technical solutions and advantages of the present invention more clear, the technical solutions in the embodiments of the present invention are clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention:

[0062] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0063] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0064] A flowchart and schematic diagram of an acoustic simultaneous localization and mapping method based on robust potential probability hypothesis density is shown in Figure 1 and Figure 2 As shown, the specific steps include:

[0065] S1: Model the source state and DoA observation data as a random finite set. The number of constructions is N t The source state is a random finite set s t,n Indicates the status of each sound source

[0066]

[0067] Among them, n t,n is the process noise with covariance Q.

[0068]

[0069] B t Simulating the birth process at time t, P(s t-1,n ) represents the survival target between t-1 and t;

[0070]

[0071] The DoA estimation algorithm is used to infer the instantaneous direction of each sound source relative to the robot at time t. The multi-source measurement process is assumed to be the union of the measurement value generated by the target and the clutter caused by the reverberation. The measurement data Including quantity M t DoA observation data and robot measurement data The robot measurement data includes the robot running speed y t,v and direction y t,γ The DoA observation data noise follows the Student T distribution. The DoA observation data is shown below:

[0072] ω t,m =g(s t,n )+e t,m ,e t,m ~St(0 2×1 ,R t,m ,ν) (4)

[0073] where St(·) is the Student T distribution, g(·) is the Cartesian to spherical transformation, ν is the degree of freedom parameter, and e t,m is the covariance R t,m The DoA observation data error. The Student's T distribution is expressed as follows:

[0074]

[0075] Where d is the DoA observation data dimension, Σ=ω t,m -g(s t,n ).

[0076]

[0077]

[0078] Among them, K t is a random set of false measurements generated by clutter, and each DoA observation data is Azimuth φ t,m ∈[0,2π) and the elevation angle θ t,m ∈[0,π).

[0079] S2: Evenly distribute the minimum and maximum distances from the sound source to the robot, and use the random numbers randomly generated by the uniform distribution as the missing distance information of the sound source state. Estimate the distance of each unmeasured sound source relative to the robot by using J b The sound source state is missing distance Initialize the mean value of the birth probability hypothesis density of the i-th particle

[0080]

[0081] in, Represents the minimum distance r from the sound source to the robot min and the maximum value rmax The mean of the birth probability density is assumed to be Built

[0082]

[0083] Among them, g -1 (·) is the transformation from spherical coordinates to Cartesian coordinates.

[0084] S3: Potential probability hypothesis density prediction step. The Gaussian mixture potential probability hypothesis density is used to estimate the sound source position. The mean of the Gaussian term represents the possible location of the sound source. Predicted potential probability hypothesis density potential distribution is the sum of the birth and survival target potential distributions, that is, the convolution of the birth and survival target potential distributions. Expressed as

[0085]

[0086] in, is the binomial coefficient, P S is the probability of target existence, is the birth potential distribution at time step t. The strength of the prediction It is expressed as follows:

[0087]

[0088] The strength of the survival goal is expressed as

[0089]

[0090] in, Birth intensity is expressed as

[0091]

[0092] in, The number of Gaussian mixture terms representing birth intensity, and are the corresponding prediction weights, means, and covariances of the Gaussian mixture terms, respectively.

[0093] S4: The potential probability hypothesis density update step is completed by the unscented Kalman filter. The specific method is: when the microphone array receives the sound source signal, the potential probability hypothesis density of each sound source is updated separately. The updated potential distribution and update intensity It is expressed as follows:

[0094]

[0095]

[0096] Among them, P D is the target detection probability, For z∈Ω t Detection item|Ω t | can be expressed as

[0097]

[0098] DoA observation data noise follows Student T distribution, can be rewritten as

[0099]

[0100] Among them, Gamma(·) represents the gamma distribution, α t ~Gamma(ν / 2,ν / 2), Given n targets, multi-target observations Z t The likelihood, p K,t (·) is the clutter potential distribution at time step t, is the permutation coefficient, e j (·) is an elementary symmetric function of order j.

[0101]

[0102] in, is the probability hypothesis density of the random finite set of clutter at time step t, V is the room volume, λ c is the clutter rate, and the clutter follows a uniform distribution. The update strength of the potential probability assumption density The remaining parameters required are as follows:

[0103]

[0104]

[0105]

[0106]

[0107] The unscented Kalman function is applied to the update process of the mean and variance of the probability hypothesis density. The mean of the jth Gaussian component and variance Weighted sigma points It is composed of the following formula

[0108]

[0109]

[0110] Where ε = α 2(n+κ)-n, α=0.01 is the scale parameter, κ=1, β=2. The predicted measurement is generated by

[0111]

[0112] The predicted state is given by the following formula Covariance matrix and the cross-covariance matrix renew.

[0113]

[0114]

[0115]

[0116] The updated probability hypothesis density mean and covariance are obtained using the following formula;

[0117]

[0118] in, is the Kalman gain.

[0119] S5: Use the variational Bayesian method to approximate the posterior likelihood of the potential probability hypothesis density. The specific method is: The key to the variational Bayesian method is to obtain an approximation of the target likelihood Since the target likelihood Difficult to calculate, an approximate expression can be obtained

[0120]

[0121] The variational Bayes method is calculated by minimizing the Kullback-Leibler (KL) divergence and

[0123]

[0124]

[0125] Variational Bayesian methods approximate nonlinear probabilities through an iterative process. represents the expectation of the auxiliary random variable, A measure of the difference between the observed and predicted values.

[0126]

[0127]

[0128] S6: The robot pose estimation is obtained by weighted averaging the robot pose particle weights and the Rao-Blackwellized filter particles. The specific method is: In order to obtain the robot pose estimation, the Rao-Blackwell particle filter is used to estimate The robot's posture is represented by particles with different weights in

[0129]

[0130] The single feature strategy is used to generate the robot pose particle weights. The single feature strategy formula is as follows:

[0131]

[0132]

[0133] in, and Resampling get

[0134] S7: The sound source position is estimated by trimming and merging the updated optimized potential probability hypothesis density weights, mean and covariance, and the mean of the potential probability hypothesis density corresponding to the potential probability hypothesis density weight greater than the set threshold is the sound source position, and the number of potential probability hypothesis density weights that meet the conditions is the number of sound sources.

[0135] In order to verify the effectiveness of the present invention, the Imgae model was used to simulate a 6×6×3m 3 In the room, two static sound sources are located at s1 = (1, 3, 1.7) m and s2 = (5, 4, 1.75) m. The mobile robot is equipped with a microphone array and IMU to obtain DoA observation data and robot speed y t,v and direction y t,γ To ensure that the mobile robot is within the room, it is necessary to keep the robot within 1m of the room boundary. to enforce the maximum turn. The sound source is a pure speech signal with a sampling frequency of 16kHz randomly selected from the TIMIT database. DoA is calculated using the classic phase-transformed generalized cross-correlation algorithm (GCC-PHAT). RMSE (root mean square error), MAE (mean absolute error) and OSPA (ptimal SubpatternAssignment distance) are used to evaluate the performance of the acoustic synchronous positioning and mapping method. Under different signal-to-noise ratios (SNR) and different reverberation conditions, the positioning results of 50 Monte Carlo experiments of the acoustic synchronous positioning and mapping method are shown in Tables 1 and 2. It can be seen from the test results that under low signal-to-noise ratio conditions, the impact on the positioning results is still small, which can meet the requirements of practical applications. The impact on the method in a reverberant environment is small, which can meet the requirements of practical applications. All of these show the good robustness and good working stability of the present invention. Figure 3 Table 3 and Table 3 show the positioning results of different robot trajectories and sound source positions when the reverberation time is 300ms and the signal-to-noise ratio is 25dB.

[0136] Table 1 Positioning error of acoustic synchronous localization and mapping method under different SNRs with 300ms reverberation time

[0137]

[0138]

[0139] Table 2 Positioning error of acoustic synchronous positioning and mapping method under different reverberation times at 25 dB SNR

[0140]

[0141] Table 3 Comparison of different indicators of four groups of trajectories

[0142]

[0143] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for simultaneous acoustic localization and mapping based on robust potential probability hypothesis density, characterized in that include: The sound source state and DoA observation data are modeled as random finite sets, where the noise of the DoA observation data obeys the Student T distribution, and the potential probability hypothesis density of the sound source state and DoA observation data is recursively propagated, where the potential probability hypothesis density includes potential distribution, weight, mean and covariance; The minimum and maximum distances from the sound source to the robot are evenly distributed, and the random numbers randomly generated by the even distribution are used as the missing distance information of the sound source state; The DoA observation data and the missing distance information of the sound source state are combined into the mean value of the potential probability hypothesis density of the sound source state, and the mean value of the potential probability hypothesis density of the sound source state is used as the input information of the potential probability hypothesis density prediction process; The unscented Kalman filter based on Student-T distribution is used to update the mean and covariance of the probability hypothesis density of the sound source state potential; The variational Bayes method is used to calculate the approximate potential probability hypothesis density likelihood probability, which is used to update the weight of the potential probability hypothesis density; A single feature strategy is used to generate the robot pose particle weights, and the robot pose particle weights are weighted averaged with the Rao-Blackwellized filter particles to obtain the robot pose estimation, and the robot pose particle weights are resampled. The mean and covariance of the updated potential probability hypothesis density of the sound source state are trimmed and merged. When the weight of the potential probability hypothesis density is greater than the set threshold, the corresponding mean of the potential probability hypothesis density is the sound source position, and the number of weights of the potential probability hypothesis density that meets the conditions is the number of sound sources.

2. The method for acoustic simultaneous localization and mapping based on robust potential probability hypothesis density according to claim 1, characterized in that: When the sound source state and DoA observation data are modeled as random finite sets: the number of constructions is N t The source state is a random finite set s t,n Indicates the status of each sound source Among them, n t,n is the process noise with covariance Q; The DoA estimation algorithm is used to infer the instantaneous direction of each sound source relative to the robot at time t. The multi-source measurement process is assumed to be the union of the measurement value generated by the target and the clutter caused by the reverberation. The measurement data Including quantity M t DoA observation data and robot measurement data The robot measurement data includes the robot running speed y t,v and direction, y t,γ The DoA observation data noise follows the Student T distribution. The DoA observation data is shown below: ω t,m (g(s). t,n )+e t,m ,e t,m ~St(0). 2×1 ,R t,m ,ν) where St(·) is the Student T distribution, g(·) is the Cartesian to spherical transformation, ν is the degree of freedom parameter, and e t,m is the covariance R t,m DoA observation data error; Among them, K t is a random set of false measurements generated by clutter, and each DoA observation data is Azimuth φ t,m ∈[0,2π) and the elevation angle θ t,m ∈[0,π).

3. The method for acoustic simultaneous localization and mapping based on robust potential probability hypothesis density according to claim 2, characterized in that: Use a uniformly distributed random number as the sound source state. When distance information is missing: Estimate the distance of each unmeasured sound source relative to the robot by using J b The sound source state is missing distance Initialize the mean value of the birth probability hypothesis density of the i-th particle in Represents the minimum distance r from the sound source to the robot min and the maximum value r max The mean of the birth probability density is assumed to be Built Among them, g -1 (·) is the transformation from spherical coordinates to Cartesian coordinates.

4. The method for acoustic simultaneous localization and mapping based on robust potential probability hypothesis density according to claim 3, characterized in that: The prediction process of potential probability hypothesis density is as follows: the Gaussian mixture potential probability hypothesis density is used to estimate the location of the sound source. The mean of the Gaussian term represents the possible location of the sound source. The predicted potential probability hypothesis density potential distribution is the sum of the birth and survival target potential distributions, that is, the convolution of the birth and survival target potential distributions, Expressed as in, is the binomial coefficient, P S is the probability of target existence, is the birth potential distribution at time step t, the strength of the prediction It is expressed as follows: in, Indicates birth intensity, Indicates the strength of the survival goal.

5. The method for acoustic simultaneous localization and mapping based on robust potential probability hypothesis density according to claim 4, characterized in that: The potential probability hypothesis density update process is as follows: When the mean and covariance of the sound source state potential probability hypothesis density are updated: When the microphone array receives the sound source signal, each sound source potential probability hypothesis density is updated separately, where the updated potential distribution and update intensity It is expressed as follows: Among them, P D is the target detection probability, Given n targets, multi-target observation Z t The likelihood of DoA observation data noise follows Student T distribution, for z∈Ω t Detection item|Ω t | can be expressed as Among them, Gamma(·) represents the gamma distribution, α t ~Gamma(ν / 2,ν / 2); Given the estimated robot pose The mean of the jth Gaussian component and covariance The unscented Kalman filter is used in the update process of formula (15), and the predicted state is given by the following formula Covariance matrix and the cross-covariance matrix Make an update The updated probability hypothesis density mean and covariance are obtained using the following formula; in, is the Kalman gain.

6. The method for acoustic simultaneous localization and mapping based on robust potential probability hypothesis density according to claim 5, characterized in that: When using variational Bayesian methods to calculate the approximate potential probability hypothesis density likelihood probability: obtain an approximation of the target likelihood The variational Bayes method is calculated by minimizing the Kullback-Leibler divergence and Variational Bayesian methods approximate nonlinear probabilities through an iterative process. represents the expectation of the auxiliary random variable, A measure of the difference between the observed and predicted values.

7. The method for acoustic simultaneous localization and mapping based on robust potential probability hypothesis density according to claim 6, characterized in that: The robot pose estimate is obtained by taking the weighted average of the robot pose particle weights and the Rao-Blackwellized filter particles: The robot pose estimate is obtained using the Rao-Blackwell particle filter The robot's posture is represented by particles with different weights in The single feature strategy is used to generate the robot pose particle weights. The single feature strategy formula is as follows: in, and Resampling get

Images