CN114972826A

CN114972826A - Method and device for determining target state, electronic equipment and computer storage medium

Info

Publication number: CN114972826A
Application number: CN202110214037.4A
Authority: CN
Inventors: 王博胜; 王晓亮; 王兵; 卿泉; 王刚
Original assignee: Taobao China Software Co Ltd
Current assignee: Taobao China Software Co Ltd
Priority date: 2021-02-25
Filing date: 2021-02-25
Publication date: 2022-08-30

Abstract

The embodiment of the application provides a method and a device for determining a target state, electronic equipment and a computer storage medium, wherein the method for determining the target state comprises the following steps: acquiring point cloud data of a target to be determined at the current moment; aiming at the point cloud data at the current moment, a preset point cloud registration algorithm is used for registration, so that the error in point cloud registration can be reduced, the accuracy of a registration processing result is improved, and the accurate speed distribution of the target to be determined at the current moment can be obtained based on the registration process; and predicting the point cloud position change of the target to be determined at the next moment according to the speed distribution of the target to be determined at the current moment. Therefore, the automatic driving vehicle can more accurately sense the surrounding objects, namely the state of the target to be determined, so that the automatic driving vehicle system can accurately sense the surrounding objects, more accurate and reasonable driving decisions are executed, and the safety driving of the automatic driving vehicle is guaranteed.

Description

Method and device for determining target state, electronic equipment and computer storage medium

Technical Field

The embodiment of the application relates to the technical field of automatic driving, in particular to a method and a device for determining a target state, electronic equipment and a computer storage medium.

Background

The autonomous driving is a technology for controlling a vehicle through a computer system to realize unmanned driving of the vehicle, has great advantages in aspects of improving driving safety, traffic efficiency and the like, and is a research hotspot in the industry.

For a high-level automatic driving vehicle system, the operation environment is complex and does not depend on the operation of a driver, so that accurate perception of the environment around the vehicle is particularly important. The system can effectively guarantee the safety of an automatic driving vehicle system by accurately sensing the motion of objects around the vehicle, and lays a foundation for subsequent decision and control of vehicle driving. Currently, three-dimensional laser radars are mostly adopted as important sensors in an automatic driving vehicle system to obtain reliable three-dimensional information of objects around the vehicle. However, three-dimensional lidar sampling data is sparse, making it inadequate in speed measurement.

Therefore, how to enable the automatic driving vehicle system to accurately sense the motion of the peripheral objects becomes a problem to be solved urgently.

Disclosure of Invention

In view of the above, embodiments of the present application provide a target state determination scheme to at least partially solve the above problems.

According to a first aspect of embodiments of the present application, there is provided a method for determining a target state, including: acquiring point cloud data of a target to be determined at the current moment; aiming at the point cloud data of the current moment, determining the speed distribution of the target to be determined by using a preset point cloud registration algorithm; and predicting the point cloud position change of the target to be determined at the next moment according to the speed distribution of the target to be determined.

According to a second aspect of the embodiments of the present application, there is provided an apparatus for determining a target state, including an obtaining module, a registering module, and a predicting module; the acquisition module is used for acquiring point cloud data of a target to be determined at the current moment; the registration module is used for determining the speed distribution of the target to be determined by using a preset point cloud registration algorithm according to the point cloud data at the current moment; the prediction module is used for predicting the point cloud position change of the target to be determined at the next moment according to the speed distribution of the target to be determined.

According to a third aspect of embodiments of the present application, there is provided an electronic apparatus, including: the system comprises a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface complete mutual communication through the communication bus; the memory is used for storing at least one executable instruction, and the executable instruction causes the processor to execute the operation corresponding to the determination method of the target state according to the first aspect.

According to a fourth aspect of embodiments of the present application, there is provided a computer storage medium having stored thereon a computer program which, when executed by a processor, implements the method of determining a target state as described in the first aspect.

According to the determination scheme of the target state provided by the embodiment of the application, point cloud data of the target to be determined at the current moment is obtained; aiming at the point cloud data at the current moment, a preset point cloud registration algorithm is used for registration, so that the error in point cloud registration can be reduced, the accuracy of a registration processing result is improved, and the speed distribution of a target to be determined can be determined through the registration process; and predicting the point cloud position change of the target to be determined at the next moment according to the speed distribution of the target to be determined. In the registration process, the registration of the point cloud position is an important aspect, and the time interval for collecting the point cloud data is short, so that the point cloud position change of the target to be determined at the next time can be predicted according to the speed distribution of the current time. Therefore, the automatic driving vehicle can more accurately sense the state of the object around the automatic driving vehicle, namely the target to be determined, so that more accurate and reasonable driving decision is executed, and the safety driving of the automatic driving vehicle is guaranteed.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments described in the embodiments of the present application, and other drawings can be obtained by those skilled in the art according to the drawings.

Fig. 1 is a flowchart illustrating a method for determining a target state according to an embodiment of the present disclosure;

fig. 2 is a schematic diagram illustrating an exemplary application scenario provided in an embodiment of the present application;

FIG. 3 is a flowchart illustrating steps of another method for determining a target state according to an embodiment of the present disclosure;

fig. 4 is a block diagram illustrating a target state determining apparatus according to an embodiment of the present disclosure;

fig. 5 is a schematic structural diagram of an electronic device according to an embodiment of the present application.

Detailed Description

In order to make those skilled in the art better understand the technical solutions in the embodiments of the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, but not all embodiments. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments in the present application shall fall within the scope of protection of the embodiments in the present application.

It should be noted that the first and second in the present application are only for distinguishing names and do not represent sequential relationships, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features, e.g. first state, second state, first category, second category.

The following further describes specific implementations of embodiments of the present application with reference to the drawings of the embodiments of the present application.

The first embodiment,

An embodiment of the present application provides a method for determining a target state, as shown in fig. 1, where fig. 1 is a flowchart of the method for determining a target state provided in the embodiment of the present application, and the method for determining a target state includes the following steps:

step S101, point cloud data of the target to be determined at the current moment is obtained.

The target to be determined in the embodiment of the present application may be any object that moves relatively, for example, in an autonomous driving scenario, the target to be determined may be any object around an autonomous vehicle, and any object moves relative to the driving autonomous vehicle, including but not limited to a person, another vehicle, a road facility, an obstacle, a rod-shaped facility (such as a telegraph pole or a traffic light pole, etc.). The point cloud data (point cloud data) records three-dimensional coordinates, color information (RGB), reflection Intensity information (Intensity), and the like of each point representing the object in the form of a point. In this example, the point cloud data of the target to be determined around the vehicle may be collected in real time, and the point cloud data at the current time may be point cloud data at any time in the point cloud data collected in real time. Taking an automatic driving scene as an example, at least one sensor is loaded on an automatic driving vehicle, the sensor can be a laser radar, a millimeter wave radar, a camera and the like, and the example of collecting point cloud data of obstacles around the vehicle in real time through a three-dimensional laser radar loaded on the vehicle is taken as an example for explanation, and the three-dimensional laser radar can be used for collecting the point cloud data of the obstacles around the vehicle at different moments in real time.

It should be noted that the method for determining the target state provided in the embodiment of the present application is applied to an automatic driving scenario, as shown in fig. 2, fig. 2 is a schematic diagram illustrating an application scenario provided in the embodiment of the present application, a driving system in fig. 2 includes an automatic driving vehicle 21 and a network 22, a driving control device 211 and a three-dimensional lidar 212 are disposed in the automatic driving vehicle 21, the driving control device 211 is configured to control information such as a traveling path, a speed, and a direction of the automatic driving vehicle 21, the three-dimensional lidar 212 is used as a sensor of the automatic driving system and can provide three-dimensional information of obstacles around the vehicle, the driving control device 211 performs interactive communication with the three-dimensional lidar 212 through the network 22, and the network 22 may be a network 22 of various connection types, such as a wired network, a wireless communication link, an optical fiber cable, or the like. The determination method of the target state provided by the embodiment of the present application is executed by the driving control apparatus 211, and accordingly, the determination means of the target state is provided in the driving control apparatus 211. It is understood that the numbers of the driving control device 211, the network 22, and the three-dimensional lidar 212 in fig. 2 are only schematic representations, and the number of the driving control device 211, the network 22, and the three-dimensional lidar is not limited in the embodiment of the present application; fig. 2 illustrates a three-dimensional lidar as an example, but of course, the lidar may also be other sensors for acquiring point cloud data of an obstacle around a vehicle, which is not limited in this embodiment of the present application.

Step S102, aiming at the point cloud data of the current moment, a preset point cloud registration algorithm is used for determining the speed distribution of the target to be determined.

The preset point cloud registration algorithm in the embodiment of the present application may be implemented by using a suitable algorithm, for example: the method includes a simulated annealing algorithm, a grid search algorithm, a genetic algorithm, a tabu search algorithm, a particle swarm algorithm, an ant colony algorithm and the like, and the method is not limited in the embodiment of the application. Because the scanning beam emitted by the three-dimensional laser radar is shielded by an obstacle or other objects, the acquisition of the three-dimensional point cloud data of the whole obstacle cannot be completed through one-time scanning, and therefore, the object needs to be scanned from different positions and angles. The point cloud registration refers to splicing point cloud data scanned at adjacent moments or preset interval moments together aiming at point cloud data acquired by the same barrier at different moments. In the embodiment of the present application, a speed distribution may be represented by a speed Probability Density distribution Function (PDF).

Aiming at the point cloud data at the current moment, the preset point cloud registration algorithm is used for registration, so that the error in point cloud registration can be reduced, the accuracy of a registration processing result is improved, and the speed distribution of the target to be determined at the current moment can be determined through the registration process.

And step S103, predicting the point cloud position change of the target to be determined at the next moment according to the speed distribution of the target to be determined.

Because the time interval of the point cloud data collected by the three-dimensional laser radar is shorter, the point cloud position change at the next time can be predicted according to the speed distribution at the current time. In an example, the point cloud data acquired at different moments also carry time sequence information, and by performing time sequence smoothing processing on the point cloud data at the current moment when the point cloud position change at the next moment is predicted, the point cloud position change at the next moment of the target to be determined can be obtained through smooth prediction.

Optionally, in the embodiment of the present application, the point cloud position change at the next time of the target to be determined is predicted according to the speed distribution at the current time and a preset rule, where the preset rule may be set appropriately by a person skilled in the art according to actual requirements, in an implementation manner, the preset rule may be set according to a rule condition when historical point cloud data is trained, and in another implementation manner, the preset rule may be determined by analyzing a large number of preset rules used when performing prediction processing on the point cloud position change at the next time on a large number of speed distributions at the current time of the target to be determined.

It should be noted that, in the embodiment of the present application, a time interval between a current time and a next time may be set by a person skilled in the art according to an actual application scenario or a sampling interval time of the three-dimensional laser radar, or may be set by a user according to a requirement of the user, which is not limited to this embodiment of the present application.

According to the method for determining the target state, point cloud data of a target to be determined at the current moment are obtained; aiming at the point cloud data at the current moment, the preset point cloud registration algorithm is used for registration, so that the error in point cloud registration can be reduced, the accuracy of the registration processing result is improved, and the speed distribution of the target to be determined at the current moment can be obtained through the registration process. And predicting the point cloud position change of the target to be determined at the next moment according to the speed distribution of the target to be determined. In the registration process, the registration of the point cloud position is an important aspect, and the time interval for collecting the point cloud data is short, so that the point cloud position change of the target to be determined at the next time can be predicted according to the speed distribution of the current time. Therefore, the automatic driving vehicle can more accurately sense the state of the object around the automatic driving vehicle, namely the target to be determined, so that more accurate and reasonable driving decision is executed, and the safety driving of the automatic driving vehicle is guaranteed.

Example II,

The second embodiment of the present application is based on the solution of the first embodiment, wherein the step S102 may be implemented by the steps S102 a-S102 c, and the step S102a determines a registration time before the current time according to a preset point cloud registration algorithm; step S102b, point cloud data of the registration moment are obtained; step S102c, using a preset point cloud registration algorithm to perform registration processing on the point cloud data at the current moment and the point cloud data at the registration moment, and obtaining the speed distribution of the target to be determined at the current moment according to the registration processing result.

It should be noted that, in the embodiment of the present application, the registration time is determined according to the registration accuracy of a preset point cloud registration algorithm, but in practical application, the registration time before the current time may also be a previous time (previous point cloud data acquisition time) of the current time for acquiring point cloud data, or may also be any time before the current time determined according to an actual situation.

If the registration time is the previous time of the current time, when the target to be determined moves slowly, the point cloud data of the current time and the point cloud data of the previous time are not distinguished obviously enough, the point cloud data of the current time and the point cloud data of the previous time are subjected to registration processing, and the error of the obtained registration processing result is larger (compared with the target to be determined with a higher moving speed), so that the point cloud data of the target to be determined at different times are registered by adopting a cross-time registration method, the registration time is determined by the registration precision of the point cloud registration algorithm, the point cloud data of the current time and the point cloud data of the registration time are registered by using a preset point cloud registration algorithm, the error during registration is reduced, the accuracy of the registration processing result is improved, and based on the more accurate registration processing result, more accurate speed distribution of the target to be determined at the current moment can be obtained.

Optionally, in an embodiment of the present application, the registration time in step S102a is determined according to the registration accuracy of the point cloud registration algorithm, the minimum instantaneous speed of the target to be determined, and the acquisition time interval for acquiring the point cloud data.

According to the method and the device, when the registration time is determined by adopting a cross-time registration method, the registration precision of a point cloud registration algorithm, the minimum instantaneous speed of a target to be determined and the acquisition time interval of the acquired point cloud data are comprehensively considered, the accuracy of the registration time is further improved, and therefore the error in registering the point cloud data at the current time and the point cloud data at the registration time is further reduced.

When determining the registration time to be registered with the point cloud data at the current time, optionally, in an embodiment of the present application, the registration time to be registered with the point cloud data at the current time is determined according to a result of (the registration accuracy of the point cloud registration algorithm, the acquisition time interval of the acquired point cloud data, and the minimum instantaneous speed of the target to be determined). Where "/" denotes division. Exemplarily, the precision of the point cloud registration algorithm is set to be s centimeters, the minimum instantaneous speed required for distinguishing the target to be determined is v centimeters per second, and the acquisition time interval of the three-dimensional laser radar sampling point cloud data is delta _ ts. Taking the current time as t and the previous time as t-1 as an example, the registration time is t-s/delta _ ts/v, that is, the registration time is no longer the previous time of the current time, but is determined by the registration accuracy of the point cloud registration algorithm, the acquisition time interval of the acquired point cloud data, and the minimum instantaneous speed of the target to be determined, where s cm, the minimum instantaneous speed vccm/s, and the acquisition time interval delta _ ts may be set by those skilled in the art according to specific situations, and the embodiment of the present application is not limited. In other words, in the embodiment of the application, before the point cloud data and the point cloud data at the registration time before the current time are subjected to registration processing, the registration time at which the point cloud data to be registered with the current time are to be determined according to the actual situation, and the registration time is no longer the previous time of the current time, so that the potential error of the point cloud position caused by slow movement of the target to be determined can be reduced, and the influence of the change of the view angle on the point cloud registration is reduced.

Optionally, in an embodiment of the present application, the step S102c may be implemented by: carrying out densification processing on the point cloud data at the current moment to obtain dense point cloud data at the current moment; carrying out densification processing on the point cloud data at the registration moment to obtain dense point cloud data at the registration moment; and performing registration processing on the dense point cloud data at the current moment and the dense point cloud data at the registration moment by using a preset point cloud registration algorithm.

In a practical application scenario, the reasons for causing the obstacle point cloud data to change may include the following two reasons: one is relative movement between obstacles, the other is change of visual angles, and the point cloud data collected by the three-dimensional laser radar is sparse, so that the point cloud data is changed violently due to the change of the visual angles. The method is characterized in that an automatic driving vehicle applied to an automatic driving scene is taken as an example, at least one three-dimensional laser radar can be carried on the automatic driving vehicle, point cloud data of obstacles around the vehicle can be collected through the three-dimensional laser radar mounted on the automatic driving vehicle, for a target to be determined, point cloud registration of the previous moment and the current moment can be achieved when geometric information of the target to be determined is known, but due to sparse sampling of the three-dimensional laser radar, complete geometric information of the target to be determined is unknown, and therefore dense processing is further carried out on the point cloud data collected at the current moment, and dense point cloud data corresponding to the current moment are obtained. Compared with point cloud data directly acquired by a three-dimensional laser radar, dense point cloud data carries richer three-dimensional information of a target to be determined, so that the registration result is more accurate when a point cloud registration algorithm is subsequently used for registration processing. The method and the device for processing the point cloud data perform densification processing on the point cloud data, namely densifying the geometric information of the target to be determined. Based on the more accurate registration result, more accurate velocity distribution at the current time can be obtained.

The point cloud data densification processing in the above steps can be realized by the following two exemplary ways. In a first example, binning is performed on point cloud data, and dense point cloud data is obtained according to binning results. In the second example, the point cloud data is subjected to meshing processing, and dense point cloud data is obtained according to the result of the meshing processing.

In the first example, a surface element form is adopted to perform densification processing on the point cloud data, illustratively, each point cloud is modeled into a local small plane, a surface element is constructed, namely a curved surface of the point cloud is reconstructed, and the point cloud data is subjected to densification processing by adopting the reconstructed point cloud surface elements to obtain dense point cloud data. The specific manner and process for constructing the surface element may be implemented by those skilled in the art by adopting any appropriate manner or algorithm according to actual needs, and the embodiment of the present application does not limit this.

In the second example, the point cloud data is subjected to a densification process in a gridding form, for example, a triangular patch, the point clouds are connected with each other according to a neighborhood relationship to form a grid expression, and dense point cloud data is output. The specific implementation of the gridding algorithm can also be implemented by those skilled in the art in any appropriate manner or algorithm according to actual needs, and this is not limited in this embodiment of the present application.

According to the method and the device, the point cloud data of the target to be determined at the current moment and the point cloud data of the registration moment are subjected to densification processing, so that the geometric information of the target to be determined is densified, and compared with the point cloud data directly acquired by a three-dimensional laser radar, the dense point cloud data carries more abundant three-dimensional information of the target to be determined, so that the registration result is more accurate when the point cloud registration algorithm is subsequently used for registration processing.

Optionally, in an embodiment of the present application, the preset point cloud registration algorithm is a general iterative closest point GICP algorithm.

The preset point cloud registration algorithm in the embodiment of the present application may be an iterative closest point algorithm ICP or a general iterative closest point GICP (Generalized-ICP) algorithm, etc.

According to the embodiment of the application, the energy function of the GICP is adopted for modeling, the point cloud registration energy function is determined, the point cloud data of the target to be determined at the current moment and the point cloud data of the target to be determined at the registration moment are subjected to registration processing according to the point cloud registration energy function, the accuracy of a registration processing result is improved, and therefore the speed distribution of the target to be determined at the current moment is obtained according to the registration processing result.

Based on the scheme of the first embodiment, after step S103, the second embodiment of the present application may further process point cloud data in which the point cloud position change exceeds a change threshold. Optionally, the point cloud data with the point cloud position change exceeding the change threshold is subjected to weight reduction or rejection processing.

Predicting the point cloud location change at the next moment of the target to be determined based on the point cloud registration according to the velocity distribution at the current moment. The point cloud registration refers to the point cloud at the current moment is registered as much as possible, but point clouds which are not well registered exist, for example, point clouds with more severe changes, the point clouds which are not well registered can be predicted through prediction, and are processed, so that adverse effects on acquisition and registration of cloud data of the next moment point are avoided. If the point cloud position changes greatly, if the point cloud position changes more than a change threshold, it indicates that the point cloud data acquired at the current time may not be accurate enough, and the point cloud data needs to be processed to obtain a more accurate result. Illustratively, the point cloud data with the point cloud position change exceeding the change threshold are screened out, and when the point cloud data at the next moment are registered, the screened out point clouds are omitted, so that the accuracy of the registration processing result is improved. Therefore, the automatic driving vehicle can more accurately sense the state of the object around the automatic driving vehicle, namely the target to be determined, so that more accurate and reasonable driving decision is executed, and the safety driving of the automatic driving vehicle is guaranteed.

When point cloud data with point cloud position changes exceeding a change threshold are processed, in one implementation mode, the point cloud data with the point cloud position changes exceeding the change threshold are removed, when the point cloud data at the next moment are matched, the removed point cloud data do not participate in point cloud registration processing at the next moment, and in the other implementation mode, the point cloud with the point cloud position changes exceeding the change threshold is matched with a lower weight value, so that the influence of the point cloud registration processing at the next moment is reduced, and the accuracy of a registration processing result is improved.

Example III,

Optionally, in an embodiment of the present application, after obtaining the speed distribution of the target to be determined at the current time according to any of the first to second embodiments, the method for determining the target state further includes the following steps: step S301 and step S302. As shown in fig. 3, fig. 3 is a flowchart illustrating steps of another method for determining a target state according to an embodiment of the present application. The method for determining the target state comprises the following steps:

step S301, respectively extracting dynamic and static characteristics of the speed distribution at the registration moment and the speed distribution at the current moment to obtain the dynamic and static characteristics of the target to be determined at the current moment.

Wherein, dynamic and static characteristics include: the dynamic probability and the static probability of the target to be determined.

It should be noted that, taking the automatic driving system applied in the automatic driving scene as an example for description, the normal operation of the automatic driving system requires the cooperative cooperation of a plurality of modules, wherein the sensing module is used as the eye of the automatic driving system and plays an important role in the safety of the automatic driving system. And the speed estimation and the dynamic and static judgment of obstacles around the vehicle are important links in the sensing module, so that the safety of the automatic driving system is effectively guaranteed, and a reliable basis is provided for subsequent decision and control. Therefore, after the speed distribution of the target to be determined at the current moment is obtained, the dynamic and static characteristics of the target to be determined at the current moment are judged according to the speed distribution of the target to be determined at the current moment.

In the embodiment of the present application, a velocity probability density distribution function may be used to represent velocity distribution, dynamic and static characteristics of the velocity distribution at the registration time and the velocity probability density distribution function at the current time are extracted, and dynamic and static characteristics of the target to be determined are calculated according to the extracted dynamic and static characteristics, where the dynamic and static characteristics in the embodiment of the present application include the probability that the target to be determined is in a dynamic state and the probability that the target to be determined is in a static state at the current time, and optionally, after the probabilities that the target to be determined is in the dynamic state and the probability that the target to be determined is in the static state are normalized, the sum obtained by adding the probability that the target to be determined is in the dynamic state and the probability that the target to be determined is in the static state is 1.

Optionally, in an embodiment of the present application, the dynamic and static features further include: for indicating the uncertainty probability of the target state uncertainty to be determined.

For convenience of understanding, in this example, an example is given in which an object to be determined is any obstacle around a vehicle, and the description is given with reference to fig. 2 in the first embodiment, in an actual application scenario, there may be an obstacle state that cannot be determined by a driving control device in an autonomous vehicle, for example, when point cloud data of a certain obstacle at a registration time and at a current time suddenly changes, so that when a preset point cloud registration algorithm is used to perform registration processing on point cloud data of the current time and point cloud data of the registration time, a large error is generated in an obtained registration processing result, accuracy of speed distribution of an obtained obstacle at the current time is low, and thus dynamic and static features of the obstacle at the current time include uncertainty. The dynamic and static characteristics of the obstacle at the current moment in the embodiment of the application further comprise uncertainty probability used for indicating the uncertainty of the state of the obstacle, and the uncertainty probability can be used for prompting or giving an alarm to a driver of the automatic driving vehicle, so that the driver of the automatic driving vehicle can manually judge the obstacle according to the actual condition of a road, judge whether the automatic driving vehicle needs to avoid or execute other more accurate and reasonable driving decisions, and provide guarantee for the safe driving of the automatic driving vehicle.

In this embodiment of the present application, the dynamic and static characteristics include a probability that the target to be determined is in a dynamic state at the current time, a probability that the target to be determined is in a static state, and a probability that the target to be determined is in an uncertain state, and optionally, after the probabilities that the target to be determined is in the dynamic state, the probability that the target to be determined is in the static state, and the probability that the target to be determined is in the uncertain state are normalized, a sum obtained by adding the probabilities that the target to be determined is in the dynamic state, the probability that the target to be determined is in the static state, and the probability that the target to be determined is in the uncertain state is 1.

Alternatively, in an embodiment of the present application, step S301 may be implemented as step S301a and step S301 b.

Step S301a, obtaining a speed direction consistency measurement, a state uncertainty measurement, and a stationary state confidence of the target to be determined according to the speed distribution at the registration time and the speed distribution at the current time.

The consistency of the speed direction of the target to be determined between the registration time and the current time, that is, the variance of the speed direction, may be expressed by using a circular statistics technique, for example, it may be understood that other techniques for expressing the consistency of the speed direction may also be used, and the embodiment of the present application is only an exemplary representation, and is not limited thereto. In determining the velocity direction consistency measure, in one implementation, a weighted average of the velocity probability density distribution function is employed, and in another implementation, a topk form is employed, and the velocity direction consistency measure is obtained from the variance of the velocity direction of the velocity probability density distribution function with the first K being large. The obtained speed direction consistency measurement is set to be h, the h of the dynamic characteristic and the h of the static characteristic are counted, the distribution of the speed direction consistency is fitted in at least one mode, the h is normalized to be between 0 and 1 through a normalization function according to the fitted distribution, and the normalized h is set to be ph.

The method includes the steps that state uncertainty measurement of a target at a registration time and a current time is to be determined, unimodal characteristics of a velocity probability density distribution function are modeled through uncertainty modeling, state uncertainty measurement u is determined according to a standard deviation of the velocity probability density distribution function, the u is normalized to be between 0 and 1 through a normalization function, and the normalized u is set to be pu.

The confidence coefficients of the registration time and the current time of the target to be determined are determined, that is, in the velocity probability density distribution function, the confidence coefficient s of the static state is determined according to the probability density with the velocity of 0, then s is normalized to be between 0 and 1 through the normalization function, and the normalized s is set to be ps.

The normalization function in the embodiment of the present application may be a softmax function, or may be another mapping function for normalization, which is not limited in the embodiment of the present application.

Step S301b, obtaining the dynamic and static characteristics of the target to be determined at the current moment according to the speed direction consistency measurement, the state uncertainty measurement and the static state confidence.

Optionally, in an embodiment of the present application, step S301b may be implemented by at least one of the following: determining a difference between a preset standard value and a product of the state uncertainty measure and the stationary state confidence; determining the dynamic probability of the target to be determined according to the product of the difference and the speed direction consistency measurement; and/or determining the difference between a preset standard value and the speed direction consistency measurement, and determining the static probability of the target to be determined according to the difference; and/or determining the uncertainty probability of the target to be determined according to the product of the speed direction consistency measurement, the state uncertainty measurement and the static state confidence coefficient.

It should be noted that, in the embodiment of the present application, if the speed direction consistency measurement is represented by ph, pu represents the state uncertainty measurement, and ps represents the static state confidence, because ph, pu, and ps are normalized values, a preset standard value is denoted by 1 in the embodiment of the present application, and it can be understood that, if the speed direction consistency measurement is represented by h, the state uncertainty measurement is represented by u, and the static state confidence is represented by s in the embodiment of the present application, the preset standard value may be set by a person skilled in the art after comprehensively analyzing h, u, and s.

In the embodiment of the application, ph represents the consistency measurement of the speed direction, pu represents the state uncertainty measurement, ps represents the confidence coefficient of the static state, and 1 represents a preset standard value, wherein if the dynamic and static characteristics include the dynamic probability, the static probability and the uncertainty probability of the target to be determined, the dynamic probability is represented as (1-pu × ps) × ph, the static probability is represented as 1-ph, and the uncertainty probability is represented as pu × ps × ph.

If the dynamic and static characteristics include the dynamic probability and the static probability of the target to be determined, the dynamic probability is represented as [ (1-pu × ps) × ph ]/(1-pu × ps × ph), the dynamic probability is represented as (1-ph)/(1-pu × ps × ph), and the symbol "/" in the formula is divided by.

Step S302, determining the state of the target to be determined at the current moment according to the dynamic and static characteristics of the target to be determined at the current moment.

Step S302 may be implemented by way of a first state determination operation and a second state determination operation. Illustratively, the first state determining operation includes: according to the dynamic and static characteristics of the target to be determined at the current moment, carrying out first classification processing on the dynamic and static characteristics through a hidden Markov model, and determining that the target to be determined is in a motion state, a static state or an uncertain state at the current moment according to a first classification result.

According to the method and the device, the state sequence is modeled according to a Hidden Markov Model (HMM for short), and the HMM is used for carrying out first classification processing on the dynamic and static features. And setting a transition probability matrix according to the dynamic and static characteristics, classifying and solving the dynamic and static characteristics by using an HMM (hidden Markov model), and determining that the target to be determined is in a motion state or a static state or an uncertain state at the current moment according to a first classification result.

Illustratively, the second state determining operation includes: if the first classification results of the continuous multiple moments indicate that the target to be determined is in a set state, taking the last moment of the continuous multiple moments as a reference moment, wherein the set state comprises a static state or an uncertain state; and after the reference time, updating the current time at preset time intervals, acquiring the real-time speed distribution of the current time, and determining the state of the target to be determined according to the real-time speed distribution and the speed distribution corresponding to the reference time.

The setting state of the embodiment of the application may include a static state or an uncertain state, and if the first classification results at a plurality of consecutive times indicate that the target to be determined is in the setting state, it is unknown whether the target to be determined is moving, the state of the target to be determined does not need to be determined according to the first state determination operation, and the state of the target to be determined needs to be determined according to the second state determination operation. For example, originally, the point cloud data of the current time and the registration time of the target to be determined is subjected to point cloud registration at a time interval of 0.1s, the first classification results of a plurality of continuous 0.1s indicate that the target to be determined is in a set state, the last 0.1s time of the plurality of continuous 0.1s is taken as a reference time, the current time is updated at a time interval of 1s after the reference time, the point cloud data of the current time is registered with the point cloud data of the reference time to obtain the real-time speed distribution of the current time, and the state of the target to be determined is determined according to the real-time speed distribution and the speed distribution corresponding to the reference time. The time span between the real-time speed distribution of the current time and the reference time is longer, so that not only time and computing resources are saved, but also the state of the target to be determined, which is determined according to the real-time speed distribution of the current time and the speed distribution in the set state, is more accurate. It is understood that the preset time interval can be set by those skilled in the art according to practical situations, and the embodiments of the present application are illustrated by taking 0.1s and 1s as examples, and do not represent that the embodiments of the present application are limited thereto.

When the second state determining operation is started, optionally, in an embodiment of the present application, the method for determining the target state further includes: and updating the state of the target to be determined by the first state determination operation by using the determined state of the target to be determined.

In the embodiment of the application, when the second state determining operation starts to work, the state of the target to be determined by the first state determining operation is replaced by the state determined by the second state operation, that is, whether the target to be determined is moving or not can not be judged according to the first state determining operation, and the state of the target to be determined corresponding to the target to be determined needs to be replaced by the state of the target to be determined by the second state operation, so that the determined state of the target to be determined is more accurate, the driving control device of the automatic driving vehicle is facilitated to execute accurate and reasonable driving decision according to the accurate state of the target to be determined, and safety driving of the automatic driving vehicle is guaranteed.

Optionally, in an embodiment of the present application, the method for determining the target state further includes: and monitoring the state of the object to be determined by the first state determining operation, and if the object to be determined is monitored to be converted from a static state or an uncertain state into a motion state, finishing the second state determining operation.

The second state determining operation is an operation used only when the state of the target to be determined by the first state determining operation is a static state or an uncertain state, the state of the target to be determined by the first state determining operation is monitored, if the state of the target to be determined is monitored to be changed from the static state or the uncertain state to a motion state, and the state is no longer suitable for the second state determining operation when the state of the target to be determined is determined, the second state determining operation is finished, so that not only are computing resources saved, but also the state of the target to be determined according to the first state determining operation in subsequent moments is more accurate, a driving control device of the automatic driving vehicle is facilitated to execute an accurate and reasonable driving decision according to the accurate state of the target to be determined, and safety driving of the automatic driving vehicle is guaranteed.

Further, the determination of the target state in the embodiment of the present application is described by taking a specific example, and the processing of the data includes speed estimation and dynamic and static classification.

Velocity estimation

When the speed of the obstacle at the current moment is estimated, a speed probability density distribution function of the obstacle at the current moment is obtained through a point cloud registration algorithm and a point cloud data registration error, the speed probability density distribution function represents speed distribution, and point cloud position change of the obstacle at the next moment is predicted according to the speed distribution at the current moment, which is concretely shown as follows.

(1) Global point cloud registration

The method and the device for estimating the movement speed of the obstacle are used for estimating the movement speed of the obstacle from the change, the position and the rotation of the point cloud data of the obstacle at the current moment. Illustratively, point cloud data at the current moment and point cloud data at the registration moment before the current moment are registered through a point cloud registration algorithm, the speed distribution of the obstacle at the current moment is obtained according to the registration processing result, and then the point cloud position change of the obstacle at the next moment is predicted according to the speed distribution of the current moment.

The point cloud registration algorithm may adopt various global optimization algorithms, for example: a simulated annealing algorithm, a grid searching algorithm and the like.

According to the method and the device, the point cloud data of the obstacles around the vehicle can be collected through the three-dimensional laser radar installed on the automatic driving vehicle, for the obstacles, when the geometrical information of the obstacles is known, the point cloud registration of the registration time and the current time can be achieved, but due to the sparse sampling performance of the three-dimensional laser radar, the complete geometrical information of the obstacles is unknown, and therefore the point cloud data collected at the current time are subjected to dense processing, and the dense point cloud data corresponding to the current time are obtained. Compared with point cloud data directly acquired by a three-dimensional laser radar, dense point cloud data carries richer three-dimensional information of obstacles, so that the registration result is more accurate when a point cloud registration algorithm is subsequently used for registration processing.

The embodiment of the application carries out the densification processing on the point cloud data in the following way, thereby obtaining dense point cloud data: (1) in a surface element form, modeling each point cloud as a local small plane, constructing a surface element, namely point cloud curved surface reconstruction, and performing densification processing on the point cloud data by adopting the reconstructed point cloud surface element to obtain dense point cloud data; (2) gridding, such as a triangular patch, connects the point clouds to each other according to the neighborhood relationship to form a grid expression, thereby outputting dense point cloud data.

The method and the device for acquiring the speed distribution of the obstacle at the current moment are used for registering the dense point cloud data at the current moment and the dense point cloud data at the registration moment before the current moment through a point cloud registration algorithm, and acquiring the speed distribution of the obstacle at the current moment according to a registration processing result. Compared with point cloud data directly acquired by a three-dimensional laser radar, the dense point cloud data carries richer three-dimensional information of obstacles, so that the registration result is more accurate when a point cloud registration algorithm is subsequently used for registration processing.

(2) Point cloud data registration error

If the registration time is the last time of the current time, when the obstacle moves slowly, the point cloud data of the current time and the point cloud data of the last time are not distinguished sufficiently, the point cloud data of the current time and the point cloud data of the last time are subjected to registration processing, and the error of the obtained registration processing result is larger (compared with the case that the movement speed of the obstacle is higher), so that the point cloud data of the obstacle at different times are registered by adopting a cross-time registration method.

In a practical application scenario, the reasons for causing the obstacle point cloud data to change may include the following two reasons: one is relative movement between obstacles, the other is change of visual angles, and the three-dimensional laser radar is sparse, so that the change of the visual angles causes drastic change of point cloud data, and the speed probability density distribution function of the obstacles is inaccurate. Therefore, the method and the device model the potential errors of the point cloud positions, and reduce the influence of the view angle change on the registration of the point cloud data at the current moment and the point cloud data at the previous moment. The registration time is not only the last time of the current time, that is, the time between the registration time and the current time is over-time, the precision of the point cloud registration algorithm is set to be s cm, the minimum instantaneous speed required for distinguishing the obstacles is v cm/sec, the acquisition time interval of the three-dimensional laser radar sampling point cloud data is delta _ ts, taking the current time as t and the last time as t-1 as an example, the registration time is (t-s/delta _ ts/v), and the symbol "/" in the formula represents division, that is, the registration time is not the last time of the current time any more, but is the time determined by the registration precision of the point cloud registration algorithm, the acquisition time interval of the acquired point cloud data and the minimum instantaneous speed of the obstacles.

According to the embodiment of the application, a GICP (Generalized-ICP) energy function is adopted for modeling to obtain a point cloud registration energy function, and point cloud data of the obstacle at the current moment and point cloud data of the registration moment are subjected to registration processing according to the point cloud registration energy function, so that the speed distribution of the obstacle at the current moment is obtained according to the registration processing result.

Furthermore, because the time interval of the three-dimensional laser radar for collecting the point cloud data is short, the change of the point cloud position at the next moment can be predicted according to the speed distribution of the obstacle at the current moment.

(II) dynamic and static classification

According to the method and the device, the dynamic and static characteristics are extracted through the speed probability density distribution function at the current moment and the registration moment, and are classified, so that the state of the obstacle at the current moment is obtained, and the method and the device are as follows.

(1) Dynamic and static feature extraction

The embodiment of the application also extracts dynamic and static characteristics of the speed probability density distribution function, if the dynamic and static characteristics comprise the dynamic probability, the static probability and the uncertainty probability of the barrier, the dynamic and static characteristics are expressed as three-dimensional vectors [ the dynamic probability, the static probability and the uncertainty probability ], and the probability accumulation of the three is 1.0; if the dynamic and static characteristics comprise the dynamic probability and the static probability of the obstacle, expressing the dynamic and static characteristics as [ the dynamic probability and the static probability ] and accumulating the probabilities to be 1.0. The dynamic and static characteristics extracted by the application are as follows:

1.1, consistency of speed and direction of two frames (point cloud data collected twice) before and after, namely variance of the speed and direction, and the consistency of speed and direction orientation is expressed by using a circular classification technology in the embodiment of the application, and a similar technology can also be adopted. In determining the velocity direction consistency measure, in one implementation, a weighted average of the velocity probability density distribution is used, and in another implementation, a topk form is used, and the velocity direction consistency measure is obtained from the variance of the velocity direction of the velocity probability density distribution with the top K being large. Firstly, the speed direction consistency measurement is set as h, h of the dynamic characteristic and the static characteristic is counted, the distribution of the speed direction consistency is fitted in various modes, h is normalized to be between 0 and 1 by the fitted distribution, the normalized h is set as ph, and the final speed direction consistency measurement is represented by the ph.

1.2, modeling uncertainty, mainly for modeling unimodal characteristics of velocity probability density distribution function, expressing uncertainty measurement u by using standard deviation of velocity distribution, and then mapping u to 0-1 by mapping function, which may be any function for normalization, such as softmax function, and setting normalized u to pu, and characterizing final state uncertainty measurement by pu.

And 1.3, determining the confidence coefficient s of the static state according to the probability density with the speed of 0 in a speed probability density distribution function, then mapping s to a range from 0 to 1 through a mapping function, wherein the mapping function can be any function for normalization, such as a softmax function, and setting the normalized s as ps, and characterizing the final confidence coefficient of the static state by using ps.

If the dynamic and static characteristics include the dynamic probability, the static probability and the uncertainty probability of the obstacle, the dynamic probability is expressed as (1-pu × ps) × ph, the static probability is expressed as 1-ph, and the uncertainty probability is expressed as pu × ps × ph. If the dynamic and static characteristics include the dynamic probability and the static probability of the obstacle, the dynamic probability is represented by [ (1-pu × ps) × ph ]/(1-pu × ps × ph), the dynamic probability is represented by (1-ph)/(1-pu × ps × ph), and the symbol "/" in the equation indicates division.

(2) Dynamic and static classification

The method comprises the following steps of modeling a state sequence according to a Hidden Markov Model (HMM), setting a proper transition probability matrix by using the extracted dynamic and static characteristics, and solving the dynamic and static characteristics by using an existing algorithm, wherein the process can be called short-term dynamic and static classification, the embodiment of the application also provides long-term dynamic and static classification short-term dynamic and static classification, and the detection of the ultra-low-speed obstacle is carried out, and the specific flow is as follows:

when the obstacles in the short-term dynamic and static classification result are in a static state or uncertain, the long-term dynamic and static classification starts to work, and the long-term dynamic and static classification work is as follows:

and 2.1, counting the speed direction consistency characteristics of all static moments from the static state, and counting to obtain a speed direction consistency measure ph.

And 2.2, recording the static moment of the first frame, and registering the point clouds of the first frame at each subsequent moment or at each certain interval moment. Point cloud registration is carried out at certain intervals, namely point clouds with longer interval time are selected to register the point clouds of the first frame, so that time can be saved, efficiency is improved, and uncertainty measurement pu and static state confidence coefficient ps are obtained through statistics.

And 2.3, replacing the final result of the short-term dynamic and static classification with the result of the long-term dynamic and static classification when the long-term dynamic and static classification starts to work.

And 2.4, if the classification result of the short-term dynamic and static classification is a motion state, finishing the long-term dynamic and static classification.

Example four,

An apparatus for determining a target state is provided in an embodiment of the present application, as shown in fig. 4, fig. 4 is the apparatus for determining a target state provided in the embodiment of the present application, and the apparatus 40 for determining a target state includes an obtaining module 401, a registering module 402, and a predicting module 403;

the obtaining module 401 is configured to obtain point cloud data of a target to be determined at a current moment;

the registration module 402 is configured to determine, for the point cloud data at the current time, a speed distribution of the target to be determined by using a preset point cloud registration algorithm;

the predicting module 403 is configured to predict a point cloud position change of the target to be determined at the next time according to the velocity distribution of the target to be determined.

Optionally, in an embodiment of the present application, the registration module 402 is further configured to determine a registration time before the current time according to a preset point cloud registration algorithm; acquiring point cloud data at the registration moment; and performing registration processing on the point cloud data at the current moment and the point cloud data at the registration moment by using a preset point cloud registration algorithm, and obtaining the speed distribution of the target to be determined at the current moment according to the registration processing result.

Optionally, in an embodiment of the present application, the registration time is determined according to the registration accuracy of the point cloud registration algorithm, the minimum instantaneous speed of the target to be determined, and the acquisition time interval of the point cloud data.

Optionally, in an embodiment of the present application, the apparatus for determining a target state 40 further includes a densification processing module, where the densification processing module is configured to perform densification processing on the point cloud data at the current time to obtain dense point cloud data at the current time; carrying out densification processing on the point cloud data at the registration moment to obtain dense point cloud data at the registration moment; the registration module 402 is configured to perform registration processing on the dense point cloud data at the current time and the dense point cloud data at the registration time by using a preset point cloud registration algorithm.

Optionally, in an embodiment of the application, the denseness processing module is further configured to perform binning on the point cloud data, and obtain the denseness point cloud data according to a binning result; or carrying out meshing processing on the point cloud data, and obtaining dense point cloud data according to a result of the meshing processing.

Optionally, in an embodiment of the present application, the target state determining apparatus 40 further includes a feature module and a state determining module, where the feature module is configured to perform dynamic and static feature extraction on the velocity distribution at the registration time and the velocity distribution at the current time, respectively, to obtain dynamic and static features of the target to be determined at the current time, where the dynamic and static features include: determining the dynamic probability and the static probability of a target to be determined; the state determining module is used for determining the state of the target to be determined at the current moment according to the dynamic and static characteristics of the target to be determined at the current moment.

Optionally, in an embodiment of the present application, the feature module is further configured to obtain a speed direction consistency metric, a state uncertainty metric, and a stationary state confidence of the target to be determined according to the speed distribution at the registration time and the speed distribution at the current time; and obtaining the dynamic and static characteristics of the target to be determined at the current moment according to the speed direction consistency measurement, the state uncertainty measurement and the static state confidence coefficient.

Optionally, in an embodiment of the present application, the feature module is further configured to determine a difference between a preset standard value and a product of the state uncertainty measure and the stationary state confidence; determining the dynamic probability of the target to be determined according to the product of the difference and the speed direction consistency measurement; and/or determining the difference between a preset standard value and the speed direction consistency measurement, and determining the static probability of the target to be determined according to the difference; and/or determining the uncertainty probability of the target to be determined according to the product of the speed direction consistency measurement, the state uncertainty measurement and the static state confidence coefficient.

Optionally, in an embodiment of the present application, the state determining module further includes a first state determining unit, where the first state determining unit is configured to perform a first state determining operation, and includes: according to the dynamic and static characteristics of the target to be determined at the current moment, carrying out first classification processing on the dynamic and static characteristics through a hidden Markov model, and determining that the target to be determined is in a motion state, a static state or an uncertain state at the current moment according to a first classification result.

Optionally, in an embodiment of the application, the state determining module further includes a second determining unit, where the second determining unit is configured to, if the first classification results at multiple consecutive times all indicate that the target to be determined is in the set state, take a last time of the multiple consecutive times as a reference time, where the set state includes a static state or an uncertain state; and after the reference time, updating the current time at preset time intervals, acquiring the real-time speed distribution of the current time, and determining the state of the target to be determined according to the real-time speed distribution and the speed distribution corresponding to the reference time.

Optionally, in an embodiment of the present application, the apparatus for determining a state of a target 40 further includes an updating module, configured to update the state of the target to be determined by the first state determining operation, using the determined state of the target to be determined.

Optionally, in an embodiment of the present application, the apparatus for determining a state of an object 40 further includes an ending module, where the ending module is configured to monitor a state of the object to be determined, which is determined by the first state determining operation, and end the second state determining operation if it is monitored that the object to be determined is changed from a stationary state or an uncertain state to a moving state.

Optionally, in an embodiment of the present application, the apparatus for determining a target state 40 further includes a processing module, and the processing module is configured to perform weight reduction or rejection processing on the point cloud data whose point cloud position changes exceed a change threshold.

The target state determining device 40 in the embodiment of the present application is used to implement the corresponding target state determining method in the foregoing multiple method embodiments, and has the beneficial effects of the corresponding method embodiments, and details are not repeated here. In addition, for the function implementation of each module in the apparatus for determining the target state in the embodiment of the present application, reference may be made to the description of the corresponding part in the foregoing method embodiment, and details are not repeated here.

Of course, these algorithm modules may vary depending on the type of autonomous vehicle. For example, different algorithm modules may be involved for logistics vehicles, public service vehicles, medical service vehicles, terminal service vehicles. The algorithm modules are illustrated below for these four autonomous vehicles, respectively:

the logistics vehicle refers to a vehicle used in a logistics scene, and may be, for example, a logistics vehicle with an automatic sorting function, a logistics vehicle with a refrigeration and heat preservation function, and a logistics vehicle with a measurement function. These logistics vehicles may involve different algorithm modules.

For example, the logistics vehicles can be provided with an automatic sorting device, and the automatic sorting device can automatically take out, convey, sort and store the goods after the logistics vehicles reach the destination. This relates to an algorithm module for sorting goods, which mainly implements logic control of goods taking out, carrying, sorting and storing.

For another example, in a cold-chain logistics scenario, the logistics vehicle may further include a refrigeration and insulation device, and the refrigeration and insulation device may implement refrigeration or insulation of transported fruits, vegetables, aquatic products, frozen foods, and other perishable foods, so that the transported fruits, vegetables, aquatic products, frozen foods, and other perishable foods are in a suitable temperature environment, thereby solving the problem of long-distance transportation of perishable foods. The algorithm module is mainly used for dynamically and adaptively calculating the proper temperature of cold meal or heat preservation according to the information such as the property, the perishability, the transportation time, the current season, the climate and the like of food (or articles), and automatically adjusting the cold-storage heat preservation device according to the proper temperature, so that a transport worker does not need to manually adjust the temperature when the vehicle transports different foods or articles, the transport worker is liberated from the complicated temperature regulation and control, and the efficiency of cold-storage heat preservation transportation is improved.

For another example, in most logistics scenarios, the fee is charged according to the volume and/or weight of the parcel, but the number of logistics parcels is very large, and the measurement of the volume and/or weight of the parcel by a courier is only dependent, which is very inefficient and has high labor cost. Therefore, in some logistics vehicles, a measuring device is added, so that the volume and/or the weight of the logistics packages can be automatically measured, and the cost of the logistics packages can be calculated. This relates to an algorithm module for logistics package measurement, which is mainly used to identify the type of logistics package, determine the measurement mode of logistics package, such as volume measurement or weight measurement or combined measurement of volume and weight, and can complete the measurement of volume and/or weight according to the determined measurement mode and complete the cost calculation according to the measurement result.

The public service vehicle is a vehicle providing some public service, and may be, for example, a fire truck, an ice removal truck, a watering cart, a snow scraper, a garbage disposal vehicle, a traffic guidance vehicle, and the like. These public service vehicles may involve different algorithm modules.

For example, in the case of an automatically driven fire fighting vehicle, the main task is to perform a reasonable fire fighting task on the fire scene, which involves an algorithm module for the fire fighting task, which at least needs to implement logic such as identification of the fire situation, planning of the fire fighting scheme, and automatic control of the fire fighting device.

For another example, in the case of an ice removing vehicle, the main task is to remove ice and snow formed on the road surface, and this involves an algorithm module for ice removal, which at least needs to recognize the ice and snow condition on the road surface, formulate an ice removal scheme according to the ice and snow condition, such as which road sections need to be removed, which road sections need not to be removed, whether a salt spreading manner, the number of salt spreading grams, etc. are adopted, and logic such as automatic control of an ice removing device in the case of determining the ice removal scheme.

The medical service vehicle is an automatic driving vehicle capable of providing one or more medical services, the vehicle can provide medical services such as disinfection, temperature measurement, dispensing and isolation, and the algorithm modules relate to algorithm modules for providing various self-service medical services.

The terminal service vehicle is a self-service automatic driving vehicle which can replace some terminal devices and provide certain convenient service for users, and for example, the vehicles can provide services such as printing, attendance checking, scanning, unlocking, payment and retail for the users.

For example, in some application scenarios, a user often needs to go to a specific location to print or scan a document, which is time consuming and labor intensive. Therefore, a terminal service vehicle capable of providing printing/scanning service for a user appears, the service vehicles can be interconnected with user terminal equipment, the user sends a printing instruction through the terminal equipment, the service vehicle responds to the printing instruction, documents required by the user are automatically printed, the printed documents can be automatically sent to the position of the user, the user does not need to queue at a printer, and the printing efficiency can be greatly improved. Or, the scanning instruction sent by the user through the terminal equipment can be responded, the scanning vehicle is moved to the position of the user, the user finishes scanning on the scanning tool of the service vehicle on which the document to be scanned is placed, queuing at the printing/scanning machine is not needed, and time and labor are saved. This involves an algorithm module providing print/scan services that needs to identify at least the interconnection with the user terminal equipment, the response to print/scan instructions, the positioning of the user's location, and travel control.

For another example, as new retail business is developed, more and more electronic stores sell goods to large office buildings and public areas by using vending machines, but the vending machines are placed at fixed positions and are not movable, and users need to go by the vending machines to purchase the needed goods, which is still poor in convenience. Therefore, self-service driving vehicles capable of providing retail services appear, the service vehicles can carry commodities to move automatically and can provide corresponding self-service shopping APP or shopping entrances, a user can place an order for the self-service driving vehicles providing retail services through the APP or shopping entrances by means of a terminal such as a mobile phone, the order comprises names and numbers of commodities to be purchased, and after the vehicle receives an order placement request, whether the current remaining commodities have the commodities purchased by the user and whether the quantity is sufficient can be determined. This involves algorithm modules that provide retail services that implement logic primarily to respond to customer order requests, order processing, merchandise information maintenance, customer location, payment management, etc.

Example V,

Based on the determination method of any target state described in the first to third embodiments, the present application provides an electronic device, and it should be noted that, the determination method of the target state in the present application may be executed by any suitable electronic device with determination capability of the target state, including but not limited to: server, mobile terminal (such as mobile phone, PAD, etc.), PC, etc. As shown in fig. 5, fig. 5 is a structural diagram of an electronic device according to an embodiment of the present application. The specific embodiments of the present application do not limit the specific implementation of the electronic device. The electronic device 50 may include: a processor (processor)502, a Communications Interface (Communications Interface)504, a memory 506, and a communication bus 508.

Wherein: the processor 502, communication interface 504, and memory 506 communicate with one another via a communication bus 508.

A communication interface 504 for communicating with other electronic devices or servers.

The processor 502 is configured to execute the computer program 510, and may specifically execute relevant steps in the above-described method for determining the target state.

In particular, the computer program 510 may comprise computer program code comprising computer operating instructions.

The processor 502 may be a central processing unit CPU, or an application Specific Integrated circuit asic, or one or more Integrated circuits configured to implement embodiments of the present application. The intelligent device comprises one or more processors which can be the same type of processor, such as one or more CPUs; or may be different types of processors such as one or more CPUs and one or more ASICs.

A memory 506 for storing a computer program 510. The memory 506 may comprise high-speed RAM memory, and may also include non-volatile memory (non-volatile memory), such as at least one disk memory.

The determination methods of the target states in the foregoing embodiments provided by the embodiments of the present application can all be executed by the electronic apparatus 50, and accordingly, the determination devices of the target states in the foregoing embodiments are provided in the electronic apparatus 50.

For specific implementation of each step in the program 510, reference may be made to corresponding steps and corresponding descriptions in units in the foregoing method for determining a target state, which are not described herein again. It can be clearly understood by those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described devices and modules may refer to the corresponding process descriptions in the foregoing method embodiments, and are not described herein again.

It should be noted that, according to the implementation requirement, each component/step described in the embodiment of the present application may be divided into more components/steps, and two or more components/steps or partial operations of the components/steps may also be combined into a new component/step to achieve the purpose of the embodiment of the present application.

The above-described methods according to the embodiments of the present application may be implemented in hardware, firmware, or as software or computer code that may be stored in a recording medium such as a CD ROM, RAM, floppy disk, hard disk, or magneto-optical disk, or as computer code downloaded through a network, originally stored in a remote recording medium or a non-transitory machine-readable medium, and to be stored in a local recording medium, so that the methods described herein may be stored in such software processes on a recording medium using a general purpose computer, a dedicated processor, or programmable or dedicated hardware such as an ASIC or FPGA. It will be appreciated that the computer, processor, microprocessor controller or programmable hardware includes memory components (e.g., RAM, ROM, flash memory, etc.) that can store or receive software or computer code that, when accessed and executed by the computer, processor or hardware, implements the method of determining a target state described herein. Further, when a general-purpose computer accesses code for implementing the determination method of the target state shown herein, execution of the code converts the general-purpose computer into a special-purpose computer for executing the determination method of the target state shown herein.

Those of ordinary skill in the art will appreciate that the various illustrative elements and method steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the embodiments of the present application.

The above embodiments are only used for illustrating the embodiments of the present application, and not for limiting the embodiments of the present application, and those skilled in the relevant art can make various changes and modifications without departing from the spirit and scope of the embodiments of the present application, so that all equivalent technical solutions also belong to the scope of the embodiments of the present application, and the scope of patent protection of the embodiments of the present application should be defined by the claims.

Claims

1. A method of determining a target state, comprising:

acquiring point cloud data of a target to be determined at the current moment;

aiming at the point cloud data of the current moment, determining the speed distribution of the target to be determined by using a preset point cloud registration algorithm;

and predicting the point cloud position change of the target to be determined at the next moment according to the speed distribution of the target to be determined.

2. The method according to claim 1, wherein the determining the velocity distribution of the target to be determined by using a preset point cloud registration algorithm for the point cloud data of the current time comprises:

determining a registration moment before the current moment according to the preset point cloud registration algorithm;

acquiring point cloud data of the registration moment;

and performing registration processing on the point cloud data at the current moment and the point cloud data at the registration moment by using the preset point cloud registration algorithm, and obtaining the speed distribution of the target to be determined at the current moment according to the registration processing result.

3. The method of claim 2, wherein the registration time is determined according to a registration accuracy of the point cloud registration algorithm, a minimum instantaneous velocity of the target to be determined, and an acquisition time interval of acquiring point cloud data.

4. The method according to claim 2, wherein the registration processing of the point cloud data at the current time and the point cloud data at the registration time by using the preset point cloud registration algorithm comprises:

carrying out densification processing on the point cloud data at the current moment to obtain dense point cloud data at the current moment;

carrying out densification processing on the point cloud data at the registration moment to obtain dense point cloud data at the registration moment;

and performing registration processing on the dense point cloud data at the current moment and the dense point cloud data at the registration moment by using the preset point cloud registration algorithm.

5. The method of claim 4, wherein the densification process comprises:

performing binning processing on the point cloud data, and obtaining dense point cloud data according to binning processing results;

alternatively, the first and second electrodes may be,

and carrying out meshing processing on the point cloud data, and obtaining dense point cloud data according to a meshing processing result.

6. The method according to any one of claims 1-5, wherein the preset point cloud registration algorithm is a General Iterative Closest Point (GICP) algorithm.

7. The method of claim 2, wherein the method further comprises:

respectively extracting dynamic and static characteristics of the speed distribution at the registration moment and the speed distribution at the current moment to obtain the dynamic and static characteristics of the target to be determined at the current moment, wherein the dynamic and static characteristics comprise: the dynamic probability and the static probability of the target to be determined;

and determining the state of the target to be determined at the current moment according to the dynamic and static characteristics of the target to be determined at the current moment.

8. The method of claim 7, wherein the dynamic and static features further comprise: and the uncertainty probability is used for indicating the uncertainty of the target state to be determined.

9. The method according to claim 7 or 8, wherein the performing dynamic and static feature extraction on the velocity distribution at the registration time and the velocity distribution at the current time respectively to obtain the dynamic and static features at the current time of the target to be determined comprises:

obtaining a speed direction consistency measurement, a state uncertainty measurement and a static state confidence of the target to be determined according to the speed distribution of the registration moment and the speed distribution of the current moment;

and obtaining the dynamic and static characteristics of the target to be determined at the current moment according to the speed and direction consistency measurement, the state uncertainty measurement and the static state confidence.

10. The method of claim 9, wherein the obtaining the dynamic and static characteristics of the target to be determined at the current time according to the speed direction consistency measure, the state uncertainty measure and the static state confidence comprises:

determining a difference between a preset standard value and a product of the state uncertainty measure and the stationary state confidence; determining a dynamic probability of the target to be determined according to a product of the difference and the speed direction consistency measure;

and/or the presence of a gas in the gas,

determining the difference between the preset standard value and the speed direction consistency measurement, and determining the static probability of the target to be determined according to the difference;

and/or the presence of a gas in the gas,

and determining the uncertainty probability of the target to be determined according to the product of the speed direction consistency measurement, the state uncertainty measurement and the static state confidence coefficient.

11. The method of claim 8, wherein the determining the state of the target to be determined at the current time according to the dynamic and static characteristics of the target to be determined at the current time comprises:

a first state determination operation comprising: and according to the dynamic and static characteristics of the current moment of the target to be determined, carrying out first classification processing on the dynamic and static characteristics through a hidden Markov model, and according to a first classification result, determining that the current moment of the target to be determined is in a motion state, a static state or an uncertain state.

12. The method of claim 11, wherein the method further comprises:

a second state determination operation comprising:

if the first classification results of a plurality of continuous moments indicate that the target to be determined is in a set state, taking the last moment of the plurality of continuous moments as a reference moment, wherein the set state comprises the static state or the uncertain state;

and after the reference time, updating the current time at preset time intervals, acquiring the real-time speed distribution of the current time, and determining the state of the target to be determined according to the real-time speed distribution and the speed distribution corresponding to the reference time.

13. The method of claim 12, wherein the method further comprises:

updating the state of the target to be determined by the first state determining operation by using the determined state of the target to be determined.

14. The method of claim 12, wherein the method further comprises:

and monitoring the state of the object to be determined by the first state determining operation, and if the object to be determined is monitored to be converted from a static state or an uncertain state into a motion state, ending the second state determining operation.

15. The method of claim 1, wherein the method further comprises: and carrying out weight reduction or rejection processing on the point cloud data with the point cloud position change exceeding the change threshold.

16. The device for determining the target state comprises an acquisition module, a registration module and a prediction module;

the acquisition module is used for acquiring point cloud data of a target to be determined at the current moment;

the registration module is used for determining the speed distribution of the target to be determined by using a preset point cloud registration algorithm according to the point cloud data at the current moment;

the prediction module is used for predicting the point cloud position change of the target to be determined at the next moment according to the speed distribution of the target to be determined.

17. An electronic device, comprising: the system comprises a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface complete mutual communication through the communication bus;

the memory is used for storing at least one executable instruction, and the executable instruction causes the processor to execute the operation corresponding to the determination method of the target state according to any one of claims 1-15.

18. A computer storage medium having stored thereon a computer program which, when executed by a processor, carries out the method of determining a target state according to any one of claims 1 to 15.