阅读说明：本技术 一种基于仿人记忆机制的机器人概率地图更新方法 (Robot probability map updating method based on humanoid memory mechanism ) 是由 张波涛 王亚东 吴秋轩 吕强 仲朝亮 于 2020-12-24 设计创作，主要内容包括：本发明公开了一种基于仿人记忆机制的机器人概率地图更新方法,包括：针对目标物建立记忆和遗忘模型,将目标物在任意位置的概率赋予记忆属性和遗忘属性,以记忆量和遗忘速率组成遗忘曲线表示目标物在该位置的概率变化曲线；当目标物被重复识别时,触发重复识别量化机制,将每一个已发现过该目标物的位置对应的剩余记忆量和新增记忆量叠加得到对应位置的阶段起始记忆量,并赋予阶段遗忘速率,其中每个位置的新增记忆量根据历史统计数据中目标物出现次数占比得出；任意时刻下目标物在每个位置的概率,为该位置记忆量占所有位置记忆量的比重。本发明利用记忆模型提高长航时动态环境下地图更新的自适应性,从而为机器人提供相对准确的概率信息。(The invention discloses a robot probability map updating method based on a humanoid memory mechanism, which comprises the following steps: establishing a memory and forgetting model aiming at a target object, giving a memory attribute and a forgetting attribute to the probability of the target object at any position, and forming a forgetting curve by using the memory amount and the forgetting rate to represent the probability change curve of the target object at the position; when the target object is repeatedly identified, triggering a repeated identification quantification mechanism, overlapping the residual memory amount and the newly added memory amount corresponding to each position where the target object is found to obtain the stage initial memory amount of the corresponding position, and giving a stage forgetting rate, wherein the newly added memory amount of each position is obtained according to the ratio of the occurrence times of the target object in the historical statistical data; the probability of the target object at each position at any time is the proportion of the position memory amount to all the position memory amounts. The invention utilizes the memory model to improve the adaptivity of map updating in the dynamic environment during long voyage, thereby providing relatively accurate probability information for the robot.)

1. A robot probability map updating method based on a humanoid memory mechanism is characterized by comprising the following steps:

establishing a memory and forgetting model aiming at a target object, giving a memory attribute and a forgetting attribute to the probability of the target object at any position, and forming a forgetting curve by using the memory amount and the forgetting rate to represent the probability change curve of the target object at the position;

when the target object is repeatedly identified, triggering a repeated identification quantification mechanism, overlapping the residual memory amount and the newly added memory amount corresponding to each position where the target object is found to obtain the stage initial memory amount of the corresponding position, and giving a stage forgetting rate, wherein the newly added memory amount of each position is obtained according to the ratio of the occurrence times of the target object in the historical statistical data;

the probability of the target object at each position at any time is the proportion of the position memory amount to all the position memory amounts.

2. The robot probability map updating method based on the humanoid memory mechanism as claimed in claim 1, wherein the expression of the forgetting curve is as follows:

p_(x,y)＝p₀e^-kt,t∈(0,∞)；

wherein p (x, y) represents the amount of memory in a particular thing; p is a radical of₀Representing an initial amount of memory; k denotes the forgetting rate and t denotes time.

3. The method for updating the probability map of the robot based on the humanoid memory mechanism as claimed in claim 2, wherein the calculation method of the stage initial memory amount is as follows:

wherein d is_nIs the newly added memory amount, k, of this stage_nIs the stage forgetting rate, and the subscript n or n-1 represents the corresponding stage sequence number.

4. The robot probability map updating method based on the humanoid memory mechanism as claimed in claim 3, wherein the forgetting rate is calculated in a manner that:

wherein b is a forgetting adjustment coefficient; Δ t_nIs the interval time of two adjacent identifications; k is a radical of_nIs the stage forgetting rate.

5. The method as claimed in claim 3, wherein the robot probability map updating method based on the humanoid memory mechanism is characterized in that the new memory d of the target object at each position_nThe calculation mode is i/j, wherein i is the number of times that the target object is found at the position till the stage, and j is the sum of the number of times that the target object is found at all the positions till the stage.

6. The method for updating the probability map of the robot based on the humanoid memory mechanism as claimed in claim 4 or 5, wherein the forgetting adjustment coefficient has a value of 2.

7. The method for updating the probability map of the robot based on the humanoid memory mechanism as claimed in claim 4 or 5, further comprising an accelerated forgetting mechanism, wherein the accelerated forgetting mechanism is triggered when the target object is not identified again at any position within a preset time period, and comprises the following steps: adjusting the forgetting rate of the forgetting curve corresponding to the position of the found target object, and accelerating the forgetting rateWherein k is_nIs the forgetting rate before modification, m is the acceleration coefficient, and the value range is 0<m<1。

Technical Field

The invention belongs to the field of map information updating, and relates to a robot probability map updating method based on a human memory mechanism.

Background

The probability map is a process of quantitatively modeling the probability of existence of a target object under the background of taking target search as a task, and is an accurate measurement map. The probability map is mostly constructed by depending on sample statistics, the position probability of the current target object can be objectively reflected according to historical information, and the robot is prevented from performing global traversal when executing a target search task. The robot takes the probability information of the target object as reference to plan the path, and the task execution time can be effectively reduced. The probability map can objectively reflect the position probability information of the target object, provides reliable information support for a target search task, is widely applied to the fields of robot navigation, target search and the like, and has wide applicability. In the prior art, many path planning methods based on probability maps exist, such as an industrial robot path search optimization algorithm based on probability maps of application number CN201610257825.0, which is generally more efficient than the traditional path planning.

The premise of the above effect is the accuracy and effectiveness of the probability map. The unpredictable change of the environment can cause the outdating and the ineffectiveness of the probability map, so that the planning of the path is problematic, and the probability map can be better used in the related field only by solving the problems. Therefore, technical improvements to the probability map itself are also at hand.

In the related art, the invention of the authorization notice number CN1967151B provides a map data updating system and a map data updating method, which can prevent the mismatch of the map data between grids. However, for updating the probability map, a perfect scheme is lacking at present.

Disclosure of Invention

Aiming at the problem that the probability map in the prior art is lack of a reliable updating mode, the invention provides a robot probability map updating method of a human memory mechanism, which introduces a human memory forgetting rule, establishes a unique updating model related to the probability of a target object, designs a fading mode of forgetting rate during repeated identification on the basis, can accurately update the probability map, and improves the real-time property of the probability map to cope with the uncertainty change of the environment.

The technical scheme of the invention is as follows.

A robot probability map updating method based on a humanoid memory mechanism comprises the following steps:

establishing a memory and forgetting model aiming at a target object, giving a memory attribute and a forgetting attribute to the probability of the target object at any position, and forming a forgetting curve by using the memory amount and the forgetting rate to represent the probability change curve of the target object at the position; when the target object is repeatedly identified, triggering a repeated identification quantification mechanism, overlapping the residual memory amount and the newly added memory amount corresponding to each position where the target object is found to obtain the stage initial memory amount of the corresponding position, and giving a stage forgetting rate, wherein the newly added memory amount of each position is obtained according to the ratio of the occurrence times of the target object in the historical statistical data; the probability of the target object at each position at any time is the proportion of the position memory amount to all the position memory amounts.

The invention refers to the human memory model, because the memory quantity and the probability are reduced along with the time lapse, and the attenuation of the probability in the probability map is in accordance with the frame, therefore, the invention endows the change of the probability with new attribute, the memory quantity replaces the probability to calculate, and finally the probability is expressed by the proportion between the memory quantities; the self-adaptability of map updating in a dynamic environment during long-term navigation is improved, so that relatively accurate probability information is provided for the robot. When the robot carries out other work tasks, the memory amount is updated every time the robot finds a target object, and new probability information is generated. The superposition of the new memory capacity and the residual probability of the previous stage together form the initial value of the present stage. The probability normalization of all coordinate points is the global probability information.

Preferably, the expression of the forgetting curve is as follows:

p_(x,y)＝p₀e^-kt,t∈(0,∞)；

wherein p (x, y) represents the amount of memory in a particular thing; p is a radical of₀Representing an initial amount of memory; k represents forgetting rate and t representsAnd (3) removing the solvent. The expression is obtained by fitting a negative exponential function based on an Ebenhaos forgetting curve. This indicates that the memory amount decreases with the lapse of time, and the curve gradually flattens from steep as a whole.

Preferably, the calculation method of the phase initial memory amount is as follows:

wherein d is_nIs the newly added memory amount, k, of this stage_nIs the stage forgetting rate, and the subscript n or n-1 represents the corresponding stage sequence number. The memory amount is updated synchronously to adapt to dynamic external environment due to the repeated identification of the object, and when the object is found in one position, the memory amounts in the other positions are updated synchronously, which is different from the value of the newly-added memory amount.

Preferably, the forgetting rate is calculated by:

wherein b is a forgetting adjustment coefficient; Δ t_nIs the interval time of two adjacent identifications; k is a radical of_nIs the stage forgetting rate. Because the time intervals of repeated recognition are different, the forgetting rate should be different, and a reasonable calculation formula is provided to maximally help the curve to represent valuable probability information.

Preferably, the new memory d of the object at each position_nThe calculation mode is i/j, wherein i is the number of times that the target object is found at the position till the stage, and j is the sum of the number of times that the target object is found at all the positions till the stage. The calculation formula is activated for all positions at the same time in each repeated recognition, and results of different positions of the calculation formula are different, so that the change of respective probabilities is represented.

Preferably, the forgetting adjustment coefficient takes a value of 2. When the value of the forgetting adjustment coefficient b is larger, the difference of forgetting rates between the front stage and the rear stage is smaller, and after repeated verification, the real situation can be reflected most when the value is about 2, so that 2 is taken as an optimal value.

Preferably, the system further comprises an accelerated forgetting mechanism, wherein the accelerated forgetting mechanism is triggered when the target object is not identified again at any position within a preset time length, and the method comprises the following steps: adjusting the forgetting rate of the forgetting curve corresponding to the position of the found target object, and accelerating the forgetting rateWherein k is_nIs the forgetting rate before modification, m is the acceleration coefficient, and the value range is 0<m<1. Since in special cases, the target object cannot be found for a long time, and therefore the attenuation degree of the original curve cannot actually reflect the fact objectively, an accelerated forgetting mechanism is introduced to help the subsequent possible memory reconstruction or the target object is not in the map range.

The substantial effects of the invention include: the method makes up the defects of the traditional updating mode of the probability map, gives a human memory forgetting model, and improves the adaptivity of map updating in the dynamic environment during long voyage, thereby providing relatively accurate probability information for the robot.

Drawings

Fig. 1 is a schematic diagram of a forgetting curve in a repetitive identification process according to an embodiment of the present invention.

Detailed Description

The technical solution of the present application will be described with reference to the following examples. In addition, numerous specific details are set forth below in order to provide a better understanding of the present invention. It will be understood by those skilled in the art that the present invention may be practiced without some of these specific details. In some instances, methods, means, elements and circuits that are well known to those skilled in the art have not been described in detail so as not to obscure the present invention.

Example (b):

a robot probability map updating method based on a humanoid memory mechanism comprises the following steps:

establishing a memory and forgetting model aiming at a target object, giving a memory attribute and a forgetting attribute to the probability of the target object at any position, and forming a forgetting curve by using the memory amount and the forgetting rate to represent the probability change curve of the target object at the position; when the target object is repeatedly identified, triggering a repeated identification quantification mechanism, overlapping the residual memory amount and the newly added memory amount corresponding to each position where the target object is found to obtain the stage initial memory amount of the corresponding position, and giving a stage forgetting rate, wherein the newly added memory amount of each position is obtained according to the ratio of the occurrence times of the target object in the historical statistical data; the probability of the target object at each position at any time is the proportion of the position memory amount to all the position memory amounts.

Fig. 1 is a schematic diagram of a forgetting curve in a repeated recognition process, which shows the change of the memory of a target object in a single position within a few time periods, wherein the specific calculation mode is shown as follows.

The expression of the forgetting curve of the present embodiment is:

p_(x,y)＝p₀e^-kt,t∈(0,∞)；

wherein p (x, y) represents the amount of memory in a particular thing; p is a radical of₀Representing an initial amount of memory; k denotes the forgetting rate and t denotes time. The expression is obtained by fitting a negative exponential function based on an Ebenhaos forgetting curve. This indicates that the memory amount decreases with the lapse of time, and the curve gradually flattens from steep as a whole.

The calculation method of the initial memory amount of the phase is as follows:

wherein d is_nIs the newly added memory amount, k, of this stage_nIs the stage forgetting rate, the remaining memory amount in FIG. 1 is r_nInstead, the lower subscripted number, n, or n-1, represents the corresponding phase number. The memory amount will be updated synchronously to adapt to dynamic external environment due to repeated identification of the target object, and in addition, when one position finds the target object, the rest positionsThe memory amount of (2) is also updated synchronously, with the difference that the value of the newly added memory amount is different.

New memory d of the object at each position_nThe calculation mode is i/j, wherein i is the number of times that the target object is found at the position till the stage, and j is the sum of the number of times that the target object is found at all the positions till the stage. The calculation formula is activated for all positions at the same time in each repeated recognition, and results of different positions of the calculation formula are different, so that the change of respective probabilities is represented.

The forgetting rate of this embodiment is calculated as follows:

wherein b is a forgetting adjustment coefficient; Δ t_nIs the interval time of two adjacent identifications; k is a radical of_nIs the stage forgetting rate. Because the time intervals of repeated recognition are different, the forgetting rate should be different, and a reasonable calculation formula is provided to maximally help the curve to represent valuable probability information.

The forgetting adjustment coefficient b in this embodiment takes a value of 2. When the value of the forgetting adjustment coefficient b is larger, the difference of forgetting rates between the front stage and the rear stage is smaller, and after repeated verification, the real situation can be reflected most when the value is about 2, so that 2 is taken as an optimal value.

The embodiment further comprises an accelerated forgetting mechanism, wherein the accelerated forgetting mechanism is used for triggering when the target object is not identified again at any position within a preset time, and the accelerated forgetting mechanism comprises the following steps: adjusting the forgetting rate of the forgetting curve corresponding to the position of the found target object, and accelerating the forgetting rateWherein k is_nIs the forgetting rate before modification, m is the acceleration coefficient, and the value range is 0<m<1. Since in a special case there is a case where the target object cannot be found for a long time, the degree of attenuation of the original curve does not actually reflect the fact objectively, so thatAn accelerated forgetting mechanism is introduced to help the subsequent possible memory reconstruction or the condition that the target object is not in the map range.

The present embodiment refers to a human memory model, since both the memory amount and the probability will decrease with the lapse of time, and the attenuation of the probability in the probability map also conforms to the above framework, the present embodiment gives a new attribute to the change of the probability, calculates with the memory amount instead of the probability, and finally represents the probability by the specific gravity between the memory amounts; the self-adaptability of map updating in a dynamic environment during long-term navigation is improved, so that relatively accurate probability information is provided for the robot. When the robot carries out other work tasks, the memory amount is updated every time the robot finds a target object, and new probability information is generated. The superposition of the new memory capacity and the residual probability of the previous stage together form the initial value of the present stage. The probability normalization of all coordinate points is the global probability information.

The substantial effects of the present embodiment include: the method makes up the defects of the traditional updating mode of the probability map, gives a human memory forgetting model, and improves the adaptivity of map updating in the dynamic environment during long voyage, thereby providing relatively accurate probability information for the robot.

The embodiments provided in the present application are implemented in the form of software functional units, which can be stored in a readable storage medium. Based on such understanding, the technical solutions of the embodiments of the present application may be essentially or partially contributed to by the prior art, or all or part of the technical solutions may be embodied in the form of a software product, where the software product is stored in a storage medium and includes several instructions to enable a device (which may be a single chip, a chip, or the like) or a processor (processor) to execute all or part of the steps of the methods of the embodiments of the present application. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.

The above description is only for the specific embodiments of the present application, but the scope of the present application is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present application, and shall be covered by the scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.