CN114442625B - Environment map construction method and device based on multi-strategy combined control agent - Google Patents
Environment map construction method and device based on multi-strategy combined control agent Download PDFInfo
- Publication number
- CN114442625B CN114442625B CN202210076715.XA CN202210076715A CN114442625B CN 114442625 B CN114442625 B CN 114442625B CN 202210076715 A CN202210076715 A CN 202210076715A CN 114442625 B CN114442625 B CN 114442625B
- Authority
- CN
- China
- Prior art keywords
- environment
- boundary
- map
- strategy
- intelligent
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000010276 construction Methods 0.000 title claims abstract description 27
- 238000001514 detection method Methods 0.000 claims abstract description 58
- 230000033001 locomotion Effects 0.000 claims abstract description 44
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 24
- 238000013528 artificial neural network Methods 0.000 claims abstract description 22
- 239000003795 chemical substances by application Substances 0.000 claims description 77
- 238000000034 method Methods 0.000 claims description 61
- 230000008569 process Effects 0.000 claims description 32
- 238000005259 measurement Methods 0.000 claims description 16
- 238000012549 training Methods 0.000 claims description 14
- 239000003344 environmental pollutant Substances 0.000 claims description 10
- 231100000719 pollutant Toxicity 0.000 claims description 10
- 230000008859 change Effects 0.000 claims description 6
- 230000008447 perception Effects 0.000 claims description 5
- 238000007405 data analysis Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 abstract description 4
- 230000004888 barrier function Effects 0.000 abstract description 2
- 238000012544 monitoring process Methods 0.000 description 6
- 230000007613 environmental effect Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 230000004807 localization Effects 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 239000000356 contaminant Substances 0.000 description 1
- 238000011109 contamination Methods 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course or altitude of land, water, air, or space vehicles, e.g. automatic pilot
- G05D1/02—Control of position or course in two dimensions
- G05D1/021—Control of position or course in two dimensions specially adapted to land vehicles
- G05D1/0231—Control of position or course in two dimensions specially adapted to land vehicles using optical position detecting means
- G05D1/0238—Control of position or course in two dimensions specially adapted to land vehicles using optical position detecting means using obstacle or wall sensors
- G05D1/024—Control of position or course in two dimensions specially adapted to land vehicles using optical position detecting means using obstacle or wall sensors in combination with a laser
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course or altitude of land, water, air, or space vehicles, e.g. automatic pilot
- G05D1/02—Control of position or course in two dimensions
- G05D1/021—Control of position or course in two dimensions specially adapted to land vehicles
- G05D1/0255—Control of position or course in two dimensions specially adapted to land vehicles using acoustic signals, e.g. ultra-sonic singals
Abstract
The invention relates to an environment map construction method and device based on multi-strategy combined control intelligent agent, which mainly comprises three stages of boundary detection, global detection and map construction. Firstly, detecting the boundary of an unknown environment by adopting an improved wall-following strategy; secondly, dividing the area surrounded by the boundary into grids, sequentially taking randomly selected target points in each grid as the end points of the movement of the intelligent body by using a simplified spiral traversal strategy, and controlling the intelligent body to move by combining a D-Lite algorithm, so that the time for completing the global detection of the unknown environment is reduced; and finally, generating barrier distribution and concentration field distribution of the unknown environment through a space occupying grid map algorithm and an RBF neural network respectively. The invention realizes detection of the whole unknown environment by combining various strategies to control the intelligent agent, improves the calculation efficiency and can completely and accurately construct the environment map.
Description
Technical Field
The invention belongs to the field of intersection of information science and environmental science, and particularly relates to an environment map construction method and device based on multi-strategy combined control intelligent agent.
Background
Environmental monitoring is of great importance to public safety and sustainability of the ecosystem. Pollution source localization is one of the most critical issues in environmental monitoring applications. By positioning the pollution source, effective measures can be timely taken to prevent further expansion of pollution and reduce the risk of leakage of harmful substances. In recent years, automatic searching and positioning of pollution sources based on agents is becoming a new research direction. The problems of obstacle avoidance, path planning and the like are related to the positioning process of the pollution source of the intelligent body, and specific information of the monitoring environment needs to be known in advance. In the actual application process, the monitoring environment is mostly unknown, so that the movement of an intelligent agent needs to be controlled through an intelligent algorithm, and an environment map is quickly and accurately constructed.
The environment map construction is a process that an agent detects environment information by using a sensor carried by the agent and establishes an identifiable model. Currently, environmental map construction methods can be roughly classified into three categories: a geometric map-based method, a topological map-based method, and a grid map-based method. The first two methods are difficult to meet the actual application demands in terms of precision and real-time performance, but the grid map-based method has no limit, can be used for positioning and navigation of an intelligent body, and is the most widely-used method at present. The grid map-based method mainly comprises a rapid SLAM (Simultaneous Localization And Mapping, synchronous positioning and mapping), a micro SLAM, a Cartographer algorithm and the like. However, all the above three methods involve motion estimation and update matching, which cannot guarantee robustness and computational efficiency at the same time, and cannot meet the actual application requirements. In addition, the D site algorithm is often used for dynamic path planning in an unknown environment, but the method can only detect a grid area near a planned path in advance, cannot completely detect the whole environment, and cannot solve the problem that a target point is unknown. In order to detect the environment as much as possible, multiple applications are often required, so that part of the environment is repeatedly detected, and the calculation efficiency of the algorithm is reduced.
Disclosure of Invention
The invention aims to solve the technical problems of low calculation efficiency, incomplete map construction and the like of the existing method, and provides a novel method capable of effectively considering accuracy and timeliness. In order to achieve the technical purpose, the technical scheme of the invention is to provide an environment map construction method and device based on multi-strategy combined control intelligent agent.
According to one aspect of the invention, the invention provides an environment map construction method based on multi-strategy joint control agent, comprising the following steps:
step S1: based on an agent sensing model, an improved wall-following strategy is adopted to control the agent to move along the boundary of an unknown environment for one circle to carry out boundary detection, and the boundary of the unknown environment is obtained according to the pose of the agent in the boundary detection process and the measurement data analysis of the laser radar;
step S2: dividing a region surrounded by a boundary into a plurality of grids, randomly selecting a target point from each grid, and sequentially taking each target point as an end point of the movement of the intelligent body by adopting a simplified spiral traversal strategy;
step S3: dynamically planning paths of the intelligent agents reaching each end point by adopting a D-Lite algorithm, enabling the intelligent agents to sequentially reach the target points according to planned tracks, and performing global detection;
step S4: generating an obstacle distribution map of an unknown environment by adopting a space occupying grid map algorithm based on the pose of an intelligent agent and the measurement data of a laser radar in the global detection process;
step S5: training the RBF neural network based on self position information recorded in the movement process of the intelligent body and the collected pollutant concentration, and fitting the trained RBF neural network to generate a concentration distribution map of an unknown environment after the training is completed;
step S6: and combining the obstacle distribution map and the concentration distribution map to generate an environment map, and completing the map construction of the unknown environment.
Preferably, in step S1, the agent is a mobile carrier on which two ultrasonic sensors, a laser radar, a concentration sensor, and a photoelectric encoder are mounted; the two ultrasonic sensors are respectively positioned at the front and the right side of the intelligent body and are used for detecting whether obstacles exist at the front and the right side; the concentration sensor, the laser radar and the photoelectric encoder are positioned at the central position of the intelligent body, the concentration sensor is used for collecting the concentration of pollutants, the laser radar is used for scanning the whole environment and calculating the distribution condition of detection obstacles, and the photoelectric encoder is used for calculating the pose of the intelligent body, including coordinates and angles.
Preferably, in step S1, the improved wall-down strategy includes:
the intelligent body is placed at any boundary point, the front and back directions are parallel to the boundary, and the intelligent body is controlled to advance or turn according to the detection distance result of the ultrasonic sensor, so that the intelligent body is always on the left side of the boundary along the movement direction.
Preferably, in step S1, the improved wall-down strategy includes:
let the flag bit of ultrasonic sensor i be U i The value is as follows:
where l is the distance threshold and d is the distance between the agent and the boundary, when U i When the value is 1, the ultrasonic sensor i faces to the boundary, otherwise, the boundary is not present;
when the intelligent agent detects different boundary conditions, the zone bit U 1 U 2 There are four combinations, namely 00, 01, 10 and 11;
assuming that the current movement direction of the intelligent body is J, the value range is 0-3, and the values respectively correspond to the lower direction, the right direction, the upper direction and the left direction, the formula for controlling the change of the movement direction of the intelligent body according to the mark bit combination is as follows:
where Δj is the motion direction change value, i.e. the difference between the new motion direction and the current motion direction.
Preferably, in step S2, the simplified spiral traversal strategy includes:
and selecting clockwise, taking each target point as an end point of the movement of the intelligent agent in turn, and ensuring that the target points in each grid can be accessed only once.
Preferably, step S4 includes:
setting the scanning angle range of the laser radar to be-23 DEG to 23 DEG and the step angle to be 2.86 DEG in the global detection process, namely dividing the scanning range of the laser radar into 16 different scanning angles;
measuring data in the coverage range of a laser beam is obtained through laser radar scanning, and the probability p (I|z) that each point in an unknown environment is occupied by an obstacle is calculated by adopting a space-occupying grid algorithm t ) The specific formula is as follows:
wherein I represents the scanned point, z t Is the set of the measurement data of the laser radar and the pose of the intelligent agent at the time t, r max Is the farthest distance which can be detected by the laser radar, r I Is the distance from the center of the laser radar to I, phi I Is the relative angle between the laser radar and I, phi k Is equal to phi I The closest scan angle, k is the index number of the scan angle, r k Is phi k Detection distance of laser radar in direction, beta is the scanned step angle, p min And p max Is a constant, satisfy 0<p min <0.5<p max <1；
Updating the confidence coefficient of the moment I according to the probability that each point is occupied by an obstacle, and calculating the occupation probability based on the confidence coefficient, wherein the specific formula is as follows:
wherein C is t,I And C t-1,I The confidence levels at time t and time t-1I, respectively, and the initial confidence level C 0,I ＝0，p I Is the occupancy probability of I, the closer the occupancy probability is to 0,I, the more likely it is to be a free regionA domain; conversely, the closer the occupancy probability is to 1, the more likely it is that i is an obstacle region;
and the intelligent body continuously updates the occupation probability of each point detected by the laser radar according to the planned path movement, and generates an obstacle distribution map of an unknown environment.
Preferably, in step S5, the RBF neural network has a network structure with two inputs and one output and including 5 hidden layer nodes, the input is the position coordinates of each point in the environment area, and the output is the concentration value; and (3) learning and training the RBF neural network through the position and concentration value sample pairs recorded in the boundary detection and the global detection to obtain a trained RBF neural network, and completing fitting of concentration field distribution of the whole unknown environment.
According to a second aspect of the present invention, the present invention further provides an environment map construction device based on a multi-strategy joint control agent, including the following modules:
the boundary detection module is used for controlling the intelligent body to move along the boundary of the unknown environment for one circle by adopting an improved wall-following strategy based on the intelligent body perception model to carry out boundary detection, and analyzing the pose of the intelligent body and the measurement data of the laser radar in the boundary detection process to obtain the boundary of the unknown environment;
the spiral traversing module is used for dividing the area surrounded by the boundary into a plurality of grids, randomly selecting a target point from each grid, and sequentially taking each target point as an end point of the movement of the intelligent agent by adopting a simplified spiral traversing strategy;
the global detection module is used for dynamically planning paths of the intelligent agents reaching each end point by adopting a D-Lite algorithm, and enabling the intelligent agents to sequentially reach the target points according to the planned tracks to carry out global detection;
the obstacle distribution map generation module is used for generating an obstacle distribution map of an unknown environment by adopting a space-occupying grid map algorithm based on the pose of the intelligent agent and the measurement data of the laser radar in the global detection process;
the concentration distribution map generation module is used for training the RBF neural network based on the self position information recorded in the movement process of the intelligent body and the collected pollutant concentration, and after the training is finished, the trained RBF neural network is adopted for fitting to generate a concentration distribution map of an unknown environment;
the environment map generation module is used for combining the obstacle distribution map and the concentration distribution map to generate an environment map and complete the map construction of the unknown environment.
The technical scheme provided by the invention has the technical effects that:
the method and the device for constructing the environment map of the multi-strategy combined control intelligent agent are mainly divided into three stages of boundary detection, global detection and map construction. Firstly, detecting the boundary of an unknown environment through an improved wall-following strategy; secondly, dividing the area surrounded by the boundary into grids, sequentially taking randomly selected target points in each grid as the end points of the movement of the intelligent body by using a simplified spiral traversal strategy, and controlling the intelligent body to move by combining a D-Lite algorithm, so that the time for completing the global detection of the unknown environment is reduced; and finally, generating barrier distribution and concentration field distribution of the unknown environment through a space occupying grid map algorithm and an RBF neural network respectively. According to the invention, the movement of the intelligent agent is controlled through multiple strategies, the obstacle area is generated in the scanning and monitoring environment, and meanwhile, the concentration field distribution of the pollution source is fitted according to the concentration information acquired by the sensor and the position information of the sensor. The multi-strategy combined control intelligent body moves in an unknown monitoring environment, so that an environment map can be quickly and accurately constructed, and the problems of low calculation efficiency and incomplete map construction in the existing method are effectively solved.
Drawings
The invention will be further described with reference to the accompanying drawings and examples, in which:
FIG. 1 is a schematic diagram of an environment map construction method based on multi-strategy joint control agent of the present invention;
FIG. 2 is a map of six unknown environments and the present invention, wherein FIG. 2 (a) is a real environment map and FIG. 2 (b) is a constructed environment map;
FIG. 3 is a motion profile of an agent in six environments of the present invention;
FIG. 4 is a comparison of the map construction results of three different methods according to the present invention, wherein FIG. 4 (a) is a real environment map, and FIGS. 4 (B) to 4 (d) are environment maps constructed according to methods A, B and the present invention, respectively;
fig. 5 is a concentration profile of an RBF neural network fit of the invention, wherein fig. 5 (a) is a true concentration map and fig. 5 (b) is a fit concentration map.
Fig. 6 is a block diagram of an environment map construction apparatus based on a multi-strategy joint control agent according to the present invention.
Detailed Description
For a clearer understanding of technical features, objects and effects of the present invention, a detailed description of embodiments of the present invention will be made with reference to the accompanying drawings.
In this embodiment, the Python is used to write the corresponding code, and the Pygame module is used to perform the simulation experiment. In this embodiment, six different environments (150 cm×105cm in length and width) are designed, and different numbers and shapes of obstacles are distributed in each environment, so as to simulate a complex practical environment. A single gaussian concentration distribution model is adopted to simulate point pollution sources in the environment, and the function is defined as follows:
wherein, (x) 0 ,y 0 ) Is the position of the pollution source, (x, y) is the coordinate of any point in the environment, c (x, y) is the concentration of the point, and the value range is (0, 1)]。
Referring to fig. 1, fig. 1 is a schematic diagram of an environment map construction method based on multi-strategy joint control agent, which comprises the following specific implementation steps:
step S1: based on an agent perception model, an improved wall-following strategy is adopted to control an agent to move along the boundary of an unknown environment for one circle to carry out boundary detection, and the boundary of the unknown environment is obtained according to the pose of the agent in the boundary detection process and the measurement data analysis of a laser radar.
The intelligent body is a mobile carrier provided with two ultrasonic sensors, a laser radar, a concentration sensor and a photoelectric encoder, and has the functions of detecting distance, scanning environment and detecting pollutant concentration. The two ultrasonic sensors are respectively positioned at the front and the right side of the intelligent body and are used for detecting whether obstacles exist at the front and the right side; the concentration sensor, the laser radar and the photoelectric encoder are positioned at the central position of the intelligent body, the concentration sensor is used for collecting the concentration of pollutants, the laser radar is used for scanning the whole environment and calculating the distribution condition of the detected obstacles, and the photoelectric encoder is used for calculating the pose of the intelligent body, including coordinates and angles.
The above-described system for sensing an unknown environment around an agent using the functions of detecting distance, scanning the environment, and detecting the concentration of contaminants is referred to as an agent sensing model.
And in the boundary detection process of the intelligent agent perception model, obtaining the boundary of the unknown environment according to the pose of the intelligent agent and the measurement data of the laser radar.
In step S1, the improved wall-down strategy specifically includes:
let the flag bit of ultrasonic sensor i be U i The value is as follows:
where l is the distance threshold and d is the distance between the agent and the boundary, when U i When the value is 1, the ultrasonic sensor i faces to the boundary, otherwise, the boundary is not present;
when the intelligent agent detects different boundary conditions, the zone bit U 1 U 2 There are four combinations, namely 00, 01, 10 and 11;
assuming that the current movement direction of the intelligent body is J, the value range is 0-3, and the values respectively correspond to the lower direction, the right direction, the upper direction and the left direction, the formula for controlling the change of the movement direction of the intelligent body according to the mark bit combination is as follows:
where Δj is the motion direction change value, i.e. the difference between the new motion direction and the current motion direction.
In this embodiment, the distance threshold l=1cm of the ultrasonic sensor is set, the movement direction of the intelligent body is adjusted according to the formula (3), the intelligent body is ensured to be always at the left side of the boundary along the movement direction, and the boundary detection of the unknown environment is completed.
Step S2: dividing the area surrounded by the boundary into a plurality of grids, randomly selecting a target point in each grid, and sequentially taking each target point as an end point of the movement of the intelligent body by adopting a simplified spiral traversing strategy.
In this embodiment, the unknown area surrounded by the boundary is divided into 9 grids, one target point is randomly selected from each large grid, and the target points are sequentially used as the end points of the movement of the intelligent agent in a clockwise direction.
Step S3: dynamically planning paths of the intelligent agents reaching each end point by adopting a D-Lite algorithm, enabling the intelligent agents to sequentially reach the target points according to planned tracks, and performing global detection;
in the global detection process, the intelligent agent only needs to sequentially reach the target points according to the planned path, and the whole environment area is not required to be fully covered.
Step S4: based on the pose of the intelligent agent and the measurement data of the laser radar in the global detection process, an obstacle distribution map of an unknown environment is generated by adopting a space-occupying grid map algorithm.
In this embodiment, step S4 specifically includes:
setting the scanning angle range of the laser radar to be-23 DEG to 23 DEG and the step angle to be 2.86 DEG in the global detection process, namely dividing the scanning range of the laser radar into 16 different scanning angles;
measuring data in the coverage range of a laser beam is obtained through laser radar scanning, and the probability p (I|z) that each point in an unknown environment is occupied by an obstacle is calculated by adopting a space-occupying grid algorithm t ) The specific formula is as follows:
wherein, table IShowing the scanned points, z t Is the set of the measurement data of the laser radar and the pose of the intelligent agent at the time t, r max Is the farthest distance which can be detected by the laser radar, r I Is the distance from the center of the laser radar to I, phi I Is the relative angle between the laser radar and I, phi k Is equal to phi I The closest scan angle, k is the index number of the scan angle, r k Is phi k Detection distance of laser radar in direction, beta is the scanned step angle, p min And p max Is a constant, satisfy 0<p min <0.5<p max <1, empirical value p min ＝0.3、p max =0.7. The first case in the formula (4) corresponds to the undetected area by the point, and the probability value is 0.5, which indicates whether the point is occupied or not; the second case indicates that the point is likely to be occupied, and a larger probability needs to be given; the third case indicates that the point is less likely to be occupied, giving a smaller probability.
Updating the confidence coefficient of the moment I according to the probability that each point is occupied by an obstacle, and calculating the occupation probability based on the confidence coefficient, wherein the specific formula is as follows:
wherein C is t,I And C t-1,I The confidence levels at time t and time t-1I, respectively, and the initial confidence level C 0,I ＝0，p I Is the occupancy probability of I, the closer the occupancy probability is to 0,I the more likely it is to be a free region; conversely, the closer the occupancy probability is to 1, the more likely it is that i is an obstacle region;
in the embodiment, probability of being occupied by an obstacle is given to each point in the environment map according to the formula (4), confidence of the point is updated according to the formula (5), the occupation probability is calculated based on the confidence, the occupation probability of each point detected by the laser radar is continuously updated in the movement of the intelligent body according to the planned path, and finally an obstacle distribution map of an unknown environment is generated.
Step S5: and training the RBF neural network based on the self position information recorded in the body movement process (comprising the boundary detection process and the global detection process) and the acquired pollutant concentration, and fitting by adopting the trained RBF neural network to generate a concentration distribution map of an unknown environment after the training is completed.
Step S6: and combining the obstacle distribution map and the concentration distribution map to generate an environment map, and completing the map construction of the unknown environment.
In this embodiment, the multi-strategy joint control agent algorithm provided by the invention is applied to map reconstruction of an unknown environment. In order to verify the effectiveness of the present invention, experiments are performed on six different environments, and a constructed environment map and a motion track of an intelligent agent are drawn, specifically please refer to fig. 2 and 3. Comparing fig. 2 (a) and fig. 2 (b), the map of six unknown environments is constructed more accurately, and only a few edge areas have deletions (the ratio of the missing parts is less than 1% of the full map). This is because the scanning range of the lidar is a sector, and when the sector area is divided by the grid to calculate the occupancy probability, a situation that a part of the grid area overlaps or the sector part area is not divided by the grid occasionally occurs, so that the occupancy probability of the area is calculated erroneously, and thus the finally generated obstacle map is missing. As can be seen from fig. 3, the agent first moves around the boundary for one circle, detects the environment boundary, then moves to each target point in a spiral traversing manner, finally stops at the last target point, and basically completes coverage detection of all unknown environments.
To further embody the advancement of the present invention, taking environment 2 and environment 3 as examples, the present invention and two other methods are compared: the method A does not adopt a wall walking strategy and a spiral traversing strategy, and randomly selected points are used as target points when an intelligent body moves; and the method B does not adopt a wall-following strategy to control the intelligent agent to move. The environment map constructed by the three methods is shown in fig. 4. As can be seen from fig. 4, the method a has the worst effect, many parts of the environment are not detected, and the generated map is far from the real environment; the method B completes detection of most of the environment areas, and a small amount of boundary areas are not detected; the invention completely detects the whole environment, and the result is closest to the real map.
Fig. 5 shows a concentration profile of RBF neural network fits. Although the predicted concentration near the source of contamination is slightly below the true value, the predicted concentration profile trend and the location of the concentration peak are substantially consistent with the true situation.
In some embodiments, referring to fig. 6, there is also provided an environment map construction apparatus based on a multi-strategy joint control agent, including the following modules:
the boundary detection module 1 is used for controlling an intelligent body to move along the boundary of an unknown environment for one circle by adopting an improved wall-following strategy based on the intelligent body perception model to carry out boundary detection, and analyzing and obtaining the boundary of the unknown environment according to the pose of the intelligent body and the measurement data of the laser radar in the boundary detection process;
the spiral traversing module 2 is used for dividing the area surrounded by the boundary into a plurality of grids, randomly selecting a target point from each grid, and sequentially taking each target point as an end point of the movement of the intelligent agent by adopting a simplified spiral traversing strategy;
the global detection module 3 is used for dynamically planning paths of the intelligent agents reaching each end point by adopting a D-Lite algorithm, and enabling the intelligent agents to sequentially reach the target points according to the planned tracks to carry out global detection;
the obstacle distribution map generation module 4 is used for generating an obstacle distribution map of an unknown environment by adopting a space-occupying grid map algorithm based on the pose of the intelligent agent and the measurement data of the laser radar in the global detection process;
the concentration distribution map generation module 5 is used for training the RBF neural network based on the self position information recorded in the movement process of the intelligent body and the collected pollutant concentration, and fitting the RBF neural network after the training is completed to generate a concentration distribution map of an unknown environment;
the environment map generation module 6 is used for combining the obstacle distribution map and the concentration distribution map to generate an environment map and complete the map construction of the unknown environment.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or system that comprises the element.
The foregoing embodiment numbers of the present invention are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments. In the unit claims enumerating several means, several of these means may be embodied by one and the same item of hardware. The use of the terms first, second, third, etc. do not denote any order, but rather the terms first, second, third, etc. are used to interpret the terms as labels.
The foregoing description is only of the preferred embodiments of the present invention, and is not intended to limit the scope of the invention, but rather is intended to cover any equivalents of the structures or equivalent processes disclosed herein or in the alternative, which may be employed directly or indirectly in other related arts.
Claims (8)
1. The environment map construction method based on the multi-strategy combined control agent is characterized by comprising the following steps:
step S1: based on an agent sensing model, an improved wall-following strategy is adopted to control the agent to move along the boundary of an unknown environment for one circle to carry out boundary detection, and the boundary of the unknown environment is obtained according to the pose of the agent in the boundary detection process and the measurement data analysis of the laser radar;
step S2: dividing a region surrounded by a boundary into a plurality of grids, randomly selecting a target point from each grid, and sequentially taking each target point as an end point of the movement of the intelligent body by adopting a simplified spiral traversal strategy;
step S3: dynamically planning paths of the intelligent agents reaching each end point by adopting a D-Lite algorithm, enabling the intelligent agents to sequentially reach the target points according to planned tracks, and performing global detection;
step S4: generating an obstacle distribution map of an unknown environment by adopting a space occupying grid map algorithm based on the pose of an intelligent agent and the measurement data of a laser radar in the global detection process;
step S5: training the RBF neural network based on self position information recorded in the movement process of the intelligent body and the collected pollutant concentration, and fitting the trained RBF neural network to generate a concentration distribution map of an unknown environment after the training is completed;
step S6: and combining the obstacle distribution map and the concentration distribution map to generate an environment map, and completing the map construction of the unknown environment.
2. The method for constructing an environment map based on a multi-strategy joint control agent according to claim 1, wherein in step S1, the agent is a mobile carrier on which two ultrasonic sensors, a laser radar, a concentration sensor and a photoelectric encoder are mounted; the two ultrasonic sensors are respectively positioned at the front and the right side of the intelligent body and are used for detecting whether obstacles exist at the front and the right side; the concentration sensor, the laser radar and the photoelectric encoder are positioned at the central position of the intelligent body, the concentration sensor is used for collecting the concentration of pollutants, the laser radar is used for scanning the whole environment and calculating the distribution condition of detection obstacles, and the photoelectric encoder is used for calculating the pose of the intelligent body, including coordinates and angles.
3. The method for building an environment map based on multi-strategy joint control agent according to claim 1, wherein in step S1, the improved wall-down strategy comprises:
the intelligent body is placed at any boundary point, the front and back directions are parallel to the boundary, and the intelligent body is controlled to advance or turn according to the detection distance result of the ultrasonic sensor, so that the intelligent body is always on the left side of the boundary along the movement direction.
4. The method for building an environment map based on multi-strategy joint control agent according to claim 1, wherein in step S1, the improved wall-down strategy comprises:
let the flag bit of ultrasonic sensor i be U i The value is as follows:
where l is the distance threshold and d is the distance between the agent and the boundary, when U i When the value is 1, the ultrasonic sensor i faces to the boundary, otherwise, the boundary is not present;
when the intelligent agent detects different boundary conditions, the zone bit U 1 U 2 There are four combinations, namely 00, 01, 10 and 11;
assuming that the current movement direction of the intelligent body is J, the value range is 0-3, and the values respectively correspond to the lower direction, the right direction, the upper direction and the left direction, the formula for controlling the change of the movement direction of the intelligent body according to the mark bit combination is as follows:
where Δj is the motion direction change value, i.e. the difference between the new motion direction and the current motion direction.
5. The method for building an environment map based on multi-strategy joint control agent according to claim 1, wherein in step S2, the simplified spiral traversal strategy comprises:
and selecting clockwise, taking each target point as an end point of the movement of the intelligent agent in turn, and ensuring that the target points in each grid can be accessed only once.
6. The environment map construction method based on multi-strategy joint control agent as claimed in claim 1, wherein the step S4 comprises:
setting the scanning angle range of the laser radar to be-23 DEG to 23 DEG and the step angle to be 2.86 DEG in the global detection process, namely dividing the scanning range of the laser radar into 16 different scanning angles;
measuring data in the coverage range of a laser beam is obtained through laser radar scanning, and the probability p (I|z) that each point in an unknown environment is occupied by an obstacle is calculated by adopting a space-occupying grid algorithm t ) The specific formula is as follows:
wherein I represents the scanned point, z t Is the set of the measurement data of the laser radar and the pose of the intelligent agent at the time t, r max Is the farthest distance which can be detected by the laser radar, r I Is the distance from the center of the laser radar to I, phi I Is the relative angle between the laser radar and I, phi k Is equal to phi I The closest scan angle, k is the index number of the scan angle, r k Is phi k Detection distance of laser radar in direction, beta is the scanned step angle, p min And p max Is a constant, satisfy 0<p min <0.5<p max <1；
Updating the confidence coefficient of the moment I according to the probability that each point is occupied by an obstacle, and calculating the occupation probability based on the confidence coefficient, wherein the specific formula is as follows:
wherein C is t,I And C t-1,I The confidence levels at time t and time t-1I, respectively, and the initial confidence level C 0,I ＝0，p I Is the occupancy probability of I, the closer the occupancy probability is to 0,I the more likely it is to be a free region; conversely, the closer the occupancy probability is to 1, the more likely it is that i is an obstacle region;
and the intelligent body continuously updates the occupation probability of each point detected by the laser radar according to the planned path movement, and generates an obstacle distribution map of an unknown environment.
7. The method for building an environment map based on multi-strategy joint control agent according to claim 1, wherein in step S5, the RBF neural network is a network structure having two inputs and one output and including 5 hidden layer nodes, the input is position coordinates of each point in the environment area, and the output is a concentration value; and (3) learning and training the RBF neural network through the position and concentration value sample pairs recorded in the boundary detection and the global detection to obtain a trained RBF neural network, and completing fitting of concentration field distribution of the whole unknown environment.
8. The environment map construction device based on the multi-strategy joint control agent is characterized by comprising the following modules:
the boundary detection module is used for controlling the intelligent body to move along the boundary of the unknown environment for one circle by adopting an improved wall-following strategy based on the intelligent body perception model to carry out boundary detection, and analyzing the pose of the intelligent body and the measurement data of the laser radar in the boundary detection process to obtain the boundary of the unknown environment;
the spiral traversing module is used for dividing the area surrounded by the boundary into a plurality of grids, randomly selecting a target point from each grid, and sequentially taking each target point as an end point of the movement of the intelligent agent by adopting a simplified spiral traversing strategy;
the global detection module is used for dynamically planning paths of the intelligent agents reaching each end point by adopting a D-Lite algorithm, and enabling the intelligent agents to sequentially reach the target points according to the planned tracks to carry out global detection;
the obstacle distribution map generation module is used for generating an obstacle distribution map of an unknown environment by adopting a space-occupying grid map algorithm based on the pose of the intelligent agent and the measurement data of the laser radar in the global detection process;
the concentration distribution map generation module is used for training the RBF neural network based on the self position information recorded in the movement process of the intelligent body and the collected pollutant concentration, and after the training is finished, the trained RBF neural network is adopted for fitting to generate a concentration distribution map of an unknown environment;
the environment map generation module is used for combining the obstacle distribution map and the concentration distribution map to generate an environment map and complete the map construction of the unknown environment.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210076715.XA CN114442625B (en) | 2022-01-24 | 2022-01-24 | Environment map construction method and device based on multi-strategy combined control agent |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210076715.XA CN114442625B (en) | 2022-01-24 | 2022-01-24 | Environment map construction method and device based on multi-strategy combined control agent |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114442625A CN114442625A (en) | 2022-05-06 |
| CN114442625B true CN114442625B (en) | 2023-06-06 |
Family
ID=81370603
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210076715.XA Active CN114442625B (en) | 2022-01-24 | 2022-01-24 | Environment map construction method and device based on multi-strategy combined control agent |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114442625B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2024016101A1 (en) * | 2022-07-18 | 2024-01-25 | Intelligent Cleaning Equipment Holdings Co., Ltd. | Systems and methods for robotic navigation, teaching and mapping |
| CN115316887B (en) * | 2022-10-17 | 2023-02-28 | 杭州华橙软件技术有限公司 | Robot control method, robot, and computer-readable storage medium |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2013071190A1 (en) * | 2011-11-11 | 2013-05-16 | Evolution Robotics, Inc. | Scaling vector field slam to large environments |
| CN106840168A (en) * | 2017-03-16 | 2017-06-13 | 苏州大学 | Complete coverage path planning method under clean robot and its dynamic environment |
| CN108983781A (en) * | 2018-07-25 | 2018-12-11 | 北京理工大学 | A kind of environment detection method in unmanned vehicle target acquisition system |
| CN109059931A (en) * | 2018-09-05 | 2018-12-21 | 北京航空航天大学 | A kind of paths planning method based on multiple agent intensified learning |
| CN109213169A (en) * | 2018-09-20 | 2019-01-15 | 湖南万为智能机器人技术有限公司 | The paths planning method of mobile robot |
| WO2019190395A1 (en) * | 2018-03-28 | 2019-10-03 | Agency For Science, Technology And Research | Method and system for returning a displaced autonomous mobile robot to its navigational path |
| CN111220999A (en) * | 2019-11-29 | 2020-06-02 | 大连海事大学 | Restricted space detection system and method based on instant positioning and mapping technology |
| CN111598916A (en) * | 2020-05-19 | 2020-08-28 | 金华航大北斗应用技术有限公司 | Preparation method of indoor occupancy grid map based on RGB-D information |
| CN112558599A (en) * | 2020-11-06 | 2021-03-26 | 深圳拓邦股份有限公司 | Robot work control method and device and robot |
| WO2022007350A1 (en) * | 2020-07-08 | 2022-01-13 | 格力电器（武汉）有限公司 | Global path planning method and apparatus, terminal, and readable storage medium |
-
2022
- 2022-01-24 CN CN202210076715.XA patent/CN114442625B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2013071190A1 (en) * | 2011-11-11 | 2013-05-16 | Evolution Robotics, Inc. | Scaling vector field slam to large environments |
| CN106840168A (en) * | 2017-03-16 | 2017-06-13 | 苏州大学 | Complete coverage path planning method under clean robot and its dynamic environment |
| WO2019190395A1 (en) * | 2018-03-28 | 2019-10-03 | Agency For Science, Technology And Research | Method and system for returning a displaced autonomous mobile robot to its navigational path |
| CN108983781A (en) * | 2018-07-25 | 2018-12-11 | 北京理工大学 | A kind of environment detection method in unmanned vehicle target acquisition system |
| CN109059931A (en) * | 2018-09-05 | 2018-12-21 | 北京航空航天大学 | A kind of paths planning method based on multiple agent intensified learning |
| CN109213169A (en) * | 2018-09-20 | 2019-01-15 | 湖南万为智能机器人技术有限公司 | The paths planning method of mobile robot |
| CN111220999A (en) * | 2019-11-29 | 2020-06-02 | 大连海事大学 | Restricted space detection system and method based on instant positioning and mapping technology |
| CN111598916A (en) * | 2020-05-19 | 2020-08-28 | 金华航大北斗应用技术有限公司 | Preparation method of indoor occupancy grid map based on RGB-D information |
| WO2022007350A1 (en) * | 2020-07-08 | 2022-01-13 | 格力电器（武汉）有限公司 | Global path planning method and apparatus, terminal, and readable storage medium |
| CN112558599A (en) * | 2020-11-06 | 2021-03-26 | 深圳拓邦股份有限公司 | Robot work control method and device and robot |
Non-Patent Citations (6)
| Title |
|---|
| Bernd-Helge Schafer ; Christopher Armbrust ; Tobias Fohst ; Karsten Berns.The Application of Design Schemata in Off-Road Robotics.《 IEEE Intelligent Transportation Systems Magazine 》.2013,全文. * |
| Concurrent dynamic programming for grid-based problems and its application for real-time path planning;Stephen Cossell , José Guivant;《Robotics and Autonomous Systems》;全文 * |
| Yaru Kang ; Dianxi Shi.A Research on Area Coverage Algorithm for Robotics.《2018 IEEE International Conference of Intelligent Robotic and Control Engineering (IRCE)》.2018,全文. * |
| 任泉林 ; 臧铁钢 ; 崔海程.粮仓库区移动机器人导航系统设计.《电子技术与软件工程》.2021,全文. * |
| 基于电子海图栅格化的无人水面艇全局路径规划;范云生;赵永生;石林龙;张月;;中国航海(01);全文 * |
| 障碍物场景下基于信息熵的气体源定位搜索;王艺博;《中国优秀硕士学位论文全文数据库 工程科技Ⅰ辑》;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114442625A (en) | 2022-05-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Fox | Markov localization-a probabilistic framework for mobile robot localization and navigation. | |
| CN114442625B (en) | Environment map construction method and device based on multi-strategy combined control agent | |
| Marjovi et al. | Multi-robot olfactory search in structured environments | |
| EP3336489A1 (en) | Method and system for automatically establishing map indoors by mobile robot | |
| Wolf et al. | Mobile robot simultaneous localization and mapping in dynamic environments | |
| Crowley | Navigation for an intelligent mobile robot | |
| CN112000754A (en) | Map construction method and device, storage medium and computer equipment | |
| Rekleitis | Cooperative localization and multi-robot exploration | |
| Hu et al. | Landmark‐based navigation of industrial mobile robots | |
| Bosse | ATLAS: a framework for large scale automated mapping and localization | |
| Wallgrün | Autonomous construction of hierarchical voronoi-based route graph representations | |
| Lee et al. | A reliable position estimation method of the service robot by map matching | |
| JP2009223757A (en) | Autonomous mobile body, control system, and self-position estimation method | |
| CN112880694A (en) | Method for determining the position of a vehicle | |
| CN114879660B (en) | Robot environment sensing method based on target drive | |
| Carinena et al. | Landmark detection in mobile robotics using fuzzy temporal rules | |
| Zhang et al. | Feature extraction for outdoor mobile robot navigation based on a modified Gauss–Newton optimization approach | |
| Li et al. | Robot map building from sonar sensors and DSmT | |
| Ruiz-Mayor et al. | Perceptual ambiguity maps for robot localizability with range perception | |
| Jae-Bok | Mobile robot localization using range sensors: Consecutive scanning and cooperative scanning | |
| Kwon et al. | Thinning-based topological exploration using position possibility of topological nodes | |
| Zelinsky | Environment mapping with a mobile robot using sonar | |
| Lee et al. | Probabilistic localization of the service robot by map matching algorithm | |
| Li et al. | Robot navigation method based on intelligent evolution | |
| Wang et al. | Multistrategy-Based Environment Map Building Using a Mobile Agent |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| EE01 | Entry into force of recordation of patent licensing contract | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20220506 Assignee: Hubei Hongrui Membrane Technology Co.,Ltd. Assignor: CHINA University OF GEOSCIENCES (WUHAN CITY) Contract record no.: X2024980001471 Denomination of invention: A method and device for constructing environmental maps based on multi strategy joint control intelligent agents Granted publication date: 20230606 License type: Common License Record date: 20240129 Application publication date: 20220506 Assignee: Wuhan Xingqi Technology Co.,Ltd. Assignor: CHINA University OF GEOSCIENCES (WUHAN CITY) Contract record no.: X2024980001469 Denomination of invention: A method and device for constructing environmental maps based on multi strategy joint control intelligent agents Granted publication date: 20230606 License type: Common License Record date: 20240129 Application publication date: 20220506 Assignee: Wuhan Xintiande Energy Technology Co.,Ltd. Assignor: CHINA University OF GEOSCIENCES (WUHAN CITY) Contract record no.: X2024980001464 Denomination of invention: A method and device for constructing environmental maps based on multi strategy joint control intelligent agents Granted publication date: 20230606 License type: Common License Record date: 20240129 Application publication date: 20220506 Assignee: Wuhan Shitu Technology Co.,Ltd. Assignor: CHINA University OF GEOSCIENCES (WUHAN CITY) Contract record no.: X2024980001462 Denomination of invention: A method and device for constructing environmental maps based on multi strategy joint control intelligent agents Granted publication date: 20230606 License type: Common License Record date: 20240129 Application publication date: 20220506 Assignee: Wuhan Rongguo Biotechnology Co.,Ltd. Assignor: CHINA University OF GEOSCIENCES (WUHAN CITY) Contract record no.: X2024980001461 Denomination of invention: A method and device for constructing environmental maps based on multi strategy joint control intelligent agents Granted publication date: 20230606 License type: Common License Record date: 20240129 |