CN115268467B - Navigation control system and control method of luggage van - Google Patents

Navigation control system and control method of luggage van Download PDF

Info

Publication number
CN115268467B
CN115268467B CN202211170194.0A CN202211170194A CN115268467B CN 115268467 B CN115268467 B CN 115268467B CN 202211170194 A CN202211170194 A CN 202211170194A CN 115268467 B CN115268467 B CN 115268467B
Authority
CN
China
Prior art keywords
path
state
luggage van
information
control
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
Application number
CN202211170194.0A
Other languages
Chinese (zh)
Other versions
CN115268467A (en
Inventor
马列
马海兵
沈亮
马琼
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Jiangsu Tianyi Aviation Industry Co Ltd
Original Assignee
Jiangsu Tianyi Aviation Industry Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Jiangsu Tianyi Aviation Industry Co Ltd filed Critical Jiangsu Tianyi Aviation Industry Co Ltd
Priority to CN202211170194.0A priority Critical patent/CN115268467B/en
Publication of CN115268467A publication Critical patent/CN115268467A/en
Application granted granted Critical
Publication of CN115268467B publication Critical patent/CN115268467B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

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
    • G05D1/0212—Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory
    • G05D1/0214—Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory in accordance with safety or protection criteria, e.g. avoiding hazardous areas
    • 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
    • G05D1/0212—Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory
    • G05D1/0221—Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory involving a learning process

Landscapes

  • Engineering & Computer Science (AREA)
  • Aviation & Aerospace Engineering (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Control Of Position, Course, Altitude, Or Attitude Of Moving Bodies (AREA)

Abstract

The invention discloses a navigation control system and a control method of a luggage van, wherein the system comprises a man-machine interaction unit, a sensing unit, a path planning unit and a control unit; the path planning unit carries out path planning according to the obtained target position information, the environment information and the state information of the luggage van, and when the luggage van cannot detect the dynamic barrier, a global path plan is generated; when a dynamic obstacle is detected; generating a local path plan according to a machine learning algorithm; and sending the path plan to a control unit and a man-machine interaction unit; the control unit controls the luggage van to carry luggage according to the path planning result, and a user can check the path planning information through the man-machine interaction unit and move to a target position along with the luggage van, so that automatic obstacle avoidance control of the luggage van is realized. The invention can keep a constant distance between the luggage van and the mobile terminal carried by the user, generate a smooth, safe and efficient path track and realize effective dynamic obstacle avoidance.

Description

Navigation control system and control method of luggage van
Technical Field
The invention belongs to the technical field of automation, and particularly relates to a navigation control system and a control method for a luggage van.
Background
The luggage van can help people to carry various luggage articles in a trip or work, reduces the carrying burden of passengers, and frees both hands. However, the existing luggage barrow mostly needs to be pushed by hand, so that more manpower is needed to be paid out, and people are difficult to do other things at the same time because both hands are occupied when the barrow goes forward. The automatic luggage van also has the defects of poor ability of following the user, obstacle avoidance and path planning, easy falling into local optimum, and the conditions of collision or detour and the like.
Disclosure of Invention
The invention aims to provide a navigation control system and a control method of a luggage van. Can realize that the luggage van follows automatically, keeps away the barrier automatically.
The technical scheme of the invention is as follows: a navigation control system for a luggage cart, the system comprising: the system comprises a human-computer interaction unit, a sensing unit, a path planning unit and a control unit;
the man-machine interaction unit comprises a display screen and a voice input module, wherein the display screen is used for displaying an environment map and path planning information; the voice input module is used for voice interaction with a user, so that the starting and stopping of the luggage van are realized, and a voice consultation function is realized;
the sensing unit comprises a laser radar, a vision sensor and a vehicle-mounted state sensor, and obtains environment information, obstacle information and state information of the luggage van around the luggage van through the laser radar, the vision sensor and the vehicle-mounted state sensor, and constructs an environment map;
the path planning unit is used for planning paths according to the obtained environment information around the luggage van, the obtained obstacle information and the obtained state information of the luggage van, the path planning comprises global path planning and local path planning, and when the luggage van cannot detect the dynamic obstacle, the luggage van drives according to the global path planning; when the dynamic barrier is detected, planning to drive according to the local path; the local path planning comprises the steps of predicting dynamic obstacles according to an environment map and detected dynamic obstacle information, generating a sampling space moving to a plurality of track states in a state space based on the environment map, an obstacle prediction result, a current point and positions of a target point, and generating a plurality of control actions corresponding to the plurality of track states; obtaining an expected reward of each control action in the plurality of control actions based on a machine learning algorithm, scoring the path through an evaluation function, and obtaining a track with the highest score as a local optimal path; and sending the path plan to a control unit and a man-machine interaction module;
the control unit controls the luggage van to transport luggage according to the path planning result, and a user can check path planning information through the display screen and move to a target position along with the luggage van, so that automatic obstacle avoidance control of the luggage van is achieved.
Further, the machine learning algorithm comprises an Actor network and a Critic network, wherein the Actor network is used for determining a control action corresponding to a path state to form a new motion state; the Critic network is used for determining the reward of the control action based on the given path state; the Actor network observes the state according to the current particlesAnd an objectgSelecting an appropriate control actionaObtaining an expected reward by calculating a reward functionrThen, the state is fromsIs transferred tos′Will bes，g，a，r，s′Combined into one tuple X =: (s, g,a,r,s′) And store it in the experience playback pool; the expected reward for each action is accumulated to calculate a merit function,
Figure 83087DEST_PATH_IMAGE001
wherein E is the mathematical expectation,
Figure 991000DEST_PATH_IMAGE002
as a cost factor; iterating according to a Bellman equation until strategy parameters are converged to be optimal; the bellman equation is described as follows:
Figure DEST_PATH_IMAGE003
，
Figure 329578DEST_PATH_IMAGE004
for the observed state of the luggage van at time t,
Figure 828692DEST_PATH_IMAGE005
in state for control strategy
Figure 924824DEST_PATH_IMAGE006
A reward is issued;
Figure 652740DEST_PATH_IMAGE007
is the state transition probability;
Figure 783507DEST_PATH_IMAGE008
to make a state
Figure 453523DEST_PATH_IMAGE009
The strategy that gets the highest reward.
Further, the system further comprises: the sensing unit senses the obstacle information in the environment and controls the speed of the luggage van according to the obstacle information and the distance between the luggage van and the terminal equipment carried by the user.
The invention provides a navigation control method of a luggage van, which is characterized by comprising the following steps:
step 1: obtaining environmental information, obstacle information and state information of the luggage van around the luggage van through a laser radar, a vision sensor and a vehicle-mounted state sensor in a sensing unit, and constructing an environmental map;
step 2: acquiring a target position set by a human-computer interaction unit; the path planning unit plans a path according to the obtained target position information, the environment information, the obstacle information and the state information of the luggage van, wherein the path planning comprises global path planning and local path planning, and when the luggage van cannot detect the dynamic obstacle, the luggage van drives according to the global path planning; when the dynamic barrier is detected, planning to drive according to the local path; the local path planning comprises the steps of predicting a dynamic barrier according to an environment map and detected dynamic barrier information, generating a sampling space moving to a plurality of track states in a state space based on the environment map, a barrier prediction result, a current point and the position of a target point, and generating a plurality of control actions corresponding to the plurality of track states; obtaining an expected reward of each control action in the plurality of control actions based on a machine learning algorithm, scoring the path through an evaluation function, and obtaining a track with the highest score as a local optimal path; and sending the path plan to the control unit and the man-machine interaction module;
and step 3: the control unit controls the luggage van to carry luggage according to the path planning result, and a user can check path planning information through a display screen of the man-machine interaction module and move to a target position along with the luggage van, so that automatic obstacle avoidance control of the luggage van is realized.
Further, the machine learning algorithm comprises an Actor network and a Critic network, wherein the Actor network is used for determining a control action corresponding to a path state to form a new motion state; the Critic network is used for determining the reward of the control action based on the given path state; the Actor network observes the state according to the current particlesAnd an objectgSelecting an appropriate control actionaObtaining an expected reward by calculating a reward functionrThen, the state is fromsIs transferred tos′Will bes，g，a，r，s′The combination is one tuple X =: (s, g,a,r,s′) And store it in the experience playback pool; the expected reward for each action is accumulated to calculate a merit function,
Figure 771371DEST_PATH_IMAGE001
wherein E is the mathematical expectation,
Figure 286666DEST_PATH_IMAGE002
is a cost factor; iterating according to a Bellman equation until strategy parameters are converged to be optimal; the bellman equation is described as follows:
Figure 599836DEST_PATH_IMAGE003
，
Figure 909595DEST_PATH_IMAGE004
for the observed state of the luggage van at time t,
Figure 996630DEST_PATH_IMAGE005
is in state for control strategy
Figure 50037DEST_PATH_IMAGE004
A reward is issued;
Figure 889817DEST_PATH_IMAGE007
is the state transition probability;
Figure 636056DEST_PATH_IMAGE008
to make a state of
Figure 194077DEST_PATH_IMAGE009
The strategy that gets the highest reward.
Preferably, the method comprises: the sensing unit senses the obstacle information in the environment and controls the speed of the luggage van according to the obstacle information and the distance between the luggage van and the terminal equipment carried by the user.
Preferably, the global path plan is generated based on an improved group optimization algorithm, and the specific steps include: step 2.1, rasterizing the map; each grid is in a running state or an obstacle state; initializing a particle population, wherein the particle population comprises a population scale, an initial position, an initial speed and iteration times; generating a plurality of groups of initial path point sets, namely a plurality of particles, according to the starting point and the target point; step 2.2, calculating the particle fitness by using a fitness function; step 2.3, updating the position and the speed of the particles; step 2.4, obtaining an individual optimal value and a global optimal value according to the fitness function; step 2.5 repeating steps 2.2 to 2.4 until the maximum number of iterations is reached; step 2.6, outputting a global optimal solution; step 2.7, taking the output optimal solution as a path point; the path point interpolation is processed using a cubic spline to generate a smooth path.
Further, the fitness function is:
Figure 316753DEST_PATH_IMAGE010
(ii) a WhereinX k ={
Figure 276619DEST_PATH_IMAGE011
,
Figure 928180DEST_PATH_IMAGE012
,…,
Figure 783616DEST_PATH_IMAGE013
}，X k Is as followskParticles, each particle being a path,
Figure 178826DEST_PATH_IMAGE014
is as followskA first particle ofiRoute point，
Figure 258777DEST_PATH_IMAGE015
The constant number is a constant number,
Figure 81240DEST_PATH_IMAGE016
，
Figure 348273DEST_PATH_IMAGE017
in order to be a function of avoiding obstacles,
Figure 78331DEST_PATH_IMAGE018
(ii) a Wherein
Figure 747210DEST_PATH_IMAGE019
The distance of the neighboring waypoint vector to the center of the obstacle,
Figure 819203DEST_PATH_IMAGE020
is shown asjThe radius of the individual obstacles is,
Figure 573532DEST_PATH_IMAGE021
a swelling factor that is an obstacle;
Figure 576123DEST_PATH_IMAGE022
as a function of the distance of the path,
Figure 630667DEST_PATH_IMAGE023
wherein
Figure 794932DEST_PATH_IMAGE024
As the distance from the starting point to the end point,
Figure 302136DEST_PATH_IMAGE025
is the path length;
Figure 842839DEST_PATH_IMAGE026
as a function of the smoothness of the path,
Figure 220731DEST_PATH_IMAGE027
，
Figure 352635DEST_PATH_IMAGE028
，i=1,2,…,nwherein
Figure 815977DEST_PATH_IMAGE029
For angular variation of adjacent paths, in the range 0, pi]；
Figure 239000DEST_PATH_IMAGE030
In order to be a function of the energy consumption,
Figure 471398DEST_PATH_IMAGE031
；
Figure 243045DEST_PATH_IMAGE032
；
Figure 724842DEST_PATH_IMAGE033
(ii) a Wherein the content of the first and second substances,
Figure 607347DEST_PATH_IMAGE034
in order to be a point of the path,
Figure 225410DEST_PATH_IMAGE035
respectively, the coordinates of the path points are represented,
Figure 167958DEST_PATH_IMAGE036
is the terrain height of the waypoint;
Figure DEST_PATH_IMAGE037
>1>
Figure 684521DEST_PATH_IMAGE038
。
the luggage van navigation control system and the control method provided by the invention have the following beneficial effects:
1. the luggage van can maintain a constant distance from the mobile terminal carried by the user. 2. The global optimal path is obtained by adopting an improved group optimization algorithm, the fitness function of the particle swarm optimization algorithm is improved, and the global path which is safer, smoother and more efficient is generated by setting the fitness function and considering the energy consumption, obstacle avoidance capability, path smoothness, path length and path blocking degree of the path. 3. The local path planning adopts an improved machine learning algorithm to realize the obstacle avoidance of the dynamic obstacle and the local path planning, and the state potential energy is introduced by setting a reward and punishment function to realize effective dynamic obstacle avoidance.
Drawings
FIG. 1 is a block diagram of the system of the present invention;
FIG. 2 is a flow chart of the method of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, belong to the protection scope of the present invention.
The first embodiment: as shown in fig. 1, the present embodiment provides a navigation control system for a luggage cart, which is characterized in that the system includes: the system comprises a human-computer interaction unit, a sensing unit, a path planning unit and a control unit;
the man-machine interaction unit comprises a display screen and a voice input module, wherein the display screen is used for displaying an environment map and path planning information; the voice input module is used for voice interaction with a user, so that the starting and stopping of the luggage van are realized, and a voice consultation function is realized;
the sensing unit comprises a laser radar, a vision sensor and a vehicle-mounted state sensor, and obtains environment information, obstacle information and state information of the luggage van around the luggage van through the laser radar, the vision sensor and the vehicle-mounted state sensor, and constructs an environment map;
the path planning unit is used for planning a path according to the obtained environmental information around the luggage van, the obtained obstacle information and the obtained state information of the luggage van, wherein the path planning comprises global path planning and local path planning, and when the luggage van cannot detect a dynamic obstacle, the luggage van is driven according to the global path planning; when the dynamic barrier is detected, planning to drive according to the local path; the local path planning comprises the steps of predicting dynamic obstacles according to an environment map and detected dynamic obstacle information, generating a sampling space moving to a plurality of track states in a state space based on the environment map, an obstacle prediction result, a current point and positions of a target point, and generating a plurality of control actions corresponding to the plurality of track states; obtaining an expected reward of each control action in the plurality of control actions based on a machine learning algorithm, scoring the path through an evaluation function, and obtaining a track with the highest score as a local optimal path; and sending the path plan to the control unit and the man-machine interaction module;
the control unit controls the luggage van to transport luggage according to the path planning result, and a user can check path planning information through the display screen and move to a target position along with the luggage van, so that automatic obstacle avoidance control of the luggage van is achieved.
Preferably, to avoid the algorithm falling into local optimality, a reward of 0.8 times the last state is added to the current reward, resulting in a reward:
Figure 370718DEST_PATH_IMAGE039
；
Figure 108867DEST_PATH_IMAGE040
；
Figure 753475DEST_PATH_IMAGE041
；
Figure 413126DEST_PATH_IMAGE042
；
wherein the content of the first and second substances,
Figure 181974DEST_PATH_IMAGE043
indicating a state when the luggage van reaches the terminal;
Figure 243471DEST_PATH_IMAGE044
indicating the luggage carrier and the obstacleA state in which the distance of the object is less than a set threshold value;
Figure 58981DEST_PATH_IMAGE045
representing that the state potential value of the current moment is less than the state potential value of the last moment;
Figure 471507DEST_PATH_IMAGE046
the potential energy value state which can be larger than the previous moment state at the current moment state is represented;
Figure 499506DEST_PATH_IMAGE047
indicating other states;
Figure 946668DEST_PATH_IMAGE048
representing a distance to the target;
Figure 401920DEST_PATH_IMAGE002
is a penalty factor;
Figure DEST_PATH_IMAGE049
a prize value for the current time;
Figure 567322DEST_PATH_IMAGE004
is the observed state of the vehicle at time t;
Figure 133433DEST_PATH_IMAGE050
is in a state
Figure DEST_PATH_IMAGE051
The control strategy of (2);
Figure 513730DEST_PATH_IMAGE052
the potential energy value of the state at the current moment represents the potential energy between the current moment and the target point; the potential energy value of the current moment state is smaller than that of the previous moment state, so that the luggage van arrives at the position close to the target from the position far away from the target, and a reward value is added to the current point; the potential energy value which can be more than the previous time state in the current time state indicates that the luggage van reaches the position far away from the target from the position close to the target, the reward value is reduced for the current point,and state potential energy is introduced, when the luggage van approaches a target point, a certain reward can be obtained, otherwise, the reward is reduced, and therefore a better convergence effect is achieved.
Preferably, the machine learning algorithm comprises an Actor network and a Critic network, wherein the Actor network is used for determining a control action corresponding to a path state to form a new motion state; the Critic network is used for determining the reward of the control action based on the given path state; the Actor network observes the state according to the current particlesAnd an objectgSelecting an appropriate control actionaObtaining an expected reward by calculating a reward functionrThen, the state is fromsIs transferred tos′Will bes，g，a，r，s′Combined into one tuple X =: (s,g, a,r,s′) And store it in the experience playback pool; the expected reward for each action is accumulated to calculate a merit function,
Figure 405462DEST_PATH_IMAGE053
wherein E is the mathematical expectation,
Figure 527002DEST_PATH_IMAGE002
as a cost factor; iterating according to a Bellman equation until strategy parameters are converged to be optimal; the bellman equation is described as follows:
Figure 896804DEST_PATH_IMAGE003
，
Figure 584137DEST_PATH_IMAGE004
for the observed state of the luggage van at time t,
Figure 381192DEST_PATH_IMAGE054
is in state for control strategy
Figure 68656DEST_PATH_IMAGE004
A reward is issued;
Figure 976569DEST_PATH_IMAGE055
is the state transition probability;
Figure 987251DEST_PATH_IMAGE056
to make a state
Figure 486365DEST_PATH_IMAGE057
The strategy that gets the highest reward.
Preferably, the system comprises: the sensing unit senses the obstacle information in the environment and controls the speed of the luggage van according to the obstacle information and the distance between the luggage van and the terminal equipment carried by the user.
The second embodiment: as shown in fig. 2, an embodiment of the present invention provides an automatic navigation control method for a baggage car, including the following steps:
step 1: obtaining environmental information, obstacle information and state information of the luggage van around the luggage van through a laser radar, a vision sensor and a vehicle-mounted state sensor in a sensing unit, and constructing an environmental map;
step 2: acquiring a target position set by a human-computer interaction unit; the path planning unit plans a path according to the obtained target position information, the environment information, the obstacle information and the state information of the luggage van, wherein the path planning comprises global path planning and local path planning, and when the luggage van cannot detect the dynamic obstacle, the luggage van drives according to the global path planning; when the dynamic barrier is detected, planning to drive according to the local path; the local path planning comprises the steps of predicting a dynamic barrier according to an environment map and detected dynamic barrier information, generating a sampling space moving to a plurality of track states in a state space based on the environment map, a barrier prediction result, a current point and the position of a target point, and generating a plurality of control actions corresponding to the plurality of track states; obtaining an expected reward of each control action in the plurality of control actions based on a machine learning algorithm, scoring the path through an evaluation function, and obtaining a track with the highest score as a local optimal path; and sending the path plan to the control unit and the man-machine interaction module;
and 3, step 3: the control unit controls the luggage van to carry luggage according to the path planning result, and a user can check path planning information through a display screen of the man-machine interaction module and move to a target position along with the luggage van, so that automatic obstacle avoidance control of the luggage van is realized.
Preferably, the global path planning in step 2 is generated based on an improved group optimization algorithm, and the specific steps include: step 2.1, rasterizing the map; each grid is in a running state or an obstacle state; initializing a particle population, wherein the particle population comprises a population scale, an initial position, an initial speed and iteration times Maxn; from the starting and target points, sets of initial path points, i.e., particles, are generatedX k ，k=1,2, \ 8230, maxn, one path per particle comprisingnA path point; step 2.2, calculating the particle fitness by using a fitness function; step 2.3, updating the position and the speed of the particles; step 2.4, obtaining an individual optimal value and a global optimal value according to the fitness function; step 2.5 repeating steps 2.2 to 2.4 until the maximum number of iterations is reached; step 2.6, outputting a global optimal solution; step 2.7, taking the output optimal solution as a path point; the path point interpolation is processed using cubic splines to generate a smooth path.
Further, the fitness function is:
Figure 582497DEST_PATH_IMAGE058
(ii) a WhereinX k ={
Figure 559680DEST_PATH_IMAGE059
,
Figure 424868DEST_PATH_IMAGE060
,…,
Figure 563725DEST_PATH_IMAGE061
}，X k Is as followskParticles, each particle being a path representing a set of discrete path points from a start point to an end point,
Figure 429044DEST_PATH_IMAGE062
is as followskA first particle ofiThe point of the path is the point of the path,
Figure 944339DEST_PATH_IMAGE063
the constant number is a constant number,
Figure 664034DEST_PATH_IMAGE064
，
Figure 301688DEST_PATH_IMAGE017
in order to be a function of avoiding obstacles,
Figure 372413DEST_PATH_IMAGE065
(ii) a Wherein
Figure 956978DEST_PATH_IMAGE019
The distance of the neighboring waypoint vector to the center of the obstacle,
Figure 796758DEST_PATH_IMAGE066
，
Figure 277418DEST_PATH_IMAGE020
denotes the firstjThe radius of the individual obstacles is,
Figure 848820DEST_PATH_IMAGE067
a swelling factor that is an obstacle; when in use
Figure 440338DEST_PATH_IMAGE068
When the number is 1, the path is safe, and the obstacle can be avoided; when in use
Figure 462521DEST_PATH_IMAGE068
A value of 0 indicates the presence of an obstacle.
Figure 379661DEST_PATH_IMAGE022
As a function of the distance of the path,
Figure 159398DEST_PATH_IMAGE069
wherein
Figure 85766DEST_PATH_IMAGE024
As is the distance from the starting point to the end point,
Figure 900138DEST_PATH_IMAGE025
is the path length; the shorter the length of the path is the shorter,
Figure 738912DEST_PATH_IMAGE070
the larger.
Figure 5946DEST_PATH_IMAGE026
As a function of the smoothness of the path,
Figure 204846DEST_PATH_IMAGE027
，
Figure 404883DEST_PATH_IMAGE028
，i=1,2,…,nwherein
Figure 398247DEST_PATH_IMAGE071
，
Figure 480472DEST_PATH_IMAGE072
，
Figure 217484DEST_PATH_IMAGE029
For angular variation of adjacent paths, in the range [0, π]The smaller the angular variation of the adjacent paths,
Figure 740869DEST_PATH_IMAGE073
the larger the value of (c), the smoother the path.
Figure 452605DEST_PATH_IMAGE074
In order to be a function of the energy consumption,
Figure 694230DEST_PATH_IMAGE031
；
Figure 234933DEST_PATH_IMAGE032
；
Figure 878404DEST_PATH_IMAGE033
(ii) a Wherein, the first and the second end of the pipe are connected with each other,
Figure 213570DEST_PATH_IMAGE034
in order to be a point of the path,
Figure 270388DEST_PATH_IMAGE035
respectively, the coordinates of the path points are represented,
Figure 349202DEST_PATH_IMAGE036
is the terrain height of the waypoint;
Figure 847180DEST_PATH_IMAGE037
>1>
Figure 900718DEST_PATH_IMAGE038
(ii) a Wherein
Figure 851356DEST_PATH_IMAGE075
Is the energy consumption coefficient when climbing an uphill,
Figure 999441DEST_PATH_IMAGE076
the energy consumption coefficient when the slope is downhill; coefficient of through energy consumption
Figure 351925DEST_PATH_IMAGE077
Distinguishing different energy consumption of the luggage van when going up a slope and when going down the slope; when the luggage van goes up a slope, the energy consumption is increased, and when the luggage van goes down a slope, the energy consumption is reduced.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and should not be taken as limiting the scope of the present invention, which is intended to cover any modifications, equivalents, improvements, etc. within the spirit and scope of the present invention.

Claims (6)

1. A navigation control system for a luggage cart, the system comprising: the system comprises a human-computer interaction unit, a sensing unit, a path planning unit and a control unit;
the man-machine interaction unit comprises a display screen and a voice input module, wherein the display screen is used for displaying an environment map and path planning information; the voice input module is used for voice interaction with a user, so that the starting and stopping of the luggage van are realized, and a voice consultation function is realized;
the sensing unit comprises a laser radar, a vision sensor and a vehicle-mounted state sensor, and is used for acquiring environmental information, obstacle information and state information of the luggage van around the luggage van through the laser radar, the vision sensor and the vehicle-mounted state sensor and constructing an environmental map;
the path planning unit is used for planning a path according to the obtained environmental information around the luggage van, the obtained obstacle information and the obtained state information of the luggage van, wherein the path planning comprises global path planning and local path planning, and when the luggage van cannot detect a dynamic obstacle, the luggage van is driven according to the global path planning; when the dynamic barrier is detected, planning to drive according to the local path; the local path planning comprises the steps of predicting a dynamic barrier according to an environment map and detected dynamic barrier information, generating a sampling space moving to a plurality of track states in a state space based on the environment map, a barrier prediction result, a current point and the position of a target point, and generating a plurality of control actions corresponding to the plurality of track states; obtaining an expected reward of each control action in the plurality of control actions based on a machine learning algorithm, scoring the path through an evaluation function, and obtaining a track with the highest score as a local optimal path; and sending the path plan to a control unit and a man-machine interaction unit;
the control unit controls the luggage van to carry luggage according to the path planning result, and a user can check path planning information through the display screen and move to a target position along with the luggage van, so that automatic obstacle avoidance control of the luggage van is realized;
the path planning unit generates a global path plan based on an improved group optimization algorithm, and the specific steps include: 1) Rasterizing a map; each grid is in a drivable state or an obstacle state; initializing a particle population, wherein the particle population comprises a population scale, an initial position, an initial speed and iteration times; generating a plurality of groups of initial path point sets, namely a plurality of particles, according to the starting point and the target point; 2) Calculating particle fitness by using a fitness function; 3) Updating the particle position and velocity; 4) Obtaining an individual optimal value and a global optimal value according to the fitness function; 5) Repeating the steps 2 to 4 until the maximum iteration number is reached; 6) Outputting a global optimal solution; 7) Taking the output optimal solution as a path point; processing the path point interpolation using a cubic spline to generate a smooth path;
the fitness function is:
Figure DEST_PATH_IMAGE002
(ii) a WhereinX k ={
Figure DEST_PATH_IMAGE004
,
Figure DEST_PATH_IMAGE006
,…,
Figure DEST_PATH_IMAGE008
}，X k Is as followskParticles, each particle being a path,
Figure DEST_PATH_IMAGE010
is as followskA first particle ofiThe point of the path is the point of the path,
Figure DEST_PATH_IMAGE012
the constant number is a constant number,
Figure DEST_PATH_IMAGE014
，
Figure DEST_PATH_IMAGE016
in order to be a function of avoiding obstacles,
Figure DEST_PATH_IMAGE018
(ii) a Wherein
Figure DEST_PATH_IMAGE020
The distance of the neighboring waypoint vector to the center of the obstacle,
Figure DEST_PATH_IMAGE022
is shown asjThe radius of the individual obstacles,
Figure DEST_PATH_IMAGE024
a swelling factor that is an obstacle;
Figure DEST_PATH_IMAGE026
as a function of the distance of the path,
Figure DEST_PATH_IMAGE028
wherein
Figure DEST_PATH_IMAGE032
As is the distance from the starting point to the end point,
Figure DEST_PATH_IMAGE034
is the path length;
Figure DEST_PATH_IMAGE036
as a function of the smoothness of the path,
Figure DEST_PATH_IMAGE038
，
Figure DEST_PATH_IMAGE040
，i=1,2,…,nwherein
Figure DEST_PATH_IMAGE042
For angular variation of adjacent paths, in the range 0, pi]；
Figure DEST_PATH_IMAGE044
In order to be a function of the energy consumption,
Figure DEST_PATH_IMAGE046
；
Figure DEST_PATH_IMAGE048
；
Figure DEST_PATH_IMAGE050
(ii) a Wherein
Figure DEST_PATH_IMAGE052
In order to have a high energy consumption coefficient,
Figure DEST_PATH_IMAGE054
in order to be a point of the path,
Figure DEST_PATH_IMAGE056
the coordinates of the path points are respectively represented,
Figure DEST_PATH_IMAGE058
is the terrain height of the waypoint;
Figure DEST_PATH_IMAGE060
>1>
Figure DEST_PATH_IMAGE062
(ii) a Wherein
Figure DEST_PATH_IMAGE064
Is the energy consumption coefficient when climbing an uphill,
Figure DEST_PATH_IMAGE066
the energy consumption coefficient when the slope is downhill;
the reward is:
Figure DEST_PATH_IMAGE068
；
Figure DEST_PATH_IMAGE070
；
Figure DEST_PATH_IMAGE072
；
Figure DEST_PATH_IMAGE074
；
wherein the content of the first and second substances,
Figure DEST_PATH_IMAGE076
indicating a state when the luggage van reaches the terminal;
Figure DEST_PATH_IMAGE078
indicating a state when the distance between the luggage van and the obstacle is less than a set threshold value;
Figure DEST_PATH_IMAGE080
representing that the state potential value of the current moment is less than the state potential value of the last moment;
Figure DEST_PATH_IMAGE082
the potential energy value state which can be larger than the previous time state at the current time state is represented;
Figure DEST_PATH_IMAGE084
represents other states;
Figure DEST_PATH_IMAGE086
representing a distance to the target;
Figure DEST_PATH_IMAGE088
is a penalty factor;
Figure DEST_PATH_IMAGE090
a prize value for the current time;
Figure DEST_PATH_IMAGE092
is the observed state of the vehicle at time t;
Figure DEST_PATH_IMAGE094
is in a state
Figure DEST_PATH_IMAGE096
The control strategy of (2);
Figure DEST_PATH_IMAGE098
the potential energy value of the state at the current moment represents the potential energy between the current moment and the target point.
2. The control system of claim 1, wherein the machine learning algorithm comprises an Actor network and a Critic network, the Actor network is used for determining a control action corresponding to a path state to form a new motion state; the Critic network is used for determining the reward of the control action based on the given path state; the Actor network observes the state according to the current particlesAnd an objectgSelecting an appropriate control actionaObtaining an expected reward by computing a reward functionrThen, the state is fromsIs transferred tos′Will bes，g，a，r，s′The combination is one tuple X =: (s,g,a,r,s′) And store it in the experience playback pool; the expected reward for each action is accumulated to calculate a merit function,
Figure DEST_PATH_IMAGE100
in whichEIn order to be the mathematical expectation,
Figure DEST_PATH_IMAGE102
in order to be a cost factor, the cost factor,
Figure DEST_PATH_IMAGE104
is in a state
Figure 861645DEST_PATH_IMAGE092
The control strategy of (2); iterating according to a Bellman equation until strategy parameters converge to be optimal; the bellman equation is described as follows:
Figure DEST_PATH_IMAGE106
，
Figure 888374DEST_PATH_IMAGE092
as an observation of the baggage car at time tIn the state of the mobile communication device,
Figure DEST_PATH_IMAGE108
is in state for control strategy
Figure 730428DEST_PATH_IMAGE092
A reward is issued;
Figure DEST_PATH_IMAGE110
is the state transition probability;
Figure DEST_PATH_IMAGE112
to make a state
Figure DEST_PATH_IMAGE114
The strategy that gets the highest reward.
3. The control system of claim 1, wherein the system comprises: the sensing unit senses the obstacle information in the environment and controls the speed of the luggage van according to the obstacle information and the distance between the luggage van and the terminal equipment carried by the user.
4. A control method of a navigation control system of a luggage van according to any one of claims 1 to 3, characterized in that the method comprises the steps of:
step 1: obtaining environmental information, obstacle information and state information of the luggage van around the luggage van through a laser radar, a vision sensor and a vehicle-mounted state sensor in a sensing unit, and constructing an environmental map;
step 2: acquiring a target position set by a human-computer interaction unit; the path planning unit plans a path according to the obtained target position information, the environment information, the obstacle information and the state information of the luggage van, wherein the path planning comprises global path planning and local path planning, and when the luggage van cannot detect the dynamic obstacle, the luggage van drives according to the global path planning; when the dynamic barrier is detected, planning to drive according to the local path; the local path planning comprises the steps of predicting a dynamic barrier according to an environment map and detected dynamic barrier information, generating a sampling space moving to a plurality of track states in a state space based on the environment map, a barrier prediction result, a current point and the position of a target point, and generating a plurality of control actions corresponding to the plurality of track states; obtaining an expected reward of each control action in the plurality of control actions based on a machine learning algorithm, scoring the path through an evaluation function, and obtaining a track with the highest score as a local optimal path; and sending the path plan to a control unit and a man-machine interaction unit;
and 3, step 3: the control unit controls the luggage van to carry luggage according to the path planning result, and a user can check path planning information through a display screen of the man-machine interaction unit and move to a target position along with the luggage van, so that automatic obstacle avoidance control of the luggage van is realized.
5. The control method according to claim 4, wherein the machine learning algorithm comprises an Actor network and a Critic network, and the Actor network is used for determining the control action corresponding to the path state to form a new motion state; the Critic network is used for determining the reward of the control action based on the given path state; the Actor network observes the state according to the current particlesAnd objectsgSelecting an appropriate control actionaObtaining an expected reward by calculating a reward functionrThen, the state is fromsIs transferred tos′Will bes，g，a，r，s′Combined into one tuple X =: (s,g,a,r,s′) And store it in the experience playback pool; the expected reward for each action is accumulated to compute a merit function,
Figure 548474DEST_PATH_IMAGE100
whereinEIn order to be the mathematical expectation,
Figure 628425DEST_PATH_IMAGE102
in order to be a cost factor, the cost factor,
Figure 982046DEST_PATH_IMAGE104
is a state
Figure 576975DEST_PATH_IMAGE092
The control strategy of (2); iterating according to a Bellman equation until strategy parameters are converged to be optimal; the bellman equation is described as follows:
Figure DEST_PATH_IMAGE116
，
Figure 41455DEST_PATH_IMAGE092
for the observed state of the luggage van at time t,
Figure 241492DEST_PATH_IMAGE108
in state for control strategy
Figure 264549DEST_PATH_IMAGE092
A reward is issued;
Figure 550037DEST_PATH_IMAGE110
is the state transition probability;
Figure 818208DEST_PATH_IMAGE112
to make a state
Figure 872751DEST_PATH_IMAGE114
The strategy that gets the highest prize.
6. The control method according to claim 4, characterized in that the method comprises: the sensing unit senses the obstacle information in the environment and controls the speed of the luggage van according to the obstacle information and the distance between the luggage van and the terminal equipment carried by the user.
CN202211170194.0A 2022-09-26 2022-09-26 Navigation control system and control method of luggage van Active CN115268467B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211170194.0A CN115268467B (en) 2022-09-26 2022-09-26 Navigation control system and control method of luggage van

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211170194.0A CN115268467B (en) 2022-09-26 2022-09-26 Navigation control system and control method of luggage van

Publications (2)

Publication Number Publication Date
CN115268467A CN115268467A (en) 2022-11-01
CN115268467B true CN115268467B (en) 2023-01-10

Family

ID=83756771

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211170194.0A Active CN115268467B (en) 2022-09-26 2022-09-26 Navigation control system and control method of luggage van

Country Status (1)

Country Link
CN (1) CN115268467B (en)

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109703607A (en) * 2017-10-25 2019-05-03 北京眸视科技有限公司 A kind of Intelligent baggage car
CN111780777A (en) * 2020-07-13 2020-10-16 江苏中科智能制造研究院有限公司 Unmanned vehicle route planning method based on improved A-star algorithm and deep reinforcement learning
CN114397896A (en) * 2022-01-10 2022-04-26 贵州大学 Dynamic path planning method for improving particle swarm optimization

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11530921B2 (en) * 2018-09-28 2022-12-20 Intel Corporation Method of generating a collision free path of travel and computing system

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109703607A (en) * 2017-10-25 2019-05-03 北京眸视科技有限公司 A kind of Intelligent baggage car
CN111780777A (en) * 2020-07-13 2020-10-16 江苏中科智能制造研究院有限公司 Unmanned vehicle route planning method based on improved A-star algorithm and deep reinforcement learning
CN114397896A (en) * 2022-01-10 2022-04-26 贵州大学 Dynamic path planning method for improving particle swarm optimization

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
"Using Particle Swarm Optimization for Robot Path Planning in Dynamic Environments with Moving Obstacles and Target";Amin Zargar Nasrollahy 等;《2009 Third UKSim European Symposium on Computer Modeling and Simulation》;20091231;全文 *
"一种动态环境下移动机器人的路径规划方法";朴松昊 等;《机器人》;20030131;第25卷(第1期);全文 *
"具有修正策略的改进NSGA－II三维路径规划";封建湖 等;《机械设计与制造》;20210531;全文 *

Also Published As

Publication number Publication date
CN115268467A (en) 2022-11-01

Similar Documents

Publication Publication Date Title
CN112677995B (en) Vehicle track planning method and device, storage medium and equipment
Lin et al. Anti-jerk on-ramp merging using deep reinforcement learning
CN109496288A (en) System and method for determining track
US11618444B2 (en) Methods and systems for autonomous vehicle inference of routes for actors exhibiting unrecognized behavior
CN115344052B (en) Vehicle path control method and control system based on improved group optimization algorithm
Shimosaka et al. Modeling risk anticipation and defensive driving on residential roads with inverse reinforcement learning
US11945441B2 (en) Explainability and interface design for lane-level route planner
US20220306156A1 (en) Route Planner and Decision-Making for Exploration of New Roads to Improve Map
CN113835421A (en) Method and device for training driving behavior decision model
EP4160146A1 (en) Quadtree based data structure for storing information relating to an environment of an autonomous vehicle and methods of use thereof
Bertolazzi et al. Efficient re-planning for robotic cars
CN109214827A (en) A kind of method and apparatus that identification vehicle brush list deceives benefit behavior
WO2022088761A1 (en) Method and apparatus for planning vehicle driving path, intelligent vehicle, and storage medium
CN115268467B (en) Navigation control system and control method of luggage van
CN113743767B (en) Vehicle dispatching method, system, computer and medium based on time and safety
CN113928341B (en) Road decision method, system, equipment and medium
US11904906B2 (en) Systems and methods for prediction of a jaywalker trajectory through an intersection
Smit et al. Informed sampling-based trajectory planner for automated driving in dynamic urban environments
CN115600734A (en) Data processing method and device and electronic equipment
CN111596668B (en) Mobile robot anthropomorphic path planning method based on reverse reinforcement learning
CN113911136A (en) Driverless vehicle lane change decision-making method, driverless vehicle lane change decision-making system, driverless vehicle lane change decision-making equipment and driverless vehicle lane change decision-making medium
Eroğlu et al. Energy Efficient Driving Optimization of Electrical Vehicles Considering the Road Characteristics
Schörner et al. Towards Multi-Modal Risk Assessment
CN109540166A (en) A kind of Safe path planning method based on Gaussian process
Cheng et al. ETC Intelligent Navigation Path Planning Method

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: 20221101

Assignee: Jiangsu Tianyi Airport Equipment Maintenance Service Co.,Ltd.

Assignor: Jiangsu Tianyi Aviation Industry Co.,Ltd.

Contract record no.: X2023980044219

Denomination of invention: A Navigation Control System and Control Method for Luggage Carts

Granted publication date: 20230110

License type: Common License

Record date: 20231024