CN108357594B

CN108357594B - Self-balancing unmanned bicycle based on intelligent evolution and competition and cooperation control method thereof

Info

Publication number: CN108357594B
Application number: CN201810081108.6A
Authority: CN
Inventors: 孟濬; 赵夕朦
Original assignee: Zhejiang University ZJU
Current assignee: Zhejiang University ZJU
Priority date: 2018-01-26
Filing date: 2018-01-26
Publication date: 2020-02-07
Anticipated expiration: 2038-01-26
Also published as: CN108357594A

Abstract

The invention discloses a self-balancing unmanned bicycle based on intelligent evolution and a competition and cooperation control method thereof. The control method comprises a self-balancing control part and an unmanned control part. The self-balancing control screens control strategies through whether the actual bicycle state change is the same as the expectation or not while a person rides a bicycle, and finally self-balancing of the bicycle is achieved based on real-time learning of competition and cooperation. The unmanned bicycle can have a self-balancing function in various motion states through a coupling control method, and meanwhile, an indirect driving method is adopted, so that the self-balancing and unmanned driving of the bicycle can be realized only by installing three controller modules on the common bicycle without further modification of the common bicycle.

Description

Self-balancing unmanned bicycle based on intelligent evolution and competition and cooperation control method thereof

Technical Field

The invention relates to the field of traffic, in particular to a self-balancing unmanned bicycle based on intelligent evolution and a competition and cooperation control method thereof.

Background

As a traditional vehicle, the bicycle has the advantages of narrow and small body, simple mechanism, small-radius rotation, convenience, flexibility, no pollution, no noise, no energy source, low selling price and the like, and plays a significant role in modern life with increasingly serious problems of road congestion, air pollution, oil price rise and the like. The unmanned bicycle can provide driving balance assistance for special people such as children and the old, and is expected to be widely applied to disaster rescue and forest operation.

As people's attention to intelligent vehicles and unmanned technologies continues to increase, unmanned bicycles or bicycle robots have been developed primarily based on this intelligent vehicle concept. At present, most researchers of unpiloted bicycles are around both aspects of dynamic modeling and new control algorithm, and the research on the unpiloted bicycles mostly stays in the stages of theoretical discussion and preliminary experiments. Due to the complex dynamic characteristics and certain lateral instability of the bicycle, the self-balancing of the bicycle still has many troublesome problems, and how to solve the self-balancing problem of the bicycle running at a static or low speed is the key point for the unmanned bicycle to break through the current development limitation.

The existing balance system applied to the motorcycle or the electric bicycle is essentially the superposition of a monocycle balance system (namely, an inverted pendulum balance system) and a two-foot balance system. The front handle of the bicycle has high degree of freedom, and the two wheels have no direct driving force. Therefore, the driving force on a motorcycle or an electric bicycle that causes the balance thereof is not present on the bicycle, and the balancing method thereof is not effective on the bicycle, which brings more difficulty to the self-balancing and unmanned driving of the bicycle.

Meanwhile, although some related self-evolution simulation researches exist at present, no related research for self-evolution based on hardware exists, and the self-evolution simulation researches are not applied to the self-balancing problem of the bicycle.

Disclosure of Invention

The invention aims to provide a self-balancing unmanned bicycle based on intelligent evolution and a competition and cooperation control method thereof aiming at the defects of the prior art.

The purpose of the invention is realized by the following technical scheme: a competition and cooperation control method of a self-balancing unmanned bicycle based on intelligent evolution comprises a balance control part and an unmanned control part; the handlebar control module, the middle part control module and the rear part control module control each mechanism of the modules according to the information provided by the sensor module, thereby carrying out the indirect control of the balance and the advancing of the bicycle; the control variables of the mechanisms of the handlebar control module, the middle part control module of the vehicle body and the rear part control module of the vehicle body are coupled with each other;

the implementation method of the balance control part comprises the following steps:

1) selecting key variables: selecting controllable considerable key variables, including bicycle variables and control variables of a handlebar control module, a vehicle body middle control module and a vehicle body rear control module;

2) real-time learning and controller establishment based on competition and cooperation: when a user rides a bicycle, the bicycle is randomly controlled by a controller, the change direction of each variable of the bicycle, which is caused by the variable change of the controller, is predicted without considering the bicycle control of a person, if the actual control result of the bicycle is opposite to the prediction, the control method is optimized, if the actual control result of the bicycle is the same as the prediction, the control method is reserved, the corresponding relation between the current bicycle variable (namely the current bicycle state) and the change of the control variable is established, and the parameters of the controller are learned in real time under various bicycle states based on the principle of competition and cooperation;

3) self-balancing is realized: the controller parameters obtained by real-time learning are respectively input into an actual handlebar control module, a vehicle body middle control module and a vehicle body rear control module, fine adjustment is carried out, and three controllers of the bicycle are built, so that self-balancing of the bicycle is realized;

the unmanned control part comprises the following implementation methods: and selecting a desired bicycle variable according to the target motion state to realize the unmanned control of the bicycle.

Furthermore, the control method can be divided into various road conditions and terrain conditions to converge to respective control schemes;

further, the control method can be applied to habit correction, after healthy riding habits of athletes or coaches are learned, the bicycle variable control method tends to be healthy through the superposition effect of the handlebar control module, the middle body control module, the rear body control module and the user on bicycle control, when the riding habits of the user are not good, the handlebar control module, the middle body control module and the rear body control module generate extra disturbance for a healthy riding mode, the user feels hard, and the healthy riding habits tend to be used for riding.

Further, the implementation of the unmanned control portion includes: selecting a target motion state, and controlling the bicycle in the target motion state; the motion state includes: starting, advancing, turning and retreating;

the bicycle control under the starting state comprises the following specific steps:

1) the integral adjustment is that the bicycle handlebar deflection angle α tends to be a constant through the variable adjustment of the handlebar control module, the vehicle body middle control module and the vehicle body rear control module, even if the bicycle tends to be an integral body from a running vehicle;

2) the center of gravity is adjusted through the variable adjustment of the handlebar control module, the middle part control module of the bicycle body and the rear part control module of the bicycle body, so that the bicycle handlebar deflection angle α tends to 0, and the bicycle body deflection angle β tends to 0, even if the bicycle is in a vertical standing state from a certain deflection angle;

the bicycle control under the advancing state comprises the following specific steps:

1) the integral adjustment, namely the bicycle handlebar deflection angle α tends to 0 through the variable adjustment of the handlebar control module, the middle part control module of the bicycle body and the rear part control module of the bicycle body, even if the bicycle is obtained and tends to an integral body when the handlebar does not rotate;

2) the center of gravity is adjusted through the variable adjustment of the handlebar control module, the middle part control module of the bicycle body and the rear part control module of the bicycle body, so that the deflection angle β of the bicycle body tends to 0, even if the bicycle is balanced by the bicycle;

3) indirect drive: the rear wheel of the bicycle is indirectly driven through the variable adjustment of the rotating wheel mechanism of the control module at the rear part of the bicycle body, so that the rotating angle of the rear wheel of the bicycle is adjusted

Varying at a certain angular speed, even if the bicycle is moving forward at a certain speed;

the bicycle control under the turning state comprises the following specific steps:

1) the integral adjustment, namely the bicycle handlebar deflection angle α tends to turn direction through the variable adjustment of the handlebar control module, the middle part control module of the bicycle body and the rear part control module of the bicycle body, even if the bicycle tends to be an integral body when the handlebar rotates;

Change at a certain angular velocity even if the vehicle is turning at a certain velocity;

the bicycle control method in the backward state comprises the following specific steps:

1) indirect drive: the rear wheel of the bicycle is indirectly driven through the variable adjustment of the rotating wheel mechanism of the control module at the rear part of the bicycle body, so that the rotating angle of the rear wheel of the bicycle is adjusted

Because there is a fore-and-aft relation in the place where handlebar contacts the ground with front wheel, in the backward state of bicycle, handlebar and front wheel are in the state of being dragged, the drag force in the handlebar junction is in the front, has dispelled the handlebar rotation torque produced while the bicycle advances, can simplify the regulation of the bicycle handlebar deflection angle α, the bicycle will tend to a whole in the backward state;

2) and (4) adjusting the center of gravity of the bicycle by variable adjustment of the handlebar control module, the middle part control module of the bicycle body and the rear part control module of the bicycle body, so that the deflection angle β of the bicycle body of the bicycle tends to be 0 even if the bicycle is balanced by the bicycle.

Further, the selecting the target motion state specifically includes:

1) macroscopic route determination: determining the integral traveling route of the bicycle in modes of navigation, manual selection and the like;

2) road surface control and obstacle avoidance: monitoring the road surface through a sensor module; carrying out terrain scanning, judging the terrain and selecting a control method corresponding to the terrain; and judging whether an obstacle exists or not, and if so, avoiding the obstacle, namely obtaining the traveling direction of the bicycle to be adjusted according to road surface information such as distance, obstacle width, obstacle motion condition and the like so as to adjust.

The invention has the beneficial effects that:

(1) the unmanned bicycle has a self-balancing function.

(2) The self-balancing bicycle has a self-balancing function under various motion conditions.

(3) The unmanned bicycle controls a multivariable coupling system through a coupling control method, and control variables of the three controller modules are coupled with each other, so that the unmanned bicycle becomes a self-balancing whole.

(4) The unmanned bicycle adopts an indirect driving method, can realize the unmanned driving of the common bicycle only by installing the three controller modules on the common bicycle, and does not need to further modify the common bicycle.

(5) The invention discloses a self-balancing control method for competition and cooperation, which is used for screening control strategies by judging whether the actual bicycle state change is the same as the expectation or not while a person rides a bicycle, and provides a new idea for realizing the self-balancing unmanned bicycle.

(6) The unmanned bicycle of the present invention can also be applied to habit correction for learning healthy riding habits of athletes or coaches, and correcting riding habits of ordinary people.

Drawings

FIG. 1 is an overall block diagram of the unmanned bicycle of the present invention;

FIG. 2 is a top plan view of the drone bicycle of this invention;

FIG. 3 is a rear elevational view of the unmanned bicycle of the present invention;

FIG. 4 is a block diagram of the steps of the contention and cooperation control method of the present invention;

FIG. 5 is a block diagram of the steps of the self-evolving control method of the present invention;

FIG. 6 is a block diagram of the steps of the control method of the present invention for adaptive evolution of environmental evolution;

FIG. 7 is a flow chart of the steps and learning objectives of the competition and cooperation control method of the present invention;

FIG. 8 is a flow chart illustrating the control rule obtaining through data according to the present invention;

FIG. 9 is a flow chart of the self-evolution specific steps in the self-evolution control method of the present invention;

FIG. 10 is a flowchart illustrating the steps of selecting an environment evolution adaptive evolution control scheme based on terrain scanning according to the present invention;

fig. 11 is a rear wheel drive schematic view of the unmanned bicycle of the present invention.

Detailed Description

In order to explain the present invention in more detail, an unmanned bicycle with a self-balancing function will be described in detail with reference to the accompanying drawings.

As shown in fig. 1, the present invention provides an unmanned bicycle with self-balancing function, which comprises a bicycle, a sensor module, a handlebar control module, a middle control module of a bicycle body, and a rear control module of the bicycle body.

The bicycle is a common bicycle on the market and comprises a front wheel (mass m1, radius r), a rear wheel (mass m1, radius r) and a frame (mass m 2).

The sensor module is used for measuring a bicycle handlebar deflection angle α, a bicycle body deflection angle β and a bicycle rear wheel rotation angle phi as shown in fig. 2, the bicycle handlebar deflection angle α is an included angle between a bicycle front wheel and a bicycle body, the bicycle handlebar deflection angle α is a positive number, indicates that a bicycle handlebar deflects rightward, and indicates that a bicycle handlebar deflects leftward as a negative number, the bicycle body deflection angle β is an included angle between a bicycle body and a vertical plane, the bicycle body deflection angle β is a positive number, indicates that a bicycle body tilts rightward, and indicates that a bicycle body tilts leftward as a negative number, the bicycle rear wheel rotation angle phi is a rotation angle of a bicycle rear wheel along a rear wheel axis, the bicycle rear wheel rotation angle phi is a positive number, indicates that a bicycle rear wheel rotates forward, and the bicycle rear wheel rotates backward, the bicycle rear wheel rotation angle phi is a negative number, the sensor module can be installed at a handlebar control module, or can be installed at a handlebar control module, a vehicle body deflection angle phi and a vehicle body deflection angle F2, a left-right-left-right pressure sensor array F84, a pressure sensor F3, a pressure sensor array F, a left-right pressure sensor F3, a pressure sensor F7, a pressure sensor array F3, a pressure sensor array F, a pressure sensor array F3, a pressure sensor array F, a pressure sensor;

handlebar control module be located the bicycle handlebar, including transversely placing in the electric slide bar mechanism on the handlebar (slider counter weight m 3.) handlebar control module carry out the regulation of handlebar focus and handlebar deflection angle α through adjusting handlebar slider position x be the distance at slider and handlebar center, handlebar slider position x indicate the slider to be located handlebar center right side when being the positive number, indicate the slider to be located handlebar center left side when being the negative number.

The bicycle body middle control module is positioned on a bicycle body and comprises a bicycle body electric eccentric wheel mechanism (radius r1, a bicycle body eccentric wheel counterweight m 4). The vehicle body middle control module adjusts the gravity center of the vehicle body by adjusting the rotation angle theta 1 of the vehicle body eccentric wheel. When the rotation angle theta 1 of the eccentric wheel of the vehicle body is positive, the eccentric wheel counterweight is positioned on the right side of the vehicle body, and when the rotation angle theta 1 of the eccentric wheel of the vehicle body is negative, the eccentric wheel counterweight is positioned on the left side of the vehicle body.

The control module at the rear part of the bicycle body is positioned above a rear wheel of the bicycle and comprises a rear seat electric eccentric wheel mechanism (radius r2, rear seat eccentric wheel counterweight m5) and an electric rotating wheel mechanism m 6. The backseat electric eccentric wheel mechanism adjusts the gravity center of the backseat by adjusting the rotation angle theta 2 of the backseat eccentric wheel, when the rotation angle theta 2 of the backseat eccentric wheel is positive, the eccentric wheel counterweight is positioned on the right side of the vehicle body, and when the rotation angle theta 2 of the backseat eccentric wheel is negative, the eccentric wheel counterweight is positioned on the left side of the vehicle body. The electric rotating wheel mechanism m6 comprises two rotating wheels which are perpendicular to each other: the bicycle comprises a horizontal rotating wheel and a vertical rotating wheel, wherein the horizontal rotating wheel is positioned right above a rear seat eccentric wheel, the center of the horizontal rotating wheel and the center of the rear seat eccentric wheel are positioned on the same vertical plane, and the vertical rotating wheel is tangent to the horizontal rotating wheel and is parallel to a rear wheel of the bicycle; the electric rotating wheel mechanism m6 performs the auxiliary balance of the rear seat part of the bicycle and the indirect control of the rotation of the rear wheel by adjusting the rotating speeds of the two rotating wheels.

The handlebar control module, the vehicle body middle control module and the vehicle body rear control module control each mechanism of the modules according to the information provided by the sensor module, thereby carrying out the indirect control of the balance and the advancing of the bicycle. The control variables of the handlebar control module, the middle control module of the vehicle body and the rear control module of the vehicle body are coupled with each other.

The control method of the unmanned bicycle with the self-balancing function comprises two parts, namely a self-balancing control method and an unmanned control method.

The self-balancing control method comprises but is not limited to a data acquisition driving control method, a bicycle model driving control method, a behavior driving control method, a key balance decomposition control method, an equivalent mapping control method, a self-evolution control method, an environment evolution self-adaptive evolution control method and a competition and cooperation control method.

The control method of competition and cooperation provides a method for carrying out the balance control learning of the unmanned bicycle based on competition and cooperation, the controller gives random control in real time while a person rides the bicycle, the fitness of the bicycle is judged according to whether the actual state change and the expectation of the bicycle are the same or not, and the steps are repeated so as to obtain the corresponding relation between the control and the state.

As shown in fig. 4, the specific steps of the contention and cooperation control method are as follows:

1) selecting controllable considerable key variables, wherein the controllable considerable key variables comprise bicycle variables and control variables of a handlebar control module, a bicycle body middle control module and a bicycle body rear control module, the bicycle variables comprise a bicycle handlebar deflection angle α, a bicycle body deflection angle β, a bicycle rear wheel rotation angle phi and primary and secondary derivatives thereof, and the control variables comprise a handlebar slide block position x, a bicycle body eccentric wheel rotation angle theta 1, a rear seat eccentric wheel rotation angle theta 2 and primary and secondary derivatives thereof;

2) real-time learning and controller establishment based on competition and cooperation: and the real-time learning of the controller parameters is carried out on the basis of the principle of competition and cooperation. When a user rides a bicycle, the bicycle is randomly controlled by a controller, the change direction of each variable of the bicycle, which is caused by the variable change of the controller, is predicted without considering the bicycle control by a person, if the actual control result of the bicycle is opposite to the prediction, the control method is optimized, if the actual control result of the bicycle is the same as the prediction, the control method is reserved, the corresponding relation between the current bicycle variable (namely the current bicycle state) and the change of the control variable is established, and the operation is repeated under various bicycle states; the control method can be divided into various road conditions and terrain conditions to converge to respective control schemes;

the behavior-driven control method provides a method for establishing control rules based on data of riding a bicycle, and directly controlling the balance of the unmanned bicycle through secondary mapping of a human body variable and a control variable, and the bicycle is directly controlled after the association between a controller variable and a bicycle variable is established.

The self-evolution control method provides a method for carrying out unmanned bicycle balance control learning based on self-evolution, after setting a basic rule and an evolution target, the bicycle is physically simulated in a three-dimensional simulated physical simulation space in a flat land evolution environment to generate a basic control rule set, and the self-evolution of the control rule is carried out through continuously improving the requirement on the control precision and continuously complicating the environment on the basis.

As shown in fig. 5, the self-evolution control method specifically includes the following steps:

2) setting basic rules and evolution targets: setting basic physical rules, namely establishing the physical rules which the key variables need to follow; setting an evolution target, namely establishing the evolution target of the key variable which needs to be met by bicycle balance; the physical rules comprise physical rules of a bicycle structure and physical rules under an evolution environment; the evolution environment comprises basic terrains such as flat land and sloping fields with various angles, and complex terrains such as semi-pipeline field and muddy land;

3) the basic control rule set is generated, namely, the physical simulation of the bicycle in a flat ground evolution environment is carried out in a simulated physical simulation space, and a bicycle control method capable of keeping basic balance is obtained by screening through an evolutionary algorithm, wherein the fitness of the bicycle to the environment is measured by whether each variable of the bicycle is in a certain value or stable in a normal interval, the normal interval can be that the bicycle is not in contact with the ground, namely, the bicycle body deflection angle β is greater than a threshold value B, data are generated through the methods, namely, the bicycle is ridden in the simulated physical simulation space through the bicycle control strategies, so that bicycle control data capable of keeping the basic balance are obtained, the establishment of the control rules is carried out through the data, and the refined rules are gradually summarized into the basic control rule set;

4) self-evolution: on the basis of a basic control rule set, continuously increasing the control rules by continuously improving the requirements on the control accuracy and continuously complicating the environment to form a new control rule set so as to carry out iteration, and finally obtaining an evolved control rule set which can adapt to a certain complex environment and has a certain control accuracy; the control precision refers to the accuracy and stability of controlling the balance of the bicycle; the continuous complexity of the environment can expand the evolution environment from flat land to sloping land, and a few to many depressions or high lands occur at random; the method for forming the new control rule set is to randomly increase the control rules or the control rule groups, if the control result of the basic control rule set and the new rules does not accord with the target, the control rules or the control rule groups are randomly increased again, otherwise, the new control rules or the control rule groups are divided into the control rule sets; the rule set after the control rule is added is the iterative evolution of a basic control rule set, and has a similar or self-similar relation with the previous basic control rule set, 2) the set basic rule has a constraint guide solution space effect on the finally evolved control rule set;

5) establishing a controller: based on the control rule derived by self-evolution, three controllers for bicycle variables are established by debugging the parameters of the handlebar control module, the middle part control module of the bicycle body and the rear part control module of the bicycle body.

The control method of the environmental evolution adaptive evolution provides a method for carrying out the balance control learning of the unmanned bicycle based on the environmental evolution adaptive evolution, after basic rules and evolution targets are set, the physical simulation of the bicycle in various evolution environments is carried out in a three-dimensional simulated physical simulation space, and key variables are gradually established to enable the bicycle to be in balance.

As shown in fig. 6, the method for controlling adaptive evolution of environmental evolution specifically includes the following steps:

3) the environmental evolution self-adaptive evolution method is an evolution algorithm and comprises but is not limited to a genetic algorithm (superior or inferior), the fitness of the bicycle to the environment in the environmental evolution self-adaptive evolution method is measured by whether the variables of the bicycle are stable in a normal interval or not, the normal interval can be that the bicycle does not contact the ground, namely the bicycle body deflection angle β is greater than a threshold value B;

4) establishing a controller: based on the correlation of key variables evolved by environment evolution self-adaption, three controllers for bicycle variables are established by debugging the parameters of the handlebar control module, the middle part control module of the bicycle body and the rear part control module of the bicycle body.

The bicycle model driving control method and/or the data acquisition driving control method can be further generalized to a model-based self-balancing control method of the unmanned bicycle with a self-balancing function, and a model is constructed through mechanism and/or data; the behavior driving control method and/or the key balance decomposition control method and/or the equivalent mapping control method can be further generalized to a behavior driving based self-balancing control method of the unmanned bicycle with a self-balancing function, and the behavior driving control method and/or the key balance decomposition control method and/or the equivalent mapping control method are directly used for controlling the balance of the bicycle; the self-evolution control method and/or the environmental evolution self-adaptive evolution control method and/or the competition and cooperation control method can be further generalized to be a self-balancing control method of the unmanned bicycle with the self-balancing function based on the intelligent evolution, the self-balancing control method is used for carrying out balanced learning through off-line and/or on-line evolution, and meanwhile, the unmanned bicycle with the self-balancing function based on the intelligent evolution also has an application and an application method of the unmanned bicycle with habit correction.

The habit-correcting unmanned bicycle application and application method provides the habit-correcting unmanned bicycle application and application method, and after healthy riding habits of athletes or coaches are learned, habit correction is performed through a bicycle variable control method that the superposition effect of three controllers and a user on bicycle control tends to be healthy.

The unmanned control method comprises a bicycle control method and method selection under various running states of starting, advancing, turning, backing and the like.

The bicycle control method under the starting state comprises the following specific steps:

2) and (3) adjusting the center of gravity of the bicycle by variable adjustment of the handlebar control module, the middle part control module of the bicycle body and the rear part control module of the bicycle body, so that the handlebar deflection angle α of the bicycle tends to 0, and the bicycle body deflection angle β of the bicycle tends to 0, even if the bicycle is in a vertical standing state from a certain deflection angle.

The bicycle control method under the forward state comprises the following specific steps:

3) indirect drive: the rear wheel of the bicycle is indirectly driven through the variable adjustment of a backseat rotating wheel mechanism of a control module at the rear part of the bicycle body, so that the rotating angle of the rear wheel of the bicycle is adjusted

At a certain angular velocity, even if the bicycle is moving forward at a certain speed.

The bicycle control method under the turning state comprises the following specific steps:

At a certain angular velocity even when the vehicle is derived to turn at a certain speed.

1) indirect drive: the rear wheel of the bicycle is indirectly driven through the variable adjustment of the rotating wheel mechanism of the control module at the rear part of the bicycle body, so that the rotating angle of the rear wheel of the bicycle is adjustedBecause there is a fore-and-aft relation in the place where handlebar contacts the ground with front wheel, in the backward state of bicycle, handlebar and front wheel are in the state of being dragged, the drag force in the handlebar junction is in the front, has dispelled the handlebar rotation torque produced while the bicycle advances, can simplify the regulation of the bicycle handlebar deflection angle α, the bicycle will tend to a whole in the backward state;

The specific steps of the selection of the bicycle control method under the various running states are as follows:

2) road surface control and obstacle avoidance: monitoring the road surface through a sensor module; carrying out terrain scanning, judging the terrain and selecting a control method corresponding to the terrain; judging whether an obstacle exists or not, if so, avoiding the obstacle, namely obtaining the traveling direction of the bicycle to be adjusted through road surface information such as distance, obstacle width, obstacle motion condition and the like so as to adjust;

example 1

The following description will specifically describe an unmanned bicycle having a self-balancing function, taking as an example the balance control of the unmanned bicycle by a competitive and cooperative control method.

At time t0, before the bicycle is put into use, the three devices, i.e., the handlebar control module, the middle body control module and the rear body control module (including the sensor module), are installed on a common bicycle.

And at the time t1, the product is directly put into use by the user. The user turns on the power switch, uses the unmanned bicycle as a common bicycle, starts the real-time competition and cooperation control methods of the three controllers to learn the balance under various road conditions and terrain conditions, and adds the unmanned control method.

2) real-time learning and controller establishment based on competition and cooperation: and the real-time learning of the controller parameters is carried out on the basis of the principle of competition and cooperation. When a user rides a bicycle, the bicycle is randomly controlled by the controller, the variable change directions of the bicycle, which are caused by the variable change of the controller, are predicted without considering the bicycle control by people, if the actual bicycle control result is opposite to the prediction, the control method is optimized, if the actual bicycle control result is the same as the prediction, the control method is reserved, the corresponding relation between the current bicycle variable (namely the current bicycle state) and the change of the control variable is established, and the operation is repeated under various bicycle states, as shown in fig. 7; the control method can be divided into various road conditions and terrain conditions to converge to respective control schemes;

at time t2, after the controller with self-balancing and unmanned functions is built, the self-balancing unmanned bicycle can be put into functional use by the user.

Example 2

In the following, a self-evolving control method is taken as an example to perform balance control of the unmanned bicycle in multiple environments, and an unmanned bicycle with a self-balancing function is specifically described.

At time t0, before the bicycle is put into use, three devices, namely a handlebar control module, a middle body control module and a rear body control module (including a sensor module), are installed on a common bicycle, a controller is established based on a self-evolving control method, and an unmanned control method is added.

3) generating basic control rules, namely performing physical simulation of the bicycle in a flat ground evolution environment in a simulated physical simulation space, and screening through an evolutionary algorithm to obtain a bicycle control method capable of keeping basic balance, wherein the fitness of the bicycle to the environment is measured by whether each variable of the bicycle is a certain value or stable in a normal interval, wherein the normal interval can be that the bicycle is not in contact with the ground, namely the bicycle body deflection angle β is greater than a threshold value B, and generating data through the methods, namely the 'brains' ride the bicycle in the simulated physical simulation space to obtain bicycle control data capable of keeping the basic balance;

4) self-evolution: on the basis of the basic control rule set, the requirements on the control accuracy are continuously improved and the environment is continuously complex, the control rules are continuously increased to form a new control rule set so as to carry out iteration, and finally, an evolved control rule set which can adapt to a certain complex environment and has a certain control accuracy is obtained, as shown in fig. 9; the control precision refers to the accuracy and stability of controlling the balance of the bicycle; the continuous complexity of the environment can expand the evolution environment from flat land to sloping land, and a few to many depressions or high lands occur at random; the method for forming the new control rule set is to randomly increase the control rules or the control rule groups, if the control result of the basic control rule set and the new rules does not accord with the target, the control rules or the control rule groups are randomly increased again, otherwise, the new control rules or the control rule groups are divided into the control rule sets; the rule set after the control rule is added is the iterative evolution of a basic control rule set, and has a similar or self-similar relation with the previous basic control rule set, 2) the set basic rule has a constraint guide solution space effect on the finally evolved control rule set;

And at the time t1, the controller with the self-balancing function and the unmanned function is built and then is put into use by the user. A user turns on a power switch, and the unmanned bicycle with the self-balancing function is started based on the bicycle control method in the starting state.

At the time t2, the user rides the self-balancing unmanned bicycle to perform acrobatic training in a complex field, and the self-balancing unmanned bicycle directly performs self-evolution control rule learning.

The self-evolution control rule learning means that when the control precision is reduced, rule evolution is carried out, and a new control rule is added to adapt to a new environment.

At the time of t3, a user gets off the vehicle, places some materials on the unmanned bicycle, sets unmanned automatic driving in a section of complex environment, and the unmanned bicycle with the self-balancing function can also perform self-evolution control rule learning, and meanwhile, the bicycle is adjusted based on selection of bicycle control methods in various running states, and is driven in an indirect driving mode.

the indirect drive is to indirectly drive the rear wheel of the bicycle through the variable adjustment of a backseat rotating wheel mechanism of a control module at the rear part of the bicycle body, so that the rotating angle of the rear wheel of the bicycle is changed

Varying at a certain angular velocity even when the bicycle is moving forward at a certain speed, as shown in fig. 11.

And at the time t4, the pilotless bicycle with the self-balancing function arrives at a specified place and waits for a next command.

Example 3

The following describes an unmanned bicycle with self-balancing function, taking an example of using a control method of environment evolution adaptive evolution to perform balance control of the unmanned bicycle in multiple environments.

At a time t0, namely before the bicycle is put into use, three devices, namely a handlebar control module, a bicycle body middle control module and a bicycle body rear control module (including a sensor module), are installed on a common bicycle, a controller is established based on a control method of environment evolution self-adaptive evolution, and an unmanned control method is added.

At the time t2, the user rides the self-balancing unmanned bicycle to perform acrobatic training in a complex field, and the self-balancing unmanned bicycle performs environment evolution self-adaptive evolution control scheme selection based on terrain scanning.

The terrain is judged by the terrain scanning according to the terrain scanning-based environment evolution adaptive evolution control scheme, and if the scanned terrain is learned during environment evolution adaptive evolution, the control is directly carried out according to the corresponding control scheme under the environment evolution adaptive evolution control method; if the scanned terrain is not learned during the environment evolution self-adaptive evolution, interaction with a server is carried out, and a control scheme corresponding to the terrain is obtained; if the scanned terrain exists in the database in the server, the controller parameters corresponding to the terrain are directly sent back to the unmanned bicycle, and if the scanned terrain does not exist in the database in the server, the environment evolution self-adaptive evolution needs to be carried out again on the basis of the scanned terrain, the controller parameters corresponding to the terrain are obtained, and the controller parameters are sent back to the unmanned bicycle. The specific steps of the selection of the environmental evolution adaptive evolution control scheme based on terrain scanning are shown in fig. 10.

At the time of t3, a user gets off the vehicle, places some materials on the unmanned bicycle, sets a section of unmanned automatic driving in a complex environment, and the unmanned bicycle with the self-balancing function can also perform environment evolution self-adaptive evolution control scheme selection based on terrain scanning, and simultaneously performs bicycle advancing adjustment based on selection of bicycle control methods in various running states, and is driven in an indirect driving mode.

the indirect drive indirectly drives the rear wheel of the bicycle through the variable adjustment of a backseat rotating wheel mechanism of a control module at the rear part of the bicycle body, so that the rotating angle phi of the rear wheel of the bicycle changes at a certain angular speed, even if the bicycle is driven to move forwards at a certain speed, as shown in fig. 11.

Example 4

The following describes an unmanned bicycle with a self-balancing function, taking the driving habit learning of the unmanned bicycle as an example.

At time t0, before the bicycle is put into use, three devices, namely a handlebar control module, a vehicle body middle control module and a vehicle body rear control module (including a sensor module), are installed on a common bicycle and are set through a self-balancing control method and an unmanned control method.

At time t1, the bicycle is used by the athlete or coach for a period of time, and the healthy driving habit with the least damage is learned, so as to obtain a healthy bicycle variable control method, namely, a good riding habit.

At time t2, the user is invested in practice and habit correction is performed. The habit correction is a bicycle variable control method which is implemented by superposing the control of a bicycle by three controllers on the control of a user on the bicycle so that the superposition effect tends to be healthy; thus, if the riding habit of the user is not good, the controller gives an additional disturbance, and the user feels hard, so that the user tends to use the healthy riding habit to ride the bicycle, and the user has a good riding habit.

Example 5

At time t1, the user is invested in the learning of the personalized driving habits. The riding habit learning is the learning of the driving habit of the user who uses the unmanned bicycle for a long time.

At time t2, another person (or a thief) rides the user's bicycle, and the bicycle also continues to learn the driving habits of the other person, thereby determining the change of the riding person. And then, the unmanned bicycle can contact the user through the server terminal for confirmation, judge whether the user borrows or rents the bicycle, and further contact police or related mechanisms through the server terminal if the user does not borrow or rents the bicycle for a period exceeding the renting period, and provide positioning for the user. If the user trades the unmanned bicycle, the driving habit of any previous user needs to be cleared through related authorization.

Example 6

In the following, an application of the shared unmanned bicycle to intelligent taxi calling and returning is taken as an example, and an unmanned bicycle with a self-balancing function is specifically described.

At time t0, before the bicycle is put into use, the three devices, namely the handlebar control module, the middle body control module and the rear body control module (including the sensor module), are installed on a common shared bicycle and are set through a self-balancing control method and an unmanned control method.

At time t1, the bicycle is directly released to the street for the user, and each unmanned bicycle should have its own parking space and support.

At time t2, the user calls the car through the mobile phone software on the street, the server searches the nearest shared unpiloted bicycle to the car-calling place, and the shared unpiloted bicycle is started and automatically driven to the car-calling place. If this unmanned bicycle is in non-vertical state, then need the unmanned aerial vehicle housekeeper to go out, hang just bicycle with the hook, make it get back to vertical state to start and autopilot. The vertical state is the state when the deflection angle of the bicycle body is less than or equal to the deflection angle of the bicycle body when the rear wheel of the bicycle is supported.

At time t3, the shared unpiloted bicycle arrives at the location of the call and is available to the user.

At the time t4, after the shared unpiloted bicycles are used up by the user, the server can automatically screen out the area with the lowest density of the shared unpiloted bicycles in a certain range, and the shared unpiloted bicycles can automatically drive to the place suitable for parking in the area and park for the next use requirement.

Claims

1. A self-balancing unmanned bicycle based on intelligent evolution comprises a bicycle and a sensor module, and is characterized in that the bicycle further comprises a handlebar control module, a middle part control module of a bicycle body and a rear part control module of the bicycle body;

the sensor module is used for measuring bicycle variables, wherein the bicycle variables comprise a bicycle handlebar deflection angle α, a bicycle body deflection angle β and a bicycle rear wheel rotation angle phi;

the handlebar control module is positioned on a handlebar of the bicycle, and the center of gravity of the handlebar is adjusted through the center of gravity adjusting mechanism, so that the adjustment of the handlebar deflection angle α is realized;

the middle control module of the bicycle body is positioned in the middle of the bicycle body, and the center of gravity of the middle of the bicycle body is adjusted through the center of gravity adjusting mechanism;

the rear part control module of the bicycle body is positioned at the rear part of the bicycle, the gravity center of the rear part of the bicycle body is adjusted through the gravity center adjusting mechanism, and the balance control and the rear wheel rotation control of the rear part of the bicycle are performed through the rotating wheel mechanism; the adjustment of the rotation angle phi of the rear wheel of the bicycle is realized through the rotation control of the rear wheel;

the gravity center adjusting mechanisms of the handlebar control module, the middle control module and the rear control module are respectively controlled by the balance of the rear control module, so that the adjustment of the bicycle body deflection angle β is realized together;

the self-balancing unmanned bicycle needs to be matched and adjusted by three modules, namely a handlebar control module, a middle control module of the bicycle body and a rear control module of the bicycle body, which are respectively distributed at three positions, in a working state so as to realize self-balancing of the bicycle;

selecting the bicycle variable and the control variables of the handlebar control module, the middle part control module of the bicycle body and the rear part control module of the bicycle body as key variables; when a user rides a bicycle, the handlebar control module, the bicycle body middle control module and the bicycle body rear control module give random control in real time, control rules are screened according to whether the bicycle state change is actually the same as expected, and the real-time learning of the controller parameters is repeated in such a way, so that the balance control of the unmanned bicycle is realized.

2. The self-balancing unmanned bicycle based on intelligent evolution of claim 1, wherein the gravity center adjusting mechanism of the handlebar control module is a sliding rod mechanism transversely placed on the handlebar, and the handlebar control module adjusts the handlebar gravity center by adjusting the position of a sliding block of the handlebar sliding rod mechanism.

3. The self-balancing unmanned bicycle based on intelligent evolution of claim 1, wherein the center of gravity adjusting mechanism of the middle vehicle body control module is an eccentric wheel, and the middle vehicle body control module adjusts the center of gravity of the middle vehicle body by adjusting the rotation angle of the eccentric wheel.

4. The self-balancing unmanned bicycle based on intelligent evolution of claim 1, wherein the center of gravity adjusting mechanism of the rear vehicle body control module is an eccentric wheel, and the rear vehicle body control module adjusts the center of gravity of the rear vehicle body by adjusting the rotation angle of the eccentric wheel.

5. The self-balancing unmanned bicycle based on intelligent evolution of claim 1, wherein the rotating wheel mechanism of the rear body control module is two rotating wheels perpendicular to each other: the vertical rotating wheel is tangent to the horizontal rotating wheel and is parallel to the rear wheel of the bicycle; the rear control module of the bicycle body performs balance control and rear wheel rotation control on the rear part of the bicycle by adjusting the rotating speeds of the two rotating wheels.

6. A competition and cooperation control method of a self-balancing unmanned bicycle based on intelligent evolution is characterized by comprising a balance control part and an unmanned control part;

2) real-time learning and controller establishment based on competition and cooperation: when a user rides a bicycle, the bicycle is randomly controlled by a controller, the change direction of each variable of the bicycle, which is caused by the variable change of the controller, is predicted without considering the bicycle control of a person, if the actual control result of the bicycle is opposite to the prediction, the control method is optimized, if the actual control result of the bicycle is the same as the prediction, the control method is reserved, the corresponding relation between the current bicycle variable and the change of the control variable is established, and the parameters of the controller are learned in real time under various bicycle states based on the principles of competition and cooperation;

7. The method according to claim 6, wherein in the step 2), the control method converges to respective control schemes according to various road conditions and terrain conditions.

8. The method according to claim 6, wherein in the step 2), the control method is applied to habit correction, after learning healthy riding habits of athletes or coaches, the bicycle variable control method tends to be healthy through the superposition effect of the handlebar control module, the middle body control module, the rear body control module and the user on bicycle control, when the riding habits of the user are not good, the handlebar control module, the middle body control module and the rear body control module generate additional disturbance for the healthy riding mode, and the user feels strenuous, so that the user tends to ride with the healthy riding habits.

9. The method of claim 6, wherein the implementation of the drone control portion includes: selecting a target motion state, and controlling the bicycle in the target motion state; the motion state includes: starting, advancing, turning and retreating;

3) indirect drive: the rear wheel of the bicycle is indirectly driven through the variable adjustment of a rotating wheel mechanism of a control module at the rear part of the bicycle body, so that the rotating angle phi of the rear wheel of the bicycle changes at a certain angular speed, even if the bicycle is driven to move forwards at a certain speed;

3) indirect drive: the rear wheel of the bicycle is indirectly driven through the variable adjustment of a rotating wheel mechanism of a control module at the rear part of the bicycle body, so that the rotating angle phi of the rear wheel of the bicycle changes at a certain angular speed, even if the bicycle turns at a certain speed;

1) the bicycle is characterized in that a rear wheel of the bicycle is indirectly driven through variable adjustment of a rotating wheel mechanism of a control module at the rear part of the bicycle body, so that the rotating angle phi of the rear wheel of the bicycle is reversely changed at a certain angular speed, even if the rear wheel of the bicycle is reversely rotated at a certain speed, a front-back relation exists at the ground contact part of a handlebar and a front wheel, when the bicycle is in a backward state, the handlebar and the front wheel are in a dragged state, the dragging force at the joint of the handlebar is in the front, the handlebar rotating torque generated when the bicycle advances is eliminated, the adjustment of the handlebar deflection angle α of the bicycle can be simplified, and the bicycle tends to be a whole in the backward;

10. The method according to claim 9, wherein the selecting a target motion state is specifically:

1) macroscopic route determination: determining the integral traveling route of the bicycle in a navigation and manual selection mode;

2) road surface control and obstacle avoidance: monitoring the road surface through a sensor module; carrying out terrain scanning, judging the terrain and selecting a control method corresponding to the terrain; and judging whether an obstacle exists or not, and if so, avoiding the obstacle, namely obtaining the traveling direction of the bicycle to be adjusted according to the distance, the width of the obstacle and the movement condition of the obstacle so as to adjust.