LU101594B1 - Position and posture optimization method and device of an underground mobile robot realized based on membrane calculation - Google Patents

Abstract

The present invention discloses a position and posture optimization method and device of an underground mobile robot realized based on membrane calculation. The position and posture optimization method comprises the following steps: SI, building a robot motion model; and S2, building an organization-type membrane system of the robot, including: S2-1, building a probabilistic membrane calculation model framework of the robot, and S2-2, dynamically optimizing a position and posture of the robot under the probabilistic membrane calculation model framework of the robot. The present invention introduces the membrane calculation into optimization of the underground robot in a coal mine, so that the underground inspection working efficiency and accuracy of the robot are improved.

Description

] LU101594 NR. P100382LU00

POSITION AND POSTURE OPTIMIZATION METHOD AND DEVICE OF UNDERGROUND MOBILE ROBOT REALIZED BASED ON MEMBRANE CALCULATION

BACKGROUND Technical Field The present invention belongs to the field of optimization and control of underground transfer robots in coal mines, and particularly relates to a position and posture optimization method and device of an underground mobile robot realized based on membrane calculation. Related Art With the integrated development of Internet plus, big data and intelligent control, coal mining is facing reformation and innovation. Automation and intelligence are new trends in future coal mining, thus forming a highly intelligent, unmanned, efficient and safe coal mining method, which improves the automation and intelligence while achieving unmanned safe, efficient and accurate coal mining. An Internet of Things oriented to accurate coal mining relies on an intelligent control and communication network to enhance the real-time online monitoring and risk recognition of a coal mine for a complex underground environment of the coal mine and solve the problem of uncertainty of traditional disaster prewarning, and a mobile robot plays an important carrier role. Underground real-time monitoring, inspection, search and rescue work in the coal mine is completed through the robot, so that accurate disaster prewarning and determination is more perfect, and the new trends of intelligent and accurate mining of the coal mine are deeply integrated. However, due to the own characteristics of the mobile robot, in the complex underground environment of the coal mine, how to make the mobile robot complete the underground inspection task in the coal mine and further exert the underground application value in the coal mine has important research significance.

SUMMARY For the deficiencies of the prior art, the present invention aims to provide a position and posture optimization method and device of an underground mobile robot realized based on membrane calculation, which improves the robot position and posture accuracyand the system robustness and realizes underground inspection of a coal mine. The objective of the present invention may be implemented through the following technical solution: A position and posture optimization method of an underground mobile robot realized based on membrane calculation is provided. The position and posture optimization method includes the following steps: S1, building a robot motion model; and S2, building an organization-type membrane system of the robot, including: S2-1, building a probabilistic membrane calculation model framework of the robot; and S2-2, dynamically optimizing a position and posture of the robot under the probabilistic membrane calculation model framework of the robot. An underground mobile robot realized based on membrane calculation is provided. A robot device includes a chassis, a device power supply, an artificial intelligence (Al) computer, a sensor, an inertial measurement unit, a laser radar and a double-sided depth camera. The double-sided depth camera is located at a topmost end of the robot. The laser radar, the inertial measurement unit and the sensor are located on a front adjustment rod of the robot side by side. The AI computer is located at a middle position of the robot. The device power supply is mounted at an upper end of the chassis. The power supply device, the sensor, the inertial measurement unit, the laser radar and the double-sided depth camera are all electrically connected with the AI computer. The present invention has the following beneficial effects.

1. The present invention introduces the membrane calculation into the optimization of the underground mobile robot in the coal mine to improve the underground inspection working efficiency and accuracy of the robot.

2. According to the complexity of the underground environment in the coal mine, the present invention combines the binocular camera and the laser radar to provide a guarantee for underground navigation of the robot, quickly and effectively receives and transmits data through the configured sensor, and can quickly complete the calculation while performing optimization control on the position and posture of the robot by using a Nvidia TX2 intelligent computer.

3. Simulations and experiments show that a provided membrane algorithm has fastconvergence and effectiveness, and the designed device can effectively solve the problem on inspection in the complex underground environment in the coal mine to provide a theoretical support and practical application value for future accurate mining of the coal mine by its efficient calculation performance and fast optimization performance.

BRIEF DESCRIPTION OF THE DRAWINGS To describe the technical solutions in the embodiments of the present invention or in the prior art more clearly, the following briefly describes the accompanying drawings required for describing the embodiments or the prior art. Apparently, persons of ordinary skill in the art may still derive other drawings from these accompanying drawings without creative efforts. FIG. 1 is an overall schematic structural diagram according to an embodiment of the present invention; FIG. 2 is a schematic diagram of a robot optimization simulation result based on a shortest path according to an embodiment of the present invention; FIG. 3 is a schematic diagram of robustness analysis of a membrane algorithm according to an embodiment of the present invention; FIG. 4 is a schematic diagram of velocity error analysis of the membrane algorithm according to the embodiment of the present invention; FIG. 5 is a schematic diagram of tracking error analysis of the membrane algorithm according to the embodiment of the present invention; and FIG. 6 is a schematic structural diagram of an underground mobile robot realized based on membrane calculation according to an embodiment of the present invention.

DETAILED DESCRIPTION The following clearly and completely describes the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments instead of all embodiments of the present invention. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention. As shown in FIG. 1, a position and posture optimization method of an undergroundmobile robot realized based on membrane calculation includes the following steps that: S1, a robot motion model is built: S1-1, due to the complexity of an underground environment of a coal mine, the robot is set in a planar environment, and a variable of a motion state of the robot is a position and posture and consists of plane coordinates (x,y) and an azimuth angle 0; a vector is denoted by (x.y,0)"; the position and posture and the environment participating in the calculation jointly form the motion state x; of the robot.

An initial position and posture x-1=(x,y,0)" and a subsequent position and posture x=(x",y’.0°)F are set, and u=(v,œ)" is controlled, wherein ¥ is a translational velocity of the robot, and o is an angular velocity.

The robot is controlled to be executed at time At.

The interference of error parameter noise of the motion model, a real velocity and a measured velocity are different, so that the model should consider controlling the noise.

When At is relatively small, the model of the robot may be determined as: X sin(0+ A,6)- "sin (8) a œ x, =(x,7,6)" + 2 cos(6) Loos (6+4,0) (1): Ad S1-2, since a probability density uses three-dimension to denote two-dimension, the formula (1) is extended.

When rotating &, the robot reaches a final position and posture, then: 0'=0+A 0 + Aa (© =v? + eo” (2); S1-3, a final model of the mobile robot may be obtained through the formulas (1) and (2) as: Zsin(0+4,@)- = sin (6) x, =(x,v,0} + = co3(8) =~ cos(0 + 4,0) A, à+A, 0 3). S2, an organization-type membrane system of the robot is built, including that:

in the organization-type membrane system, a rule is executed in a probabilistic manner, and data may be quickly processed, and an independent and collaborative data processing membrane system is formed in combination with parallel and distributed features of the membrane calculation.

A modeling principle and process for the 5 underground mobile robot in the coal mine by using the organization-type membrane system are shown in FIG. 1. S2-1, a probabilistic membrane calculation model framework of the robot is built: S2-1-1, a real-time position of the robot is determined, and robot data {x,v,6)" and position update output data (x',y',6")" are received within each period at which a membrane controller starts to be executed.

A following membrane system with a degree of 3 is built according to the distributed parallel features of the membrane calculation: T1=(M. pw, wp wy Rc}, ) wherein (1) M = {x,.3,.6,.Err HL, jE {1 2]}; (2) u= [I LI l, ], , a membrane mechanism of the organization membrane system ofthe robot; (3) w, = p(x, | 4x). density function resolution; (4) w, = p(x lux, ) , density calculation after noise interface; (5) w, =p(%, | 4x), ideal density evaluation; (6) R is a rule set, which maintains energy conservation in a transfer process; and (7) €, denotes a probability of evolution of an object based on an r rule.

S2-2, the position and posture of the robot are dynamically optimized under the probabilistic membrane calculation model framework of the robot, including: 1): determining a degree of a membrane structure as 3 according to the probabilistic membrane calculation model framework of the robot of S2-1, and initializing the object {x,7.8)7 in each layer of membrane; 2): evolving the rule set according to a robot posture dynamics model, and ensuring that data in an evolved rule membrane is optimal; and 3): continuously returning to execute 1) according to an update of the data in a surface layer of membrane executed in 2).

: 6 LU101594 The above evolution of the rule set is a set of the evolved rules. Objects meeting evolution conditions in the membrane are executed in a probabilistic manner, so as to process uncertain or random data acquired in the process of controlling the position and posture of the robot.

The optimization of the data in the membrane is based on an evolved rule. The data (objects) suitable for the evolved rule are executed according to a probability of occurrence of the evolved rule. Due to the presence of a plurality of membranes, a condition for evolution or transferring of the data in the subordinate membrane is that a sum of the probabilities of execution of all the rules is 1, thus guaranteeing concurrency of inter-membrane data processing, avoiding local (in a certain membrane) convergence and achieving a global optimization effect. Therefore, this is called optimization of the data in the membrane.

A P-lingua file suitable for dynamic change of the position and posture of the underground mobile robot in the coal mine is defined according to the above probabilistic membrane system. A file framework is shown below.

@ model<probabilistic> Def init membrane_structure() { @Mu=[[P2[13} E; } Def init_multisets() { @ ms(1)+=x,y.6; @ ms(2)=x,y; @ ms(2)=0; } Def init_rules() { Evolutionary rule-sets; } Def main() {

Call init membrane_structure(); Call init_multisets(); Call init_rules(); ;

In order to verify that the built robot motion model has feasibility under the probabilistic membrane algorithm compared to a common algorithm and that the membrane algorithm is higher in convergence than the common algorithm, the applicant has made the following verifications:

1) Feasibility analysis of the built robot motion model in the probabilistic membrane algorithm: According to the final model of the mobile robot according to formula (3), proving the feasibility of this model under the probabilistic membrane algorithm is verified, and a robot error model is built: exp) fe, a) = ——22— Vara (6). It is assumed that the robot is controlled to proceed at a determined time interval At , and the initial and subsequent positions and postures are respectively denoted as x, =(x,y,0},x =(x<,y,60)" , and u,=(v,®,)" is controlled, the probabilistic membrane algorithm is defined as follows: Algorithm — membrane — probabilistic(x,,u,,x,_,) x = su») y PAP 4 u(x —x) 2 U= (x-x)cos9+(y-y)sin@ 2(y- y )cos9—(x-x )sin0 r=Je-x+09-vy} A6 =atan2(y —y,x —x')-atan2(y-y',x-x")

Dar A0 At à = 40 At 7-220 5 At return = fy-5,°+0")- f(o-dV +0) f(y —7.V +0").

Based on the foregoing probabilistic membrane algorithm, in order to guarantee data initialization under a probabilistic P system and real-time data update under environment participation, a following data sampling membrane algorithm under the consideration of disturbance is built: Algorithm membrane data sampling (4,,x,_) : ÿ =v +sampling(v’ +”) @ = w+ sampling(v’ +0”) 7 = sampling( + ©”) 0 = 0+ At + AL x = x-2 sin 9 + sin(@+ DAL)

D D y= y = cos 0 +} cos(0 + BAN

D D return x, =(x,y,0), where x, is a sampling position and posture, and 7 is a disturbance term.

The algorithm includes the following specific execution steps: 1): population initialization: initialized robot posture sequences are arrayed and combined, and initial individuals are used as the posture sequences; sequence rules are encoded from 1 to n, i.e., an individual has a length of n; 2): calculation of an error function: the probabilistic membrane algorithm and a sampling membrane algorithm are executed to further evaluate the quality of results and update posture data; 3): calculation by selection operators: superior individuals are selected from a population according to the evaluation of population fitness, and the individuals withhigh fitness in the population replace those with low fitness; 4): calculation by crossover operators: the individuals are paired, and the paired individuals are replaced at a certain probability p to form new individuals; 5): calculation by mutation operators: individual values in the population are varied at a variation probability p, and if the varied new individuals have high fitness values, the new individuals are maintained, otherwise, the original individuals are maintained. 2) Convergence analysis of the built membrane algorithm The robot motion equation (3) meets the following natures and hypotheses: Nature 1: it is set that p(n) is the position of the underground robot, n is time and i is the number of times of iteration, so that a desired track is: p(n+1)=A(p,(n), nu, (7) + a(n) + p,(n) (6). Nature 2: a matrix A(p, (n),n) satisfies: | Ap, m= Alp, mI Al p= po |i (7). Nature 3: the matrix 4(p,(n),n) has a boundary which is a (p,(n),n) non-singular matrix.

Hypothesis 1 max max ||, (m) |IS 5, max max || 8,61) |< & (8). In order to satisfy the hypothesis 1, interference and errors are considered, and the formula (6) has: [1-LODAG EDIT.

Prove: if the nature 2 and the hypothesis 1 are considered, there is: | B,(n+Di< p,m) + AS (WI HS) AO). When recursion is made on (10), there is: 150s (1+) Taml+1 a. = In combination with an iterative control rate, there is: i, (n)=[I-L,(mA(p, (n),n)]#, (m) — L,(n) BD, (m) — L (m[A(p, (n),n), n] (12). —A(p, (n),n)u, (n) = L,(n) p,,,(n) +L, (n)[ , (n)-7,(n)] By use of the nature 1 and the nature 2, there is: Wiz, (mill = Lim) Ap, (rm), EHEN BIS. (ll (13).

When the formulas (9) and (11) are substituted into (13), there is: ld, (nla mI HE +O HAN CDI +2 DIE (14). According to a geometric progression, there is: -j- a 12h n-j-1 1 h CZ) _ — — (ym h a 2 a 2 h 10e had n hn —l QP (8,97 =D 15) o/ _ a(% 1) aN 1-(=) <x a-h It is assumed that a is large enough, there is: LE ê lim ||, ||, < — lim|[# |, 5 (16). The formula (11) is transformed to obtain: h 1-(—)” - 1, nl 17). BMI s—2—Ja jy +¢ 1D a ah a From the formulas (16) and (17), there is: h" h" 1-0) g 1-0) 18 Bl — —_4— € 45 08. liml à ll ach 1p ach g It can be concluded from the above reasoning that: when a coefficient is 0, i.e., if no disturbance exists, || & |l,.|l P; | convergence is 0, or if disturbance exists, there is bounded convergence, thus proving the effectiveness of the algorithm. 3) Analysis of simulation and experiment results of the mobile robot The probabilistic membrane system calculation model of the underground mobile robot in the coal mine considers the complexity of the underground environment and noise interference factors, and the system evolved rules are executed by probabilistic features.

Previous experiments have verified the consistency of a P_Lingua model and a MeCoSim simulation result, a membrane calculation simulation software MeCoSim platform is built to complete a probabilistic membrane system simulation experiment of the position and posture of the underground mobile robot in the coal mine.

It is assumed that the robot verifies the membrane algorithm under the built organization-type membrane system in a closed corridor imitated laneway. To verify the tracking optimization performance of the membrane controller and the effectiveness of the membrane algorithm in the summary of the invention, 10 and 20 node tags are set respectively. The number of individuals in the population is 40, a crossover probability pl is equal to 0.15, and a mutation probability P2 is equal to 0.85. According to the designed membrane controller structure and algorithm, by changing the number of generations of evolution, the quality of path optimization may be determined, and optimization results are shown in Table 1 and FIG. 5. Table 1 Analysis of positioning errors of the algorithm Positioning error of Positioning error of probabilistic membrane system Strapdown inertial navigation The X direction | The Y direction The X direction | The Y direction

0. 052 0. 499 0.022 0. 200

0. 052 0. 501 0.021 0. 200

0. 050 0. 505 0.020 0.205 To verify the feasibility of the device and the convergence of the algorithm, a position estimation center point ı:,_, = (90,90,0), motion control u, = (20cm/s, 10°/s)" and time control of 10 s are molded by translational and rotational velocities, and positions and orientations of relative positioning tags are measured. Observation is carried out according to the order at each time. Velocity errors generated by the robot are as shown in FIG. 3 according to the designed membrane control structure and algorithm. It can be obtained through FIG. 3 that compared with a traditional genetic algorithm, this algorithm has the advantages that the velocity errors of the movement of the robot are relatively small and have extremely high convergence, the probabilistic membrane calculation model is higher in accuracy of calculation of the velocity of the robot, and the membrane algorithm is higher in accuracy of calculation of the velocity of the underground mobile robot. Robustness experiment results of the algorithm are shown in FIG. 4. The results show that the designed membrane algorithm is higher in stability.

As shown in FIG. 6, an underground mobile robot realized based on membrane calculation is provided. A mobile robot device consists of a chassis 1, a device power supply 2, an artificial intelligence (AI) computer 3, a sensor 4, an inertial measurement unit 5, a laser radar 6 and a double-sided depth camera 7. The double-sided depth camera 7 is located at a topmost end of the robot. The laser radar 6, the inertial measurement unit 5 and the sensor 4 are located on a front adjustment rod of the robot side by side. The Al computer 3 is located at a middle position of the robot. The device power supply 2 is mounted at an upper end of the chassis 1. The power supply device 2, the sensor 4, the inertial measurement unit 5, the laser radar 6 and the double-sided depth camera 7 are all electrically connected with the AI computer 3.

The device power supply 2 realizes power supplying while having a counterweight function, so that the physical stability of the robot in a traveling process may be further guaranteed.

The chassis 1 is preferably a crawler-type chassis.

The power supply device 2 includes the device power supply and a backup power supply, and has the characteristics conforming to an underground anti-explosion standard.

The Al computer 3 occupies a central position in the whole device. All data generated during the inspection task are calculated through the computer according to the foregoing built membrane calculation model and algorithm. The postures of the robot and the double-sided depth camera 7 are controlled to be changed according to calculation results, and an accurate guarantee is provided for path planning during the inspection process of the robot.

The sensor 4 includes a gas concentration measuring sensor and a CO concentration measuring sensor. Measured values are transmitted back to the AI computer 3 for threshold research and determination after being acquired, so as to realize real-time whole-process CO and gas monitoring.

The inertial measurement unit 5 has dynamic compensation and orthogonal compensation characteristics, an RS485 interface and underground anti-explosion features, and has a posture angle error range of less than or equal to lo, a gyroscope range of 500 o/sec and a noise density of less than 0.005 o.

The laser radar 6, with the model number of RPLIDARA2M8, has a dimension

(length: width: height) of 8 cm: 8 cm: 5 cm, a measurement radius of 12 m and a frequency of 8000 times/s, and may realize scanning ranging within 360 o, a scanning frequency may be controlled through a pulse width modulation (PWM) signal, and the laser radar is applied to subsequent underground 3D environment construction and robot self-positioning for the coal mine.

The double-sided depth camera 7, with the model number of Intel RealSense D4351, has a dimension (length: width: height) of 9 cm: 2.5 cm: 2.5 cm, is provided with the depth sensor module 4 and the IMU, and may realize a 6 DoF tracking function in combination with visual data.

During use, a bracket of the double-sided depth camera 7 has an automatic adjustment function of full degree-of-freedom, so that the double-sided depth camera 7 may complete accurate shooting for found problems in the inspection process. An image is transmitted back to the AI computer 3 for image processing, and then is transmitted back to a ground station in real time, and predetermined operations are executed according to a processing result of the computer. Meanwhile, the double-sided depth camera 7 may be integrated with the laser radar 6 (located below the depth camera) to further satisfy the subsequent path optimization work in the complex underground environment of the coal mine.

Further, the power supply device 2, the sensor 4, the inertial measurement unit 5, the laser radar 6 and the double-sided depth camera 7 are all controlled by the AI computer 3 in a wired communication manner, so as to guarantee the communication efficiency and improve the overall calculation performance of the device.

A controller model is an organization-type membrane controller model which is reconstructed and makes use of the advantage of the distributed calculation performance of the membrane calculation to realize the real-time effectiveness of data acquisition-transmission-processing-feedback control, so that position and posture optimization is realized, and a very good support is provided to the subsequent path optimization.

The whole device realizes autonomous control of the robot based on the constructed membrane controller and the membrane algorithm, and is applicable to a new form of future intelligent mining of coal, thereby further increasing the intelligence level. Therefore, the position and posture optimization and device of the underground mobilerobot in the coal mine based on the organization-type membrane system have important theoretical research significance and practical significance.

Following the development trend of the future accurate coal mining, the present invention innovatively introduces the membrane calculation into the optimization of the underground robot in the coal mine, thereby improving the efficiency and accuracy of the underground inspection work of the robot. The specific effects are expressed as follows.

The effectiveness of the algorithm: The membrane algorithm based on the membrane calculation model makes full use of the parallelism characteristics of the membrane system. By constructing the organization-type membrane system adapted to the control of the robot, the membrane algorithm based on the system is designed, and the operation results of the algorithm prove the effectiveness of the algorithm.

Design of the membrane controller: The robot dynamics and the kinematics characteristics are fully considered, so that each parameter control quantity is quickly and parallelly operated in each membrane according to a control rate, and results are returned to a set surface membrane control unit. Optimization of the position and posture parameters of the robot is realized through finite iterative operation.

Device design: According to the complexity of the underground environment in the coal mine, the combination of the binocular camera and the laser radar guarantees the underground navigation of the robot. The data is quickly and efficiently received and transmitted according to the configured sensor. A Nvidia TX2 intelligent computer is used to quickly complete the calculation and perform optimization control on the position and posture of the robot at the same time.

Simulations and experiments show that the provided membrane algorithm has fast convergence and effectiveness, and the designed device can effectively solve the problem on inspection in the complex underground environment of the coal mine, and provides a theoretical support and practical application value for the future accurate mining of the coal mine with its efficient calculation performance and fast optimization performance.

In the descriptions of this specification, descriptions using reference terms such as "an embodiment”, "an example", or "a specific example" mean that specificcharacteristics, structures, materials, or features described with reference to the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, schematic descriptions of the foregoing terms do not necessarily directed at a same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or more embodiments or examples in an appropriate manner.

The basic principles, main features and advantages of the present invention have been shown and described above. Those skilled in the art should understand that the present invention is not limited by the foregoing embodiments, descriptions in the foregoing embodiments and the specification merely describe the principles of the present invention, various changes and improvements may be made to the present invention without departing from the spirit and scope of the present invention, and such changes and improvements shall all fall within the protection scope of the present invention.

Claims (8)

CLAIMS What is claimed is:

1. A position and posture optimization method of an underground mobile robot realized based on membrane calculation, comprising the following steps: S1, building a robot motion model; and S2, building an organization-type membrane system of the robot, comprising: S2-1, building a probabilistic membrane calculation model framework of the robot; and S2-2, dynamically optimizing a position and posture of the robot under the probabilistic membrane calculation model framework of the robot.

2. The position and posture optimization method of the underground mobile robot realized based on the membrane calculation according to claim 1, wherein the step S1 of building the robot motion model comprises the following steps: S1-1, forming the robot motion model by plane coordinates (x,y) and an azimuth angle 6, denoting a vector by (x,y,6)", forming a motion state xi of the robot by the position and posture and an environment participating in calculation, setting an initial position and posture x-1=(x,y,0)" and a subsequent position and posture x=(x",y’,0")", and controlling ut=(V,0)", wherein ¥ is a translational velocity of the robot, and w is an angular velocity; the robot is controlled to be executed at time At; when At is close to 0, the model of the robot is: sin (0 + 4,0) -=sin (9) x, =(x,»,60) + 2 cos(6) -Ÿ cos(0+ 4,0) Ad (D; S1-2, since a probability density uses three-dimension to denote two-dimension, extending the formula (1); when the robot rotates & making the robot reach a final position and posture, then: 0'=0+A, 0 + Aa a =v’ + £0’ 2);

and S1-3, obtaining a final model of the robot through the formulas (1) and (2) as: vo. NV 5sin(0+4,0)- sin (9) T |Ÿ Ÿ _ x, =(x,y,0) + 7 cos(0)- = cos (0 + A0) A,à+A,0 3).

3. The position and posture optimization method of the underground mobile robot realized based on the membrane calculation according to claim 2, wherein the step S2-1 of building the probabilistic membrane calculation model framework of the robot comprises: S2-1-1, determining a real-time position of the robot, and receiving robot data (x,v,6)" and position update output data {x’,y’,6"T within each period at which a membrane controller starts to be executed, and building a following membrane system with a degree of 3: IT=(M, sw, ww; Rc} cn) ; wherein (1) M = {x 2,20, Err :i,j e[L 2]}; (2) p= [[ Li ], | , a membrane mechanism of the organization membrane system of the robot; (3) w, = p(x |4,x_), density function solution; (4) w, = p(x | ux, ) , density calculation after noise interface; (5) w, = p(X | 4.x), ideal density evaluation; (6) R is a rule set, which maintains energy conservation in a transfer process; and (7) ce, denotes a probability of evolution of an object based on an r rule.

4. The position and posture optimization method of the underground mobile robot realized based on the membrane calculation according to claim 3, wherein the step S2-2 of dynamically optimizing the position and posture of the robot comprises the following steps: 1): determining a degree of a membrane structure as 3 according to the probabilisticmembrane calculation model framework of the robot of S2-1, and initializing the object (x,”e)T in each layer of membrane; 2): evolving the rule set according to a robot posture dynamics model, and ensuring that data in an evolved rule membrane is optimal; and 3): continuously returning to execute 1) according to an update of the data in a surface layer of membrane executed in 2).

5. An underground mobile robot realized based on membrane calculation, a robot device comprising a chassis 1, a device power supply 2, an artificial intelligence (AI) computer 3, a sensor 4, an inertial measurement unit 5, a laser radar 6 and a double-sided depth camera 7, wherein the double-sided depth camera 7 is located at a topmost end of the robot; the laser radar 6, the inertial measurement unit 5 and the sensor 4 are located on a front adjustment rod of the robot side by side; the Al computer 3 is located at the middle position of the robot; the device power supply 2 is mounted at an upper end of the chassis 1; the power supply device 2, the sensor 4, the inertial measurement unit 5, the laser radar 6 and the double-sided depth camera 7 are all electrically connected with the Al computer 3.

6. The underground mobile robot realized based on the membrane calculation according to claim 5, wherein the chassis 1 adopts a crawler-type chassis.

7. The underground mobile robot realized based on the membrane calculation according to claim 5, wherein the power supply device 2 comprises a device power supply body and a backup power supply.

8. The underground mobile robot realized based on the membrane calculation according to claim 5, wherein the sensor 4 comprises a gas concentration measuring sensor and a CO concentration measuring sensor.