Method, medium and device for realizing multi-task motion of robot based on sequential logic

Abstract

The present invention provides a method, medium and device for realizing multi-task motion of a robot based on temporal logic. The method expands the traditional TLTL to wTLTL (Weighted truncated linear temporal logic, hereinafter referred to as wTLTL), and combines DMPs with wTLTL to realize complex operation tasks with user preferences. The DMPs are used to approximate the trajectory with user preferences, and the trajectory shape is determined by the DMPs parameters, and the black box optimization strategy (Policy Improvement with Black‑Box optimization, hereinafter referred to as PI) based on the weighted average covariance matrix is ​​used to optimize the trajectory shape. BB ) for optimization. The wTLTL provides interpretable task specifications and is able to design a cost function with user preferences based on quantitative semantics. BB The algorithm is used to identify the optimal shape parameters of DMPs, thereby realizing the motion trajectory of complex operation tasks. The user preference is reflected by adjusting the weight in wTLTL. Compared with the traditional multi-task operation motion planning method, this method can quickly plan the motion trajectory of complex operation tasks with user preferences, with low computational complexity and high planning efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of robot motion planning, and in particular to a method, medium and device for realizing robot multi-task motion based on temporal logic. Background Art

[0002] With the continuous development of robotics technology, robots are gradually being used in scenarios such as offices, hospitals, and homes. These application scenarios usually require robots to be able to autonomously learn, plan, and execute complex operational tasks consisting of a series of logical actions.

[0003] Dynamic movement primitives (DMPs), as a flexible representation of motion trajectories, are widely used in learning and planning complex motion trajectories for robots, such as humanoid bipedal robots and dual-arm manipulators. DMPs are typically encoded as a combination of a simple linear dynamical system and a nonlinear component. A policy search algorithm is used to learn the optimal parameters of the nonlinear component, thereby obtaining smooth motion trajectories of arbitrary shapes. However, traditional DMPs methods are limited to simple go-to-goal tasks.

[0004] Path integrals algorithm (hereinafter referred to as PI) 2 ) is a typical representative of the DMPs trajectory parameter optimization algorithm, which adopts the parameterization strategy of probability weighted average and can use reinforcement learning to improve the strategy and optimize the path shape parameters. However, PI 2 The algorithm increases computational complexity, reduces parameter convergence speed, and cannot handle motion primitive parameter optimization in complex task operations.

[0005] Temporal logic, as a formal language, can describe a wide range of complex tasks in a concise and human-interpretable form, and is therefore increasingly used for motion planning in robotic systems. Common temporal logics used to describe robotic tasks include signal temporal logic (STL) and truncated linear temporal logic (TLTL). However, STL specifies subtasks with explicit time boundaries, while many manipulation tasks simply require subtasks to be executed in a desired order. Manually assigning time boundaries to subtasks can prevent the robot from finding a satisfactory strategy due to unexpected environmental events. TLTL's robustness, however, only considers the satisfaction of the TLTL formula in the most extreme cases, which may not lead to more robust solutions and fails to express user preferences. Summary of the Invention

[0006] The present invention proposes a method, medium and device for realizing multi-tasking motion of a robot based on sequential logic, which can solve at least one of the technical problems in the background technology.

[0007] To achieve the above object, the present invention adopts the following technical solutions:

[0008] A method for realizing multi-task motion of a robot based on sequential logic, wherein a computer device executes the following steps:

[0009] Step 1: Establish a DMPs approximate trajectory model based on the robot kinematic model;

[0010] Step 2: Encode the complex operation task specifications into an interpretable wTLTL formula and add user preferences by setting wTLTL weights;

[0011] Step 3: Design a cost function with user preference based on the quantitative semantics provided by the wTLTL formula, and based on the improved PI BB The algorithm optimizes the shape parameters of DMPs and generates multi-task operation trajectories with user preferences.

[0012] Furthermore, the step 1 is specifically as follows:

[0013] According to the robot kinematic model, the DMPs approximate trajectory model is established:

[0014]

[0015]

[0016]

[0017]

[0018] in, For location, For speed, and is the internal state; α z ,β z ,τ and α s is a positive proportional factor, θ is a parameter vector, g is the target position; nonlinear function f t Arbitrary complex motion can be generated, which consists of basis functions g, represented by piecewise linear functions with weighted Gaussian kernels as:

[0019]

[0020]

[0021] Among them, [g ti records the i-th term of g t , σ i and c i are the variance and mean, and its core idea is to obtain smooth motion of any shape through the non-linear term perturbation α z (β z (g - y t ) - z t ).

[0022] Furthermore, the specific content of the second step is as follows:

[0023] Encode complex operation tasks specifications into interpretable wTLTL formulas; where, the semantics of wTLTL is defined as:

[0024]

[0025] is a boolean value true, f(y t ) < c is a predicate, where f: maps the system state y<目标="0000018">to a constant, (not), ∨ (disjunction), ∧ (conjunction) are standard boolean operators, (eventually), □ (always), (until), (then), 〇 (next) and (if... then...) are operators;

[0026] When the weight vector of the boolean operator is a unit vector, i.e., ω i = 1, omit ω in the wTLTL formula; when all ω are unit vectors, traditional TLTL is regarded as a special case of wTLTL;

[0027] Add user preferences by setting wTLTL weights, where ω represents a positive weight vector, and the i-th term ω i of the conjunction and disjunction represents the relative importance of the defined specification or alternative priority, and the higher ω i , the higher its importance or priority; the semantics of the wTLTL formula is a finite trajectory on the system state. Let y<000目标="0000022">represent a series of states from y t to y t+k . If y t:t+k satisfies the wTLTL formula φ, then

[0028] Furthermore, the specific content of the third step is as follows:

[0029] ​Based on improved PI BB Algorithm, the shape parameters of DMPs are updated according to the following steps:

[0030] 1. Initialization, set the initial mean and covariance <θ, Σ> to <θ init ,λ init I>, where I is an identity matrix of appropriate dimension;

[0031] 2. Sampling, through Sample M sets of parameter vectors θ from N(θ,Σ) m ,m=1…M;where∈ m Obtained by sampling from a zero-mean Gaussian distribution with variance Σ;

[0032] Calculate the cost function for each set of samples:

[0033]

[0034]

[0035] For a given specification and state trajectory y t:t+k , smooth wTLTL robustness The definition is as follows:

[0036]

[0037]

[0038]

[0039]

[0040]

[0041]

[0042]

[0043]

[0044]

[0045]

[0046] Where Φ = ∧ ω φ i ,Ψ=∨ ω ψ i , ρ ω (y t:t+k ,φ is the wTLTL robustness, given the specification φ and trajectory yt:t+k , wTLTL robustness is defined as follows:

[0047]

[0048] ρ ω (y t:t+k f(y t ) <c)=c-f(y t ),

[0049]

[0050]

[0051]

[0052] in, is the maximum robustness value, ρ(y t:t+k ,φ i ),i+1,…,N is the standard TLTL robustness; and are the conjunction and disjunction aggregation functions, respectively, satisfying and And x≠0, specifically defined as:

[0053]

[0054]

[0055] in, is the normalized weight; with strict positive robustness ρ ω It shows that it satisfies the formula φ and has strict negative robustness ρ ω Then the formula φ is violated, that is, if ρ ω (y t:t+k ,φ)>0, then If ρ ω (y t:t+k ,φ)<0, then

[0056] and is a smoothed form of the traditional min and max operators, and The following design methods can be used:

[0057]

[0058]

[0059] Among them, k1, k2>0 are adjustable parameters;

[0060] 3. Evaluation, use the normalized exponential function to calculate the weight of M groups of samples The weight of the mth item is P m ∈[0,1]:

[0061]

[0062]

[0063] in h is a parameter; if h is large, only a few samples contribute to the average weight; if h = 0, no learning is possible;

[0064] 4. Update, update the parameters <θ,Σ> to <θ by weighted average new ,Σ new >:

[0065] Σ new ←boundcovar(Σ,λ min ,λ max )

[0066]

[0067]

[0068] Among them, boundcovar(Σ,λ min ,λ max ) is a restriction in [λ min ,λ max ] is a function in the covariance matrix Σ, with eigenvalue λ, if λ<λ min , then λ=λ min ; if λ>λ max , then λ=λ max ; Low-cost samples contribute more to the update process because they have higher weights, making θ move towards the minimum value θ * Move, and finally obtain the optimal shape parameters, and then generate the optimal path that meets the task description and conforms to the user's preferences based on DMPs.

[0069] In another aspect, the present invention further discloses a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the steps of the above method.

[0070] On the other hand, the present invention further discloses a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the above method.

[0071] As can be seen from the above technical solutions, the method and system for realizing multi-task motion of robots based on temporal logic of the present invention expands the traditional TLTL to wTLTL (Weighted truncated linear temporal logic, hereinafter referred to as wTLTL), and combines DMPs and wTLTL to realize complex operation tasks with user preferences. The DMPs are used to approximate the trajectory with user preferences. The trajectory shape is determined by the DMPs parameters and is optimized by the black box optimization strategy based on the weighted average covariance matrix (Policy Improvement with Black-Box optimization, hereinafter referred to as PI). BB ) for optimization. The wTLTL provides interpretable task specifications and is able to design a cost function with user preferences based on quantitative semantics, using the improved PI BB The algorithm can identify the optimal shape parameters of DMPs and realize the motion trajectory of complex operation tasks. The user preference is reflected by adjusting the weight in wTLTL.

[0072] In summary, the temporal logic-based robotic multitask motion planning method proposed in this paper encodes complex task specifications into interpretable wTLTL formulas, providing a convenient, effective, and interpretable approach for designing user-preferred cost functions. Applying this cost function to the parameter optimization algorithm for DMPs can generate user-preferred multitask trajectories while reducing computational complexity and improving the efficiency of multitask motion planning.

[0073] Specifically, the advantages and positive effects of the present invention are: encoding complex operation tasks into wTLTL formulas can not only provide interpretable task specifications, but also provide a more convenient, effective and user-preferred cost function design method. BB The proposed algorithm is suitable for identifying the optimal shape parameters of DMPs and thus enabling motion planning for complex manipulation tasks. Compared with traditional multi-task motion planning methods, this method can quickly plan motion trajectories for complex manipulation tasks with user preferences, with low computational complexity and high planning efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1 A schematic diagram of the architecture of the robot multi-task motion planning method based on temporal logic provided by the present invention;

[0075] Figure 2 The improved PI provided by the present invention BB A visualization of an optimization algorithm performing a parameter update in a two-dimensional search space. DETAILED DESCRIPTION

[0076] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments.

[0077] A robot multi-task motion planning method based on temporal logic according to an embodiment of the present invention has a schematic diagram of its architecture as shown in FIG. Figure 1 As shown in the paper, the traditional TLTL is extended to wTLTL, and DMPs and wTLTL are combined to achieve complex operation tasks with user preferences.

[0078] The DMPs are used to approximate trajectories with user preferences, and the trajectory shape is determined by the DMPs parameters and optimized by a black-box optimization strategy based on the weighted average covariance matrix.

[0079] The wTLTL provides interpretable task specifications and is able to design a cost function with user preferences based on quantitative semantics, using the improved PI BB algorithm, and then identify the optimal shape parameters of DMPs to achieve the motion trajectory of complex manipulation tasks.

[0080] The user preference is reflected by adjusting the weight in wTLTL.

[0081] The following are specific instructions:

[0082] The robot multi-task motion planning method based on temporal logic comprises the following steps:

[0083] Step 1: Establish a DMPs approximate trajectory model based on the robot kinematic model;

[0084] Step 2: Encode the complex operation task specifications into an interpretable wTLTL formula and add user preferences by setting wTLTL weights;

[0085] Step 3: Design a cost function with user preference based on the quantitative semantics provided by the wTLTL formula, and based on the improved PI BB The algorithm optimizes the shape parameters of DMPs and generates multi-task operation trajectories with user preferences.

[0086] The step 1 is specifically as follows:

[0087] According to the robot kinematic model, the DMPs approximate trajectory model is established:

[0088]

[0089]

[0090]

[0091]

[0092] where is the position, is the speed, and are the internal states. α z , β z , τ and α s are the proportionality factors, θ is the parameter vector, and g is the target position. The non - linear function f t allows the generation of arbitrarily complex motions, which are composed of basis functions g and are represented as piece - wise linear functions with weighted Gaussian kernels as:

[0093]

[0094]

[0095] where, [g t i records the i - th term of g t , σ i and c i are the variance and the mean. The core idea is to perturb α by the non - linear term z (β z (g - y t ) - z t ) to obtain smooth motions of arbitrary shapes.

[0096] The specific content of Step 2 is as follows:

[0097] Encode the complex operation task specification into an interpretable wTLTL formula. Among them, the semantics of wTLTL are defined as:

[0098]

[0099] T is the boolean value true, f(y t ) < c is a predicate, where f: maps the system state y t to a constant, (negation), ∨ (disjunction), ∧ (conjunction) are standard boolean operators, (finally), □ (always), (until), (then), 〇 (next) and (if... then...) are operators.

[0100] ​When the weight vector of the Boolean operator is a unit vector, i.e. ω i =1, Omit ω from the wTLTL formula. When all ω are unit vectors, traditional TLTL can be viewed as a special case of wTLTL.

[0101] Add user preferences by setting wTLTL weights. Where ω represents a positive weight vector, conjunction and disjunction are related to the i-th item ω i Indicates the relative importance of the defined specifications or alternatives priorities, ω i The higher the value, the higher its importance or priority. The semantics of the wTLTL formula is a finite trajectory on the system state, let y t:t+k Indicates that from y t to y t+k A series of states, if y t:t+k Satisfying the wTLTL formula φ, then

[0102] The step three is specifically as follows:

[0103] Improved PI BB The algorithm is shown in the following table:

[0104]

[0105]

[0106] According to the improved PI BB Algorithm, the shape parameters of DMPs will be updated according to the following steps:

[0107] 1. Initialization, set the initial mean and covariance <θ, Σ> to <θ init ,λ init I>, where I is an identity matrix of appropriate dimension;

[0108] 2. Sampling, through Sample M sets of parameter vectors θ from N(θ,Σ) m ,m=1…M. Where∈ m It is obtained by sampling from a zero-mean Gaussian distribution with variance Σ.

[0109] Calculate the cost function for each set of samples:

[0110]

[0111]

[0112] For a given specification and state trajectory y t:t+k , smooth wTLTL robustness The definition is as follows:

[0113]

[0114]

[0115]

[0116]

[0117]

[0118]

[0119]

[0120]

[0121]

[0122]

[0123] Where Φ = ∧ ω φ i ,Ψ=∨ ω ψ i ,ρ ω (y t:t+k ,φ) is the wTLTL robustness, given the specification φ and trajectory y t:t+k , wTLTL robustness is defined as follows:

[0124]

[0125] ρ ω (y t:t+ kf(y t ) <c)=c-f(y t ),

[0126]

[0127]

[0128]

[0129] in, is the maximum robustness value, ρ(y t:t+k ,φ i ),i+1,…,N is the standard TLTL robustness. and are the conjunction and disjunction aggregation functions, respectively, satisfying and And x≠0, specifically defined as:

[0130]

[0131]

[0132] in, is the normalized weight. With strict positive robustness ρ ω It shows that it satisfies the formula φ and has strict negative robustness ρ ω Then the formula φ is violated, that is, if ρ ω (y t:t+k ,φ)>0, then If ρ ω (y t:t+k ,φ)<0, then

[0133] and is a smoothed form of the traditional min and max operators, and The following design methods can be used:

[0134]

[0135]

[0136] Among them, k1, k2>0 are adjustable parameters.

[0137] 3. Evaluation, use the normalized exponential function to calculate the weight of M groups of samples The weight of the mth item is P m ∈[0,1]:

[0138]

[0139]

[0140] in h is a parameter. If h is large, only a few samples contribute to the average weight. If h = 0, no learning occurs. Generally speaking, low-cost samples have higher weights, and vice versa.

[0141] 4. Update, update the parameters <θ,Σ> to <θ by weighted average new ,Σ new >:

[0142] Σ new ←boundcovar(Σ,λ min ,λ max )

[0143]

[0144]

[0145] Among them, boundcovar(Σ,λ min ,λ max ) is a restriction in [λ min ,λ max ] is a function in the covariance matrix Σ, with eigenvalue λ, if λ<λ min , then λ=λ min ; if λ>λ max , then λ=λ max Low-cost samples contribute more to the update because they have higher weights, pushing θ towards the minimum θ * Move, and finally get the optimal shape parameters, and then generate the optimal path that meets the task description and user preferences according to DMPs. According to the above steps 1 to 4, improve PI BB The visualization diagram of the optimization algorithm performing a parameter update in the two-dimensional search space is as follows: Figure 2 shown.

[0146] The robot multi-task motion planning method based on temporal logic of the present invention is used to solve the problem of fast motion planning in robot multi-task operation.

[0147] The advantages and positive effects of the present invention are: encoding complex operation tasks into wTLTL formulas can not only provide interpretable task specifications, but also provide a more convenient, effective and user-preferred cost function design method. BB The proposed algorithm is suitable for identifying the optimal shape parameters of DMPs and thus enabling motion planning for complex manipulation tasks. Compared with traditional multi-task motion planning methods, this method can quickly plan motion trajectories for complex manipulation tasks with user preferences, with low computational complexity and high planning efficiency.

[0148] In another aspect, the present invention further discloses a computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor executes the steps of the above method.

[0149] On the other hand, the present invention further discloses a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the above method.

[0150] It is understandable that the system provided by the embodiment of the present invention corresponds to the method provided by the embodiment of the present invention, and the explanation, examples and beneficial effects of the relevant contents can refer to the corresponding parts of the above method.

[0151] The embodiment of the present application further provides an electronic device, comprising a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other via the communication bus.

[0152] Memory for storing computer programs;

[0153] A processor, configured to implement the aforementioned method for realizing multi-task motion of a robot based on sequential logic when executing a program stored in the memory;

[0154] The communication bus mentioned in the above electronic device can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The communication bus can be divided into an address bus, a data bus, a control bus, etc.

[0155] The communication interface is used for communication between the above electronic device and other devices.

[0156] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage. Alternatively, the memory may be at least one storage device located remote from the processor.

[0157] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0158] In another embodiment provided by the present application, a computer program product including instructions is also provided, which, when executed on a computer, enables the computer to execute any one of the methods for realizing multi-tasking motion of a robot based on sequential logic in the above embodiments.

[0159] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When software is used for implementation, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from a website, computer, server or data center to another website, computer, server or data center via a wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) method. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more available media integrations. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)).

[0160] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply the existence of any such actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or device comprising the element.

[0161] Each embodiment in this specification is described in a related manner. Similar parts between the various embodiments can be referred to in conjunction with each other. Each embodiment focuses on the differences between the other embodiments. In particular, the system embodiment is generally similar to the method embodiment, so the description is relatively simple. For related parts, refer to the description of the method embodiment.

[0162] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for realizing multi-task motion of a robot based on temporal logic, characterized in that: The following steps are included: Step 1: Establish the robot's trajectory kinematic model and the DMPs approximate trajectory model; Step 2: Encode the complex operation task specifications as an interpretable wTLTL formula and add user preferences by setting wTLTL weights; Step 3: Design a cost function with user preferences based on the quantitative semantics provided by the wTLTL formula, and BB Algorithms,optimize the shape parameters of DMPs and generate multi-task operation,trajectory with user preferences; The step 1 is specifically as follows: According to the robot kinematic model, the DMPs approximate trajectory model is established: in, For location, For speed, and is the internal state; α z ,β z ,τ and α s is a positive proportional factor, θ is a parameter vector, g is the target position; nonlinear function f t Arbitrary complex motion can be generated, which consists of basis functions g, represented by piecewise linear functions with weighted Gaussian kernels as: Among them, [g t ] i Recorded g t The i-th term of i and c i is the variance and mean, and the core idea is to use nonlinear terms Perturbation α z (β z (gy t )-z t ) to obtain smooth motion of arbitrary shapes; The step 2 is specifically as follows: The complex operation task specification is encoded into an interpretable wTLTL formula; where the semantics of wTLTL is defined as: T is a Boolean true value, f(y t ) < c is a predicate, where maps the system state y t to a constant, (not), (disjunction), ∧ (conjunction) are standard Boolean operators, (eventually), □ (always), (until), (then), are operators; When the weight vector of a Boolean operator is a unit vector, that is, Omit ω from the wTLTL formula; when all ω are unit vectors, traditional TLTL is considered as a special case of wTLTL; User preferences are added by setting the wTLTL weights, where ω represents a positive weight vector, the conjunction and disjunction of the i-th item ω i Indicates the relative importance of the defined specifications or alternatives, ω i The higher the value, the higher its importance or priority; the semantics of the wTLTL formula is a finite trajectory on the system state, let y t:t+k Indicates that from y t to y t+k A series of states, if y t:t+k Satisfying the wTLTL formula φ, then The step three is specifically as follows: Based on improved PI BB Algorithm, the shape parameters of DMPs are updated according to the following steps:

1. Initialization, set the initial mean and covariance <θ, Σ> to <θ init ,λ initI >, where I is an identity matrix of suitable dimension; 2. Sampling, through Sample M sets of parameter vectors θ from N(θ,Σ) m ,m=1…M; where∈ m Obtained by sampling from a zero-mean Gaussian distribution with variance Σ; Calculate the cost function for each set of samples: For a given specification and state trajectory y t:t+k , smooth wTLTL robustness The definition is as follows: Where Φ = ∧ ω φ i ,Ψ=∨ ω ψ i , ρ ω (y t:t+k ,φ) is the robustness of wTLTL, given the specification φ and trajectory y t:t+k , wTLTL robustness is defined as follows: ρ ω (y t:t+k f(y t )<c)=c-f(y t ), in, is the maximum robustness value, ρ(y t:t+k ,φ i ),i+1,…,N is the standard TLTL robustness; and are conjunctive and disjunctive aggregate functions, respectively, satisfying and And x≠0, specifically defined as: in, is the normalized weight; it has strict positive robustness ρ ω It shows that it satisfies the formula φ and has strict negative robustness ρ ω Then the formula φ is violated, that is, if ρ ω (y t:t+k ,φ)>0, then If ω (y t:t+k ,φ)<0, then and is a smoothed form of the traditional min and max operators, and The following methods can be used for design: Among them, k1, k2>0 are adjustable parameters; 3. Evaluation, using the normalized exponential function to calculate the weights of M groups of samples The mth item weight P m ∈[0,1]: in h is a parameter term; if a larger h value is taken, only a few samples will contribute to the average weight; if h = 0, no learning can be performed; 4. Update, update the parameters <θ,Σ> to <θ by weighted averaging new ,Σ new >: S new ←boundcovar(Σ,λ min ,l max ) Among them, boundcovar(Σ,λ min ,λ max ) is a restriction in [λ min ,λ max ] is a function in the covariance matrix Σ, with eigenvalue λ, if λ<λ min , then λ=λ min ; if λ>λ max , then λ=λ max ; Low-cost samples contribute more to the update process because they have higher weights, making θ move towards the minimum value θ * Move, and finally get the optimal shape parameters, and then generate the optimal path that meets the task description and meets the user's preferences according to DMPs.

2. The method for realizing multi-task motion of a robot based on sequential logic according to claim 1, characterized in that: Improved PI for wTLTL boot BB The algorithm is as follows:

3. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the processor is caused to perform the steps of the method according to any one of claims 1 to 2.

4. A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method according to any one of claims 1 to 2.

Images