A motion prediction method for multi-degree-of-freedom robots

Abstract

The present invention discloses a motion prediction method for a multi-degree-of-freedom robot. In order to solve the problem of delayed transmission of robot motion information caused by communication delay, the method of the present invention can predict the current actual joint position and speed of the robot in real time based on the delayed robot joint position and speed signals. First, according to the size of the delay value, multiple robot motion sub-states are established; then, the same number of secondary predictors are constructed; finally, they are connected in series to form a series predictor, in which each secondary predictor will obtain the predicted value of the corresponding sub-state, and the predicted value of the actual joint position and speed is generated by the last secondary predictor. By using a sufficient number of secondary predictors, the present method can improve the prediction accuracy of the robot motion state under a larger communication delay. In addition, the structure of each secondary predictor is the same, which is convenient to implement and has certain application prospects in the field of remote-controlled robots.

Description

Technical Field

[0001] The present invention belongs to the field of robot sensing and control, and in particular relates to a motion prediction method for a multi-degree-of-freedom robot. Background Art

[0002] Robots can replace humans in dangerous and harsh environments, thereby extending the scope of human activity. They are widely used in fields such as space and ocean exploration, telemedicine, and hazardous materials handling. With the development of communication technology, remote-controlled robotics has been widely researched and applied. However, with the long control distances, communication latency has become an unavoidable and prominent problem. Delayed information transmission not only reduces the operational performance of the robotic system but can even lead to system instability. Therefore, it is necessary to predict the robot's motion state to compensate for the impact of communication latency. However, existing motion prediction methods are only applicable to linear systems or situations with relatively small latency and cannot meet the application requirements of multi-degree-of-freedom robots and long-distance communication. Summary of the Invention

[0003] Purpose of the invention: The purpose of the present invention is to provide an effective motion prediction method for a multi-degree-of-freedom robot, that is, to predict the current actual joint position and speed of the robot in real time based on the delayed robot joint position and speed signals.

[0004] Technical solution: The motion prediction method of a multi-degree-of-freedom robot of the present invention is characterized in that, based on the delayed signals of the robot's joint positions and joint velocities, the current actual joint positions and actual joint velocities of the robot are predicted in real time by executing the following steps:

[0005] Step 1: Determine the gain parameter K of the secondary predictor based on the error e1 between the initial value of the robot joint position and the given initial value of the position prediction, and the error e2 between the initial value of the joint velocity and the given initial value of the joint velocity prediction;

[0006] Step 2: Determine the number m of robot motion sub-states that need to be established based on the gain parameter K of the secondary predictor and the measured communication delay value τ, and establish m motion sub-states;

[0007] Step 3: Based on the established m motion sub-states, construct secondary predictors with the same number as the motion sub-states;

[0008] Step 4: Connect each secondary predictor in series one by one. The output of the previous secondary predictor is used as the input of the next secondary predictor. The delayed signal of the robot's joint position and joint velocity is input to the first secondary predictor, and the predicted value of the robot's actual joint position and actual joint velocity is output by the last secondary predictor.

[0009] Furthermore, in step 1, the value range of the gain parameter K of the secondary predictor is determined by the following formula:

[0010]

[0011] Among them, e1 is the error between the initial joint position of the robot and the given initial position prediction value, and e2 is the error between the initial joint velocity of the robot and the given initial velocity prediction value.

[0012] Furthermore, in step 2, the number m of robot motion sub-states that need to be established is determined by the following formula:

[0013]

[0014] Where τ is the value of the communication delay and K is the gain parameter of the secondary predictor.

[0015] Furthermore, step 3 is as follows: T i (t) = [x 1i (t)x 2i (t)] represents m motion sub-states, where i = 1, 2, ..., m is the sub-state index, and the following two formulas are used to establish the element x in each sub-state respectively. 1i (t) and x 2i (t)

[0016]

[0017]

[0018] Where x1(t) and x2(t) represent the joint position and joint velocity of the multi-degree-of-freedom robot, respectively, and t is the time variable.

[0019] Furthermore, in step 3, the secondary predictor is constructed by a prediction algorithm represented by the following m equations:

[0020]

[0021] in and Represents the prediction result of the robot's motion sub-state. The prediction results of the robot's actual joint position and joint velocity will be given by the mth secondary predictor, that is, and f(·) is the known robot dynamics equation function, u is the robot's control input, r i (t) is the error calibration item, and its specific calculation method is:

[0022]

[0023] Where α is another gain parameter of the predictor, and the value range of α is 0.5<α<1.5, y 1i (t) and y 2i (t) is the input signal of each secondary predictor, when i=1, y 1i =x1(t-τ),y 2i =x2(t-τ), the delayed signal of the robot joint position and joint velocity; when i=2,3,...,m, That is, the output signal of the previous secondary predictor.

[0024] Beneficial Effects: Compared to existing technologies, this invention offers the following significant advantages: It utilizes multiple secondary predictors connected in series, enabling application with relatively long delays of approximately 4 seconds, thus handling a wider range of communication delays than conventional methods. Furthermore, this method is applicable to nonlinear multi-degree-of-freedom robots and offers high prediction accuracy. Furthermore, the identical structure of each secondary predictor facilitates implementation and scalability across diverse robotic systems, contributing to the development of networked robotic applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 Schematic diagram of the overall system of the robot motion prediction method of the present invention;

[0026] Figure 2 Schematic diagram for the construction of a single secondary predictor. DETAILED DESCRIPTION

[0027] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0028] The present invention addresses the problem of delayed transmission of robot motion state information caused by communication delays. Based on the nonlinear dynamic model of a multi-degree-of-freedom robot, a prediction method for the robot motion state (including joint position and velocity) is designed. By connecting multiple secondary predictors in series, this method can accurately predict the current actual joint position and velocity of the robot in real time based on the delayed robot joint position and velocity signals under large time delays. Figure 1 The prediction results help the operator to grasp the robot's motion status in a timely manner, and design control algorithms and issue instructions accordingly, thereby compensating for communication delays.

[0029] The specific implementation method of the present invention is described below with reference to the accompanying drawings.

[0030] First, based on the error e1 between the robot's initial joint position and the given initial position prediction value, and the error e2 between the initial joint velocity and the given initial velocity prediction value, the value range of the predictor gain K is determined according to the following formula:

[0031]

[0032] The above formula can be used to determine the minimum value that K must meet. Selecting a larger K value can reduce the impact of unknown system dynamics on the prediction results and reduce steady-state prediction errors. However, selecting a value that is too large may cause significant overshoot in the prediction results. Generally, K should not exceed 20.

[0033] The number m of robot motion sub-states that need to be established can be determined by the following formula:

[0034]

[0035] Where τ is the communication delay, which can be calculated by comparing the timestamps of signal transmission and reception. To simplify the designed prediction algorithm, m can be the smallest integer that satisfies this equation, or the smallest integer plus 1. For example, when K = 1 and τ = 4 seconds, m = 4 can be used.

[0036] Then, use x T i (t) = [x 1i (t) x 2i (t)] represents the m sub-states to be established, where i = 1, 2, ..., m is the sub-state index, then the element x in each sub-state vector is 1i (t) and x 2i (t) is given by the following formula:

[0037]

[0038]

[0039] Where x1(t) and x2(t) represent the joint positions and velocities of the multi-DOF robot, respectively, and t is the time variable. It's important to note that these two equations represent the physical meaning of each sub-state, not the calculation method. Once a sub-state is established, its value does not need to be calculated.

[0040] Secondly, m secondary predictors are constructed according to the following prediction algorithm expressed as a set of m equations:

[0041]

[0042] in and Represents the prediction result of the robot's motion sub-state, and the prediction results of the robot's actual joint position and velocity are given by the mth secondary predictor, that is, and f(·) is the known robot dynamics equation function, u is the robot's control input, r i(t) is the error calibration item, and its specific calculation method is:

[0043]

[0044] Where α is another gain parameter of the predictor, which is usually 0.5<α<1.5. 1i (t) and y 2i (t) is the input signal of each secondary predictor, when i=1, y 1i =x1(t-τ),y 2i =x2(t-τ), which is the robot joint position and velocity signal with time lag; when i=2,3,...,m, That is, the output signal of the previous secondary predictor.

[0045] As can be seen from the above formula, each secondary predictor has the same structure except for the different input signals. The input of the first secondary predictor is the delayed robot joint position and velocity signal, while the subsequent secondary predictors use the output of the previous secondary predictor as input. According to the above algorithm, its construction principle diagram is shown in the attached figure. Figure 2 As shown in Figure 2, constructing a secondary predictor requires: a gain module, an integration module, a summation (difference) module, and a delay module.

[0046] Finally, the secondary predictors are connected in series one by one, that is, the output of the previous secondary predictor is used as the input of the next one, thus forming a series predictor, as shown in the attached figure. Figure 1 The method of the present invention can predict the actual motion state of the robot based on the time-delayed robot motion information, namely the joint position and velocity, and the prediction result will be obtained from the last secondary predictor.

Claims

1. A motion prediction method for a multi-degree-of-freedom robot, characterized in that: According to the delayed signals of the robot's joint positions and joint velocities, the robot's current actual joint positions and actual joint velocities are predicted in real time by executing the following steps: Step 1: Determine the gain parameter K of the secondary predictor based on the error e1 between the initial value of the robot joint position and the given initial value of the position prediction, and the error e2 between the initial value of the joint velocity and the given initial value of the joint velocity prediction; In step 1, the value range of the gain parameter K of the secondary predictor is determined by the following formula: Wherein, e1 is the error between the initial joint position of the robot and the given initial position prediction value, and e2 is the error between the initial joint velocity of the robot and the given initial joint velocity prediction value; Step 2: Determine the number m of robot motion sub-states that need to be established based on the gain parameter K of the secondary predictor and the measured communication delay value τ, and establish m motion sub-states; In step 2, the minimum number m of robot motion sub-states that need to be established satisfies: Where τ is the value of the communication delay, K is the gain parameter of the secondary predictor; Step 3: Based on the established m motion sub-states, construct secondary predictors with the same number as the motion sub-states; Step 4: Connect each secondary predictor in series one by one. The output of the previous secondary predictor is used as the input of the next secondary predictor. The delayed signal of the robot's joint position and joint velocity is input to the first secondary predictor, and the predicted value of the robot's actual joint position and actual joint velocity is output by the last secondary predictor.

2. The motion prediction method of a multi-degree-of-freedom robot according to claim 1, characterized in that: Step 3 is as follows: T i (t) = [x 1i (t)x 2i (t)] represents m motion sub-states, where i = 1, 2, ..., m is the sub-state index, and the following two formulas are used to establish the element x in each sub-state respectively. 1i (t) and x 2i (t) Where x1(t) and x2(t) represent the joint position and joint velocity of the multi-degree-of-freedom robot, respectively, and t is the time variable.

3. The motion prediction method of a multi-degree-of-freedom robot according to claim 1, characterized in that: In step 3, the secondary predictor is constructed by a prediction algorithm represented by the following set of m equations: in and Represents the prediction result of the robot's motion sub-state. The prediction results of the robot's actual joint position and joint velocity will be given by the mth secondary predictor, that is, and f(·) is the known robot dynamics equation function, u is the control input of the robot; r i (t) is the error calibration item, and its specific calculation method is: Where α is another gain parameter of the predictor, and the value range of α is 0.5<α<1.5, y 1i (t) and y 2i (t) is the input signal of each secondary predictor, when i=1, y 1i =x1(t-τ),y 2i =x2(t-τ), which are the delayed signals of the robot joint position and joint velocity respectively; when i=2,3,...,m, That is, the output signal of the previous secondary predictor.

Images