Method, computer program, and apparatus for controlling plurality of joints of robot based on desired joint state

By using feasibility filters to filter the expected joint state in the robot system, the problem of poor modulation in the prior art is solved, and kinetic and dynamic constraints are met, and the infeasible states are avoided, which improves the robustness of the robot system is improved. and track performance.

CN118922279APending Publication Date: 2024-11-08SONY GROUP CORP
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202380029694.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2022-03-30
Filing Date
2023-03-23
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

The prior art is difficult to effectively modulate the desired joint state in robotic systems to satisfy kinematic and dynamic constraints, especially in the face of inaccurate sensing and dynamic disturbances, resulting in unsolid motion planning and may cause actuation. Device damage and poor tracking performance.

Method used

Using feasibility filters, the expected joint state is filtered to conform to kinematic and dynamic constraints, reduce the difference between the expected state and the actual state, avoid the ultimate infeasible state, consider joint coupling effects and nonlinear dynamics, and use Savitzky-Golay The algorithm performs state estimation and prediction.

Benefits of technology

It realizes efficient modulation of the desired joint state in a real-time environment, so that it meets the kinematic and dynamic constraints of the actuator, avoids infeasible states, improves the robustness and tracking performance of the robot system, and ensures the safety of the actuator. Sex and task success rate.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118922279A_ABST
    Figure CN118922279A_ABST
Patent Text Reader

Abstract

Examples relate to methods, computer programs, and apparatus for controlling a plurality of joints of a robot based on a desired joint state. The method for controlling a plurality of joints of a robot based on a desired joint state includes: obtaining information about the desired joint state; obtaining information about a current joint state; filtering the desired joint state based on the information about the current joint state and based on a state constraint of the robot to obtain information about a filtered desired joint state, where the filtered desired joint state conforms to the state constraint; and providing information about the filtered desired joint state to control a joint of the robot.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] Examples relate to methods, computer programs, and apparatus for controlling multiple joints of a robot based on desired joint states, and more particularly, but not exclusively, to the concept of real-time feasibility filters for rotary and prismatic actuators in robotic systems. Background Art

[0002] Robotic systems are used to complete complex movements and need to operate effectively. Robots can include multiple joints that are coordinated during operation, such as prismatic joints and revolute joints. In robotic systems, many dynamic applications (such as reaching, grabbing, and placing objects) require accurate motion planning and dynamic motor control. It is important to design motion planning strategies to provide stronger robustness in the face of imprecise sensing and braking. To this end, the motion planner can generate the desired robot motion relative to the latest state of the environment. In addition, the generated motion must be feasible for the robot to follow. This feasibility is defined at the kinematic level (e.g., position and velocity) and the dynamic level (e.g., acceleration and jerk), and therefore the generated motion must focus on these two levels to be called feasible. Operating an actuator beyond its limit will cause irreversible damage to the actuator itself and the environment. Due to the limitations of available computing resources, the kinematic and dynamic limitations of the actuator cannot always be solved at the motion planning level. In addition, due to the jerk limit, the position, velocity and acceleration cannot be changed immediately, so it may not be enough to meet the state constraints under each level separately. Summary of the invention

[0003] Examples are based on the discovery that a feasibility filter can be implemented that allows filtering a desired state to comply with given kinematic and dynamic constraints. Examples can provide a computationally efficient method that includes modulating a desired robot motion in a manner that satisfies kinematic and dynamic constraints of an actuator. The method can be integrated into existing motion control methods and can modulate a desired robot motion in a dynamic environment where the robot must react to dynamic perturbations on-line.

[0004] The example relates to a method for controlling multiple joints of a robot based on desired joint states. The method includes obtaining information about desired joint states and obtaining information about current joint states. The method also includes filtering the desired joint states based on the information about the current joint states and based on the state constraints of the robot to obtain information about the filtered desired joint states. The filtered desired joint states conform to the state constraints. The method also includes providing information about the filtered desired joint states to control the joints of the robot. The filtering of the desired joint states can be processed in real time or near real time to obtain the filtered desired joint states that conform to the state constraints of the robot.

[0005] The filtering of the desired state can be based on reducing the difference between the filtered desired state and the desired state taking into account the state constraints. Thus, the deviation between the desired state and the actual state or the current state can be effectively reduced.

[0006] For example, filtering can be based on minimizing the difference between the filtered desired state and the desired state considering the state constraints. Thus, the filtered desired state can be optimal considering the desired state of the robot given the actual state and within the boundaries given by the constraints.

[0007] In a further example, the difference may be determined by a sum or weighted sum of differences between the desired joint state and the filtered desired joint state in one or more of the location or position of the joint, the velocity of the joint, the acceleration of the joint, the jerk of the joint, and / or one or more higher order derivatives of the location or position. Examples may consider kinematic constraints and dynamic constraints.

[0008] For example, the difference can be the mean square error or weighted mean square error between the filtered desired state and the desired state taking into account the state constraints.Examples can implement a mean square error filter for the desired joint state and thereby provide an optimal or near-optimal filtered joint state for a given actual state and state constraints.

[0009] In an implementation, some examples may use filtering that includes determining filtering rules for each of the multiple joints separately. This may enable parallel filtering for multiple joints.

[0010] In addition, at least in some examples, the filtering can also be based on avoiding the final infeasible state of the robot. The final infeasible state is a state that may not violate any constraints at the moment, but due to the kinematic state and the dynamic state, the infeasible state will be generated in the future. An example of a final infeasible state is a state where the speed is too high at a given position to avoid violating the position constraint even when the maximum deceleration is applied. Thus, the example can consider the future state constraints when determining the current state. The avoidance of the final infeasible state can be based on analyzing the impact of the expected jerk after filtering on the future state. In this way, the example can also achieve future state constraint compliance.

[0011] In further examples, filtering may also be based on coupling effects between joints of the robot and / or nonlinear dynamics of joints of the robot.Thus, examples may also take into account mutual coupling effects or nonlinearities between joints of the robot.

[0012] The current joint state, the desired joint state and the filtered desired joint state may include one or more of the information about the location or position of the joint, the speed of the joint, the acceleration of the joint, the jerk of the joint, and / or one or more high-order derivatives of the location or position. The example may provide the kinematic desired state and the dynamic desired state after filtering. Therefore, the current joint state and the filtered desired joint state may include information about the location or position of the joint, the speed of the joint, the acceleration of the joint and the jerk of the joint. In some examples, the desired joint state may include three or less elements in the group of information about the location or position of the joint, the speed of the joint, the acceleration of the joint, the jerk of the joint, and / or one or more high-order derivatives of the location or position. Then, the method may also include predicting the desired joint state for subsequent time intervals based on the current joint state and the desired joint state. The desired joint state for subsequent time intervals may include prediction information about the location or position of the joint, the speed of the joint, the acceleration of the joint, the jerk of the joint, and / or one or more high-order derivatives of the location or position. Then, filtering may be based on the predicted desired joint state for subsequent time intervals.

[0013] Prediction of the desired joint state for a subsequent time interval may include using a parameter state estimate based on the desired joint state to obtain a predicted desired joint state for the subsequent time interval. Examples may provide an efficient way to predict the desired joint state. For example, the parameter state estimate may use a Savitzky-Golay algorithm.

[0014] In addition, in an example, filtering can be based on a state constraint, and the state constraint is based on the number of multiple subsequent time intervals. Thus, the example can allow a prediction range. The number of subsequent intervals can be configured to allow the movement of the joint to be stopped based on the maximum deceleration constraint of the joint. Therefore, the example can allow the movement within the prediction range to be stopped. For example, the number of time intervals can also depend on one or more of the location or position of the joint, the speed of the joint, the maximum allowed acceleration and / or jerk of the joint. Thus, in an example, kinematic constraints and dynamic constraints can determine the prediction range. Typically, the state constraints can include kinematic constraints and dynamic constraints of the joint.

[0015] Another example is a computer program with a program code for performing any of the methods described herein, when the computer program is executed on a computer, a processor or a programmable hardware component.

[0016] Another example is an apparatus for controlling an automated device, the automated device comprising a control unit for performing one of the methods described herein. Yet another example is an automated device, a robot, or a controller for a robot, such as an industrial robot comprising an example of the apparatus. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Some examples of apparatus and / or methods will be described below, by way of example only, and with reference to the accompanying drawings, in which:

[0018] Figure 1 a block diagram illustrating a flow diagram of an example of a method for controlling a plurality of joints of a robot based on desired joint states;

[0019] Figure 2 A block diagram showing an example of an apparatus for controlling a plurality of joints of a robot and an example of a robot;

[0020] Figure 3 Another example of a robot is shown;

[0021] FIG4 shows an example of a motion control architecture for controlling a robotic system;

[0022] Figure 5 Illustrative examples of final infeasible states using kinematic and kinetic diagrams are depicted;

[0023] FIG6 shows an example of the proposed motion control architecture using a feasibility filter;

[0024] Figure 7 A block diagram showing a flow diagram of an example of a feasibility filter;

[0025] Figure 8shows an example timeline of kinematic and dynamic parameters in a first scenario with three reference steps;

[0026] Fig. 9 shows an example timeline of kinematic and dynamic parameters in a second scenario with a sinusoidal reference; and

[0027] Fig.10 Example timelines of kinematic and dynamic parameters in the third scenario with step reference position constraints are shown. DETAILED DESCRIPTION

[0028] Some examples are now described in more detail with reference to the accompanying drawings. However, other feasible examples are not limited to the features of the embodiments described in detail. Other examples may include modifications of features and equivalents and substitutes of features. In addition, the terms used herein to describe some examples should not limit other feasible examples.

[0029] Throughout the description of the accompanying drawings, the same or similar reference numerals refer to the same or similar elements and / or features, which can be implemented in the same or modified form while providing the same or similar functions. For clarity, the thickness of the lines, layers and / or regions in the figures may also be exaggerated.

[0030] When two elements A and B are combined using "or", this should be understood to disclose all possible combinations, i.e. only A, only B, and A and B, unless otherwise explicitly defined in individual cases. As an alternative wording for the same combination, "at least one of A and B" or "A and / or B" can be used. The same applies to combinations of more than two elements.

[0031] If singular forms are used, such as "a", "an", and "the", and the use of only a single element is not explicitly or implicitly defined as mandatory, additional examples may also use several elements to achieve the same function. If a function is described below as being implemented using multiple elements, additional examples may use a single element or a single processing entity to achieve the same function. It should also be understood that the terms "comprises", "comprising", and / or "includes" when used describe the presence of specified features, integers, steps, operations, processes, elements, components, and / or groups thereof, but do not exclude the presence or addition of one or more other features, integers, steps, operations, processes, elements, components, and / or groups thereof.

[0032] Figure 1A block diagram of a flow diagram of an example of a method 10 for controlling multiple joints of a robot based on desired joint states is shown. The method 10 includes obtaining 12 information about desired joint states and obtaining 14 information about current joint states. The method 10 also includes filtering 16 the desired joint states based on the information about the current joint states and based on state constraints of the robot to obtain information about the filtered desired joint states. The filtered desired joint states comply with the state constraints. The method 10 also includes providing 18 information about the filtered desired joint states to control the joints of the robot.

[0033] Figure 2 Shown is a block diagram of an example of an apparatus 20 for controlling a plurality of joints of a robot and an example of a robot 200. The apparatus 20 for controlling a plurality of joints of a robot 200 includes a control module for performing one of the methods described herein. Figure 2 Also shown is an example of a robot 200 including an example of the apparatus 20 or a controller for the robot 200 (shown in dashed lines because it is optional from the perspective of the apparatus 20).

[0034] like Figure 2 As further shown in , the device 20 may include one or more interfaces 22 coupled to a corresponding control module or unit 24. One or more interfaces 22 may correspond to one or more inputs and / or outputs for receiving and / or transmitting information within a module, between modules, or between modules of different entities, and the information may be a digital (bit) value according to a specified code. For example, the interface 22 may include an interface circuit configured to receive and / or transmit information. In an example, the interface 22 may correspond to any device for obtaining, receiving, transmitting, or providing analog or digital signals or information, for example, any connector, contact, pin, register, input port, output port, conductor, channel, etc. that allows providing or obtaining signals or information. The interface 22 may be configured to communicate (transmit, receive, or both) wirelessly or wired, and may be configured to communicate (i.e., transmit and / or receive signals, information) with additional internal or external components. One or more interfaces 22 may include additional components for implementing communications in a mobile communication system, such components may include transceiver (transmitter and / or receiver) components, such as, one or more low noise amplifiers (LNA), one or more power amplifiers (PA), one or more duplexers, one or more diplexers, one or more filters or filtering circuits, one or more converters, one or more mixers, correspondingly adapted RF components, etc.

[0035] The control module 24 can be implemented using one or more processing units, one or more processing devices, any means for processing, such as a processor, a computer, or a programmable hardware component that can operate with correspondingly adapted software. In other words, the described functions of the control / processing module 24 can also be implemented in software, which is then executed on one or more programmable hardware components. Such hardware components may include general-purpose processors, digital signal processors (DSPs), microcontrollers, etc. Another example is an automated device 200 including an example of the device 20, such as an automated / autonomous vehicle or an industrial robot.

[0036] The example can implement a real-time feasibility filter for rotary actuators and prismatic actuators in a robotic system. In this specification, the following terms are used:

[0037] Task space: The state of the robot's end effector.

[0038] Joint space: the state of each joint of the robot.

[0039] State: includes, but is not limited to, position, velocity, acceleration, jerk, and / or one or more higher order derivatives of the location or position of an end effector or joint, respectively, in task or joint space.

[0040] Target state: The state the robot should be in at the target time.

[0041] Expected state: The state the robot should be in within a sampling time.

[0042] Motion Planner: An algorithm or set of algorithms that plans the robot's motion between its current state and a goal state.

[0043] Inverse Kinematics: An algorithm or set of algorithms that transforms task-space states into corresponding joint-space states.

[0044] Tracking errors: the difference between the desired state and the current state.

[0045] Infeasible State: A state in which one or more constraints are violated at the level of position, velocity, acceleration, jerk level, and / or one or more higher order derivatives of location or position.

[0046] Final infeasible state: A state that is currently feasible but will eventually violate a position, velocity, or acceleration constraint due to jerk restrictions.

[0047] Given a fixed goal state or a moving goal state, the goal of the motion planner is to calculate Figure 3 The robotic system illustrated in FIG. 1 is brought from its current state to an intermediate state of a target state in real time. Figure 3Another example of a robot 200 is shown. The example robot 200 includes a 6-degree-of-freedom robot system. Each joint is equipped with an actuator. Actuators M_1, M_5, and M_6 are prismatic, while M_2, M_3, and M_4 are rotational. The actuators are controlled at the position level independently of the other actuators. An embedded low-level controller converts the position command into a desired torque so that the actuator moves the joint to the desired position.

[0048] Planning the desired motion of a robotic system can be solved at two different levels, namely at the task space level or at the joint space level, which is further described in detail in Figure 4. Figure 4 shows an example of a motion control architecture for controlling a robotic system. At the top of Figure 4, a target state 41 is input, for example, an object position. At the top, the desired motion (desired task state) is generated at the task space level by a task space motion planner 42 relative to the target state and the current state. The desired task space values ​​are then converted to desired joint motions using inverse kinematics 43 based on the current joint states before being provided to the robot 44. The robot 44 uses a low-level controller 44a to generate desired joint torques for the actuators 44b of the robot 44.

[0049] FIG4 shows an alternative architecture for the bottom portion. At the bottom portion, the motion planner is designed to generate the desired joint states directly relative to the target state and the current state. The target state is provided to the inverse kinematics 43 to obtain the target joint state, which is then provided to the joint space motion planner 45, which generates the desired joint state based on the current joint state. The desired joint state is then provided to the robot 44. The robot 44 again uses the low level controller 44a to generate the desired joint torques for the actuators 44b of the robot 44.

[0050] To endow a robotic system with the ability to complete dynamic scenarios, the robot's motion should be updated regularly in a computationally efficient manner, and the desired state should satisfy the kinematic and dynamic constraints of the actuators before issuing commands to the robot. Therefore, (i) computational budget and (ii) constraint satisfaction at the current time step and future time steps are two key aspects of implementing dynamic scenarios.

[0051] The command motion should be generated so that it satisfies the state constraints and also avoids final infeasible states where violation of position or velocity constraints cannot be avoided by any feasible jerk value. Examples of such scenarios are Figure 5 shown. Figure 5 The kinematic and kinetic diagrams of the final infeasible state are depicted. Figure 5 Shows the joint position x51, velocity Acceleration and jerk The limits are indicated by the corresponding "minimum" and "maximum" thresholds.

[0052] In this example, the state is feasible at t_0. However, since the state at the position level is close to the upper limit, and the velocity and acceleration are also large positive values, there is no feasible jerk at t_1 to avoid violating the constraint. Figure 5 In the final infeasible state shown, position constraint violation cannot be avoided due to the jerk limit.

[0053] The commands should be updated regularly and in a computationally efficient way to reflect the latest changes in the environment. As an industrial application, one can imagine a kinetic pick and place scenario, where a robot has to reach an object moving on a conveyor, pick it up and place it at the desired location. If the robot's motion is not updated regularly and in a computationally efficient way, there is a high chance that the object will be missed and the task will not be completed. Another application could be human-robot collaboration in a warehouse, where a robot should move objects under the guidance of an operator. Due to the inherent characteristics of this scenario, the robot's motion cannot be planned in advance and needs to be updated regularly with respect to the operator's latest guidance.

[0054] Commanding desired states that violate these constraints can result in poor tracking performance and can even cause irreversible damage to the actuators due to the high torques generated by the low-level controller. Poor tracking performance can produce overshoot or undershoot. Overshoot can be dangerous because the robot may collide with the environment, for example, the robot may hit a conveyor in a kinetic pick and place scenario. Undershoot is also undesirable because the robot may miss reaching the moving target.

[0055] To solve this problem and avoid ill-defined and infeasible motions, the example can use a computationally efficient algorithm to modulate the desired joint states of the robot so that the robot satisfies the kinematic constraints and dynamic constraints of the actuators.

[0056] This can be done by Figure 1The filtering 16 in the embodiment can be implemented by a feasibility filter block, which is further described in detail in Figure 6. Figure 6 shows an example motion control architecture similar to Figure 4, having substantially the same components except for the addition of a feasibility filter 60, as an implementation of the above-mentioned device 20 for performing method 10. The feasibility filter 60 generates a filtered desired joint state based on the desired joint state and the current joint state. The filtered desired joint state is then provided to the robot 44. As described in more detail with respect to Figure 4, these two architectural variations are conceivable in the example. For example, the motion is modulated so that the difference between the desired state and the filtered desired state is minimized, while the filtered desired state satisfies all state constraints. In the example, the filtering 16 of the desired state can be based on reducing or even minimizing the difference between the filtered desired state and the desired state considering the state constraints.

[0057] Removing unwanted components or features from a signal has been extensively studied in the robotics and signal processing literature. However, the problem of modulating an online signal with a state constraint is a new area of ​​research. Linear filters (e.g., low-pass, high-pass, band-pass filters, and notch filters) have been widely used to remove a certain frequency component from a signal. These filters can be designed to achieve a maximally flat frequency response in Butterworth filters, the best approximation to the ideal response in Chebyshev filters, a maximally flat phase delay in Bessel filters, and a steepest cutoff of a specified order and ripple in elliptic filters. These filters can be used in analog systems and, due to computational efficiency, can also be used in real-time digital systems.

[0058] The second class of smoothing filters includes parametric filters, which are mainly used in digital signal processing applications. Savitzky-Golay, Kalman, and Wiener filters have been widely used to smooth, predict, and correct sensory signals in real time. The Savitzky-Golay algorithm filters the signal by fitting a low-degree polynomial to a continuous subset of adjacent data points. The Wiener filter smoothes the signal by minimizing the mean square error between the original and filtered signals. The Kalman filter is a group of algorithms that provide accurate estimates of a single measurement value by using the dynamical model of the system and the statistical noise. Even though these filters are computationally efficient and can be used in real-time applications, unfortunately, these filters cannot guarantee that the smoothed signal satisfies the state constraints at the current state, let alone avoid the final infeasible state.

[0059] Another approach to addressing the feasibility of the generated motions is to design motion planning algorithms that can directly consider state constraints when computing new intermediate states. Model predictive control (MPC) and optimal control are types of motion planning algorithms for robotic systems that explicitly consider the constraints imposed on the system by solving constrained optimization problems. Constraints include, but are not limited to, kinematic constraints and dynamic constraints. In addition, harmful interactions of the system with its environment can be avoided by defining collision constraints. MPC and optimal control are powerful techniques for solving motion planning problems while ensuring the feasibility of the motions. However, depending on the size and complexity of the underlying optimization problem, significant computational resources are required and the solver time may be too long to be used for real-time closed-loop feedback control of the dynamic scenarios under consideration.

[0060] The example can take advantage of the above algorithm to provide a computationally efficient method to modulate the desired joint state so that it satisfies the kinematic and dynamic constraints of the actuator in a real-time application. To reduce computation time, the example can decouple motion planning from filtering 16, as shown in FIG6. Unlike the Kalman filter, the example may assume that the dynamics of the desired state are not known in advance, but rather locally approximate the dynamics using a linear time-invariant system and update the dynamics in real time. FIG6 shows a schematic diagram of an example of a control architecture for controlling a robotic system using a feasibility filter 60.

[0061] Assume that the desired joint positions at time t are available and are given by Indicates, where N is the number of joints and the sampling time is The purpose of the feasibility filter 60 shown in FIG6 is to calculate the filtered desired joint state The filtered desired joint state satisfies the kinematic and dynamic constraints of the actuator and avoids the final infeasible state, i.e.,

[0062]

[0063] The desired acceleration, velocity, and position of the actuator at time [t] are calculated according to the following formulas,

[0064]

[0065] The current joint state, the desired joint state, and the filtered desired joint state may include one or more of information about the location or position of the joint, the velocity of the joint, the acceleration of the joint, the jerk of the joint, and / or one or more higher-order derivatives of the location or position. In addition, the state constraints may include kinematic constraints and dynamic constraints of the joint. In addition, the mean square error between the desired joint state and the filtered desired joint state may be minimized to maximize the chance of successfully completing the task. For example, the difference between the desired state and the filtered desired state may be calculated by the difference between the desired joint state and the filtered desired joint state in the location or position x of the joint, the velocity of the joint, and the speed of the joint. Acceleration of joints The jerkiness of the joint The difference may be a mean square error or a weighted mean square error between the desired state and the desired state after filtering taking into account the state constraints. For example, the weighted mean square error (MSE) may be expressed as the sum of W_i×MSE(desired_state_i, filtered_state_i), where W_i allows weighting of the corresponding state variables.

[0066] To this end, the feasibility filter can be formulated as the following convex optimization problem:

[0067]

[0068] in, And use equation 2 to calculate x f [i] and

[0069] satisfy:

[0070]

[0071] x is the current state of the robot, is a local approximation of the desired state and is updated relative to the desired joint state.

[0072] In Equation 5, it is more common to use Rather than (Asymmetric restriction). In practical applications, symmetric restrictions may apply. In some other applications, the above can be extended to systems where:

[0073]

[0074] For example, a rotary actuator coupled to some other component of the drive system may have such a constraint as well as a rotary actuator with directional resistance. This extension can be extended to the case where the velocity, acceleration and jerk limits are asymmetric, however, for ease of derivation and presentation of further results, symmetric constraints are assumed here.

[0075] If the desired velocity, acceleration, and jerk are available; then will just

[0076]

[0077] However, in most practical applications, these values ​​are not available. In such cases, these values ​​can be estimated by using a state estimation algorithm (e.g., parametric or nonparametric). Savitzky-Golay is an example of a parametric state estimation algorithm. An example of a nonparametric method is a moving average filter, which smooths short-term noise while highlighting long-term behavior. The main advantage of the moving average method is the computational time. Another approach can use parametric and nonparametric regression methods that approximate the training data set. Although these methods can be used to estimate the state of the system, these methods may not generalize well beyond the observed data points.

[0078] For example, the current joint state and the filtered desired joint state may include information about the location or position of the joint, the speed of the joint, the acceleration of the joint, and the jerk of the joint. The desired joint state may include three or less elements in a group of information about the location or position of the joint, the speed of the joint, the acceleration of the joint, and the jerk of the joint. Method 10 may also include predicting the desired joint state for subsequent time intervals based on the current joint state and the desired joint state. Then, the desired joint state for subsequent time intervals may include prediction information of one or more elements in a group of the location or position of the joint, the speed of the joint, the acceleration of the joint, and the jerk of the joint. Filtering 16 may be based on the predicted desired joint state for subsequent time intervals.

[0079] The prediction of the expected joint state for the subsequent time interval may include using a state estimation based on the expected joint state to obtain a predicted expected joint state for the subsequent time interval, for example, the state estimation uses a Savitzky-Golay algorithm.

[0080] is an open parameter that specifies the safe range of motion. This open parameter should be selected so that the final infeasible state situation can be avoided. Therefore, filtering 16 can also be based on avoiding the final infeasible state of the robot. The avoidance of the final infeasible state can be based on analyzing the impact of the filtered expected jerk on the future state (determined by K). For example, the parameter K will be selected as follows; subsequently, a proof will be provided. K can be calculated based on the most conservative number of steps required to change the sign of the velocity.

[0081]

[0082] The final infeasible state will only occur in four different scenarios:

[0083] 1- If the velocity is positive, and the acceleration is also positive, then the velocity may end up being infeasible.

[0084] 2- If the velocity is negative, and the acceleration is also negative, then the velocity may end up being unfeasible.

[0085] 3- If the position is positive, and the velocity and acceleration are also positive, then the position may end up being infeasible.

[0086] 4- If the position is negative, and the velocity and acceleration are also negative, then the position may end up being infeasible.

[0087] However, since the speed, acceleration and jerk limits are assumed to be symmetric, only two scenarios need to be considered below, and the most conservative case is taken.

[0088] 1- If the velocity is positive, and the acceleration is also positive, then the velocity may end up being infeasible.

[0089] 2- If the position is positive, and the velocity and acceleration are also positive, then the position may end up being infeasible.

[0090] To avoid violating the velocity constraint in the first case, we want to find out how long it takes to reduce the acceleration from maximum to zero. Given Equation 2, we can write:

[0091]

[0092] This allows the safe range of motion to be calculated as:

[0093]

[0094] In order to avoid violating the position-level constraint in the second scenario, we need to calculate how long it takes to reduce the velocity from the maximum value to zero. Given Equation 2, we can write the following by using mathematical induction:

[0095]

[0096] This yields two solutions:

[0097]

[0098] Given these three values, the most conservative safety margin can be calculated as follows:

[0099]

[0100] Right now

[0101]

[0102] In an example, filtering 16 can be based on a state constraint based on the number of multiple subsequent time intervals. The number of subsequent intervals can be configured to allow the movement of the joint to be stopped based on the maximum deceleration constraint of the joint. As shown above, the number of time intervals can also depend on one or more of the location or position of the joint, the speed of the joint, the maximum allowed acceleration of the joint, and the jerk.

[0103] Figure 7 A block diagram of a flow diagram of an example of a feasibility filter 60 is shown. The feasibility filter 60 takes as input a (partial) desired joint state and then estimates or predicts 61 the full desired joint state as described above. A constrained convex optimization problem as described in equation 1 above is then solved based on the current joint state and the constraints in block 62 to obtain the filtered desired joint state.

[0104] Figure 7 The algorithm described in Formula 1 to Fehler! Verweisquelle konnte nicht gefunden werden is shown. The proposed algorithm is a computationally efficient linear constrained convex optimization problem because the cost function (Equation 1) is a function of only N independent variables. Therefore, N independent optimization problems can be solved instead of solving a single optimization problem with N×K independent variables. In an example, filtering 16 can include determining a filtering rule for each of the plurality of joints separately.

[0105] Furthermore, the algorithm proposed in the example can be solved by analyzing the expected jerk after filtering in Equation 5 The impact on future states is used to deal with the final infeasible state.

[0106] Even though the filtered desired joint states respect the kinematic and dynamic constraints, there is no guarantee that the robot will perfectly follow the filtered desired signal in cases where the system has non-negligible dynamic coupling effects between different joints or a loosely tuned low-level controller. The example can handle this perturbation when updating the filtered desired state at every loop time. However, in cases where the communication delay between the controller and the robot is large and the dynamics of the robot are highly nonlinear, Equation 4 needs to be replaced by

[0107]

[0108] As robot dynamics The first-order Taylor expansion of . Such dynamics can be identified by a parameter system identification algorithm or approximated by a black-box deep learning model. Note that this does not come at the expense of higher computational complexity and more solver time since the first-order expansion is linear with respect to the decision variables. This situation rarely occurs since almost all industrial robot systems have low-level controllers that (i) eliminate the coupling between the robot joints and (ii) simplify the nonlinear dynamics to linear dynamics. In these cases, the proposed filter can be efficiently implemented on a high-level control loop that provides a reference for the robot arm. Therefore, filtering 16 can also be based on coupling effects between the robot's joints and / or nonlinear dynamics of the robot's joints.

[0109] To provide more insights into the proposed feasibility filter, three detailed numerical examples are given next. Figure 8 shows an example timeline of kinematic and dynamic parameters in a first scene, Fig. 9 shows an example timeline of kinematic and dynamic parameters in the second scene, and Fig.10 An example timeline of kinematic and dynamic parameters in the third scenario is shown. Figure 8 , Fig. 9 and Fig.10 In a similar manner, at the top there is a diagram depicting position, velocity, acceleration and jerk over time. The joint position at the top of the diagram always shows the filtered and unfiltered position. The diagrams of velocity, acceleration and jerk at the top show the unfiltered values, and the bottom shows the corresponding (scaled) filtered values.

[0110] exist Figures 8 to 10 , the unfiltered position commands are shown in the upper left corner as 81, 91 and 101, and the corresponding state variable limits are shown in dashed lines. The filtered position commands and the corresponding filtered velocity, acceleration, and jerk commands are shown as lines 82, 92 and 102. Figure 8 The unfiltered velocity, acceleration, and jerk signals are not shown; these are only Fig. 9 The values ​​for the three scenarios are shown in Table 1.

[0111] Figure 8 A three-step reference scenario is shown. The velocity, acceleration, and jerk values ​​are magnified at the bottom for better illustration. It can be seen that the filtered values ​​respect the constraints even when the desired velocity, acceleration, and jerk exceed the maximum values.

[0112] exist Figure 8 In the first example, a three-step reference command is applied to the system:

[0113]

[0114] It can be seen that the step command cannot be followed by the robot without violating the velocity, acceleration, and jerk constraints. On the other hand, the filtered command is shown to converge towards the command position while avoiding violations of the state constraints.

[0115] exist Fig. 9 In the second example in , the commanded position is smooth and follows a sinusoidal motion pattern with an amplitude of 1, making two full oscillations in 1 second:

[0116] x d (t) = cos(4πt)-1

[0117] By looking at the jerk plot, it can be seen that the system is operating at its jerk limit, which results in delayed tracking behavior on the position level. Note that this is the best feasible way to track the position reference given the state constraints and the cost function given in Fehler! Verweisquelle konnte nichtgefunden werden. The unfiltered command clearly violates the state constraints.

[0118] Fig. 9 Scenario 2 with a sinusoidal reference is shown. The velocity, acceleration and jerk values ​​are enlarged at the bottom for better illustration. It can be seen that the filtered values ​​look after the constraints even if the desired velocity, acceleration and jerk exceed the maximum values.

[0119] exist Fig.10 In the third example shown, consider the case where the command position is at the upper limit position constraint: d (t) = 5. Clearly, the position can only be reached if the velocity, acceleration, and jerk are all zero at the moment the position is reached. If any of these states is greater than zero, the position constraint is violated. If any state is less than zero, the commanded position cannot be reached. Finally, note that by using the filter, all constraints are satisfied.

[0120] Fig.10Scenario 3 with a step reference position constraint is shown. The velocity, acceleration and jerk values ​​are magnified at the bottom for better illustration. It can be seen that the filtered values ​​look after the constraints even though the desired velocity, acceleration and jerk exceed the maximum values.

[0121] Table 1:

[0122]

[0123]

[0124] Some examples are summarized below:

[0125] (1) A method for controlling multiple joints of a robot based on expected joint states, the method comprising: obtaining information about expected joint states; obtaining information about current joint states; filtering the expected joint states based on the information about the current joint states and based on state constraints of the robot to obtain information about the filtered expected joint states, wherein the filtered expected joint states comply with the state constraints; and providing information about the filtered expected joint states to control the joints of the robot.

[0126] (2) The method of (1), wherein the filtering of the desired state is based on reducing the difference between the filtered desired state and the desired state taking into account the state constraints.

[0127] (3) A method according to one of (1) or (2), wherein filtering is based on minimizing the difference between the filtered desired state and the desired state taking into account state constraints.

[0128] (4) A method according to one of (2) or (3), wherein the difference is determined by the sum or weighted sum of the differences between the expected joint state and the filtered expected joint state in one or more aspects of the joint location or position, joint velocity, joint acceleration, joint jerk and / or higher-order derivatives of the location or position.

[0129] (5) A method according to one of (2) to (4), wherein the difference is a mean square error or a weighted mean square error between the filtered desired state and the desired state taking into account the state constraints.

[0130] (6) A method according to any one of (1) to (5), wherein filtering includes determining a filtering rule for each of the plurality of joints separately.

[0131] (7) A method according to one of (1) to (6), wherein the filtering is also based on avoiding a final infeasible state of the robot.

[0132] (8) A method according to (7), wherein avoiding the final infeasible state is based on analyzing the impact of the filtered expected jerk on the future state.

[0133] (9) The method according to one of (1) to (8), wherein the filtering is also based on coupling effects between joints of the robot and / or nonlinear dynamics of the joints of the robot.

[0134] (10) A method according to one of (1) to (9), wherein the current joint state, the expected joint state and the filtered expected joint state include one or more of information about the location or position of the joint, the velocity of the joint, the acceleration of the joint, and / or the jerk of the joint, and / or higher-order derivatives of the location or position.

[0135] (11) The method according to any one of (1) to (9), wherein:

[0136] The current joint state and the filtered desired joint state include information about the location or position of the joint, the velocity of the joint, the acceleration of the joint, and the jerk of the joint, and wherein the desired joint state includes three or less elements of the group of information about the location or position of the joint, the velocity of the joint, the acceleration of the joint, and the jerk of the joint, and

[0137] wherein the method further comprises predicting an expected joint state for a subsequent time interval based on the current joint state and the expected joint state, wherein the expected joint state for the subsequent time interval comprises prediction information about one or more elements of the group consisting of a location or position of the joint, a velocity of the joint, an acceleration of the joint, and a jerk of the joint, and

[0138] Therein, filtering is based on predicted desired joint states for subsequent time intervals.

[0139] (12) The method of (11), wherein the prediction of the expected joint state for a subsequent time interval includes using a state estimate based on the expected joint state to obtain a predicted expected joint state for the subsequent time interval.

[0140] (13) The method according to (12), wherein the state estimation uses a Savitzky-Golay algorithm.

[0141] (14) A method according to one of (1) to (13), wherein the filtering is based on a state constraint, and the state constraint is based on the number of multiple subsequent time intervals.

[0142] (15) The method of (14), wherein the number of subsequent intervals is configured to allow stopping of the joint's motion based on a maximum deceleration constraint of the joint.

[0143] (16) A method according to one of (14) or (15), wherein the number of time intervals also depends on the location or position of the joint, the speed of the joint, and one or more of the maximum allowable acceleration and jerk of the joint.

[0144] (17) A method according to one of (1) to (16), wherein the state constraints include kinematic constraints and dynamic constraints of the joints.

[0145] (18) A computer program having a program code, wherein when the computer program is executed on a computer, a processor or a programmable hardware component, the program code is used to perform the method according to any one of (1) to (17).

[0146] (19) An apparatus for controlling a plurality of joints of a robot, comprising a control unit for executing one of the methods described in (1) to (17).

[0147] The aspects and features described with respect to a specific example in the foregoing examples may also be combined with one or more additional examples to replace the same or similar features of the additional examples, or to additionally introduce features into the additional examples.

[0148] Examples may also be or relate to a (computer) program including a program code, which is used to perform one or more of the above methods when executed on a computer, processor or other programmable hardware component. Therefore, the steps, operations or processes of the different methods in the above methods may also be performed by a programmed computer, processor or other programmable hardware component. Examples may also encompass program storage devices, such as digital data storage media, which are machine-readable, processor-readable or computer-readable and encode and / or contain machine-executable, processor-executable or computer-executable programs and instructions. For example, a program storage device may include or may be a digital storage device, a magnetic storage medium such as a disk and a tape, a hard drive, or an optically readable digital data storage medium. Other examples may also include a computer, a processor, a control unit, a (field) programmable logic array ((F)PLA), a (field) programmable gate array ((F)PGA), a graphics processor unit (GPU), an application-specific integrated circuit (ASIC), an integrated circuit (IC) or a system-on-chip (SoC) system programmed to perform the steps of the above methods.

[0149] It should also be understood that the disclosure of several steps, processes, operations or functions disclosed in the specification or claims should not be interpreted as implying that these operations must depend on the described order, unless explicitly stated in individual cases or necessary for technical reasons. Therefore, the foregoing description does not limit the execution of several steps or functions to a specific order. In addition, in other examples, a single step, function, process or operation may include and / or be decomposed into several sub-steps, functions, processes or operations.

[0150] If some aspects have been described in conjunction with a device or system, these aspects should also be understood as descriptions of corresponding methods. For example, blocks, devices or functional aspects of a device or system may correspond to features of a corresponding method, such as method steps. Therefore, aspects described about methods should also be understood as descriptions of corresponding blocks, corresponding elements, attributes or functional features of the corresponding device or corresponding system.

[0151] The attached claims are hereby incorporated into the detailed description, where each claim can stand on its own as a separate example. It should also be noted that although in the claims a dependent claim refers to a specific combination with one or more other claims, other examples may also include a combination of a dependent claim with the subject matter of any other dependent or independent claim. Unless it is stated in individual cases that a specific combination is not intended, such a combination is explicitly proposed here. In addition, with respect to any other independent claim, the features of the claim shall also be included, even if the claim is not directly defined as dependent on the other independent claim.

Claims

1. A method for controlling a plurality of joints of a robot based on desired joint states, the method comprising: obtaining information about the desired joint state; Get information about the current joint state; filtering the desired joint state based on the information about the current joint state and based on the state constraint of the robot to obtain information about the filtered desired joint state, wherein the filtered desired joint state complies with the state constraint; as well as Information about the filtered desired joint states is provided to control the joints of the robot.

2. The method according to claim 1, wherein: The filtering of the desired state is based on reducing the difference between the filtered desired state and the desired state taking into account the state constraints.

3. The method according to claim 1, wherein: The filtering is based on minimizing the difference between the filtered desired state and the desired state taking into account the state constraints.

4. The method according to claim 3, wherein: The difference is determined by the sum or weighted sum of the differences between the expected joint state and the filtered expected joint state in one or more aspects of the location or position of the joint, the velocity of the joint, the acceleration of the joint, the jerk of the joint, and / or higher-order derivatives of the location or position.

5. The method according to claim 4, wherein: The difference is the mean square error or weighted mean square error between the filtered desired state and the desired state taking into account the state constraints.

6. The method according to claim 5, wherein: The filtering includes determining a filtering rule for each of the plurality of joints, respectively.

7. The method according to claim 1, wherein: The filtering is also based on avoiding a final infeasible state of the robot.

8. The method according to claim 7, wherein: The avoidance of the final infeasible state is based on analyzing the impact of the filtered expected jerk on the future state.

9. The method according to claim 1, wherein: The filtering is also based on coupling effects between the joints of the robot and / or nonlinear dynamics of the joints of the robot.

10. The method according to claim 1, wherein: The current joint state, the desired joint state, and the filtered desired joint state include one or more of information about the location or position of the joint, the velocity of the joint, the acceleration of the joint, the jerk of the joint, and / or higher-order derivatives of the location or position.

11. The method according to claim 1, wherein: The current joint state and the filtered desired joint state include information about the location or position of the joint, the velocity of the joint, the acceleration of the joint, and the jerk of the joint, and wherein the desired joint state comprises no more than three elements of the group of information about the location or position of the joint, the velocity of the joint, the acceleration of the joint, and the jerk of the joint, and wherein the method further comprises predicting an expected joint state for a subsequent time interval based on the current joint state and the expected joint state, wherein the expected joint state for the subsequent time interval comprises prediction information about one or more elements of the group consisting of a location or position of the joint, a velocity of the joint, an acceleration of the joint, and a jerk of the joint, and Wherein the filtering is based on a predicted expected joint state for the subsequent time interval.

12. The method according to claim 11, wherein: The prediction of the expected joint state for the subsequent time interval includes using a parametric state estimate based on the expected joint state to obtain the predicted expected joint state for the subsequent time interval.

13. The method according to claim 12, wherein: The parameter state estimation uses the Savitzky-Golay algorithm.

14. The method according to claim 1, wherein: The filtering is based on a state constraint based on a number of a plurality of subsequent time intervals.

15. The method according to claim 14, wherein: The number of subsequent intervals is configured to allow stopping of the motion of the joint based on a maximum deceleration constraint of the joint.

16. The method according to claim 15, wherein: The number of time intervals also depends on the location or position of the joint, the velocity of the joint and one or more of the maximum allowed acceleration of the joint and the jerk of the joint.

17. The method according to claim 1, wherein: The state constraints include kinematic constraints and dynamic constraints of the joints.

18. A computer program having a program code for performing the method according to claim 1 when the computer program is executed on a computer, a processor or a programmable hardware component.

19. An apparatus for controlling a plurality of joints of a robot, comprising a control unit for executing the method according to claim 1.