Mecanum wheel jacking integrated driving assembly and motion and jacking collaborative prediction control method thereof

Abstract

The invention discloses a Mecanum wheel jacking integrated driving assembly and a motion and jacking collaborative prediction control method thereof. The assembly is composed of a chassis, four Mecanum wheel driving modules, a jacking execution unit, a height sensor, an encoder / posture or pressure monitoring unit and a central controller. The controller maps a non-linear state to a Koopman state space through a multi-layer perceptron, multi-step prediction and rolling optimization are carried out, and the wheel speed and the jacking execution amount are solved at the same time in the same control period; during the jacking execution period, speed constraint adaptive to the jacking state is set for the chassis, and feedforward compensation is conducted on the jacking control quantity according to the posture or the acceleration signal; and the Koopman matrix parameters are periodically updated on line to adapt to load and working condition changes. According to the scheme, the synchronism and stability of moving and jacking are improved, and the device is suitable for warehousing, carrying and other scenes.

Description

Technical Field

[0001] This invention belongs to the field of control technology, and in particular relates to a Mecanum wheel lifting integrated drive assembly and its motion and lifting coordinated predictive control method. Background Technology

[0002] Mecanum omnidirectional chassis are widely used in warehousing and flexible production lines, but the mobile chassis and lifting device are mostly designed separately, with drive and control units configured separately, and movement and lifting are often executed sequentially. This architecture is prone to height fluctuations caused by the transmission of movement vibrations to the lifting link under load changes, road surface adhesion coefficient fluctuations, and high-speed scheduling conditions. The center of gravity shift caused by lifting can also induce slippage and accumulation of posture errors. Traditional compensation methods based on empirical decoupling or single-loop PID cannot adequately address the real-time coupling effects of the two subsystems. Furthermore, existing solutions lack a unified state modeling and prediction framework, making it impossible to simultaneously determine the target speed of all four wheels and the lifting execution amount within a single control cycle. They also lack sufficient handling of power distribution and voltage fluctuation linkage, affecting operational efficiency and stability. Summary of the Invention

[0003] In view of this, the present invention aims to propose a Mecanum wheel lifting integrated drive assembly and a method for predictive control of motion and lifting coordination, so as to at least solve one of the problems in the background art.

[0004] To achieve the above objectives, the technical solution of the present invention is implemented as follows: A Mecanum wheel jacking integrated drive assembly includes: The chassis and multiple Mecanum wheel drive modules, each of which includes a motor and an encoder for obtaining wheel speed; A lifting mechanism, the lifting mechanism including an execution unit and a height sensor for acquiring the lifting height; The sensing and monitoring unit includes at least an encoder signal, a height signal from the height sensor, and a sensing signal for reflecting load or execution pressure. The central controller is communicatively connected to the electronic control unit of the Mecanum wheel drive module and the lifting mechanism, and operates according to a preset control cycle. The central controller is configured as follows: The nonlinear system state, including chassis pose-related quantities and lifting height, is encoded and mapped to the Koopman state space using a multilayer perceptron. Multi-step state prediction is performed in the Koopman state space, and rolling optimization is performed based on the prediction results. The target rotation speed and lifting target execution amount of each Mecanum wheel are solved simultaneously in the same control cycle. Under the condition of disturbance caused by chassis movement, feedforward compensation is superimposed on the lifting control quantity, and speed constraints are applied to the chassis movement speed during the lifting execution. The Koopman matrix parameters are updated online according to the control cycle.

[0005] Furthermore, the execution unit is at least one of a motor-driven linear lifting mechanism, a hydraulic cylinder, or a pneumatic cylinder, and the height sensor and the execution unit form a closed-loop control.

[0006] Furthermore, the sensing and monitoring unit further includes a sensor for characterizing attitude or acceleration, and the feedforward compensation generates a compensation term based on the pressure or load disturbance estimated by attitude or acceleration and lifting speed and superimposed on the lifting control quantity.

[0007] Furthermore, the rolling optimization sets constraints on wheel speed, angular velocity, and lifting speed or lifting pressure; when the lifting command is detected to be in execution state, the speed constraint adaptively decays the upper limit of the chassis movement speed according to the lifting speed or height error.

[0008] Furthermore, the central controller operates in the control cycle and sequentially completes the following within each control cycle: system state acquisition and encoding, multi-step prediction in the Koopman state space, rolling optimization solution, control quantity issuance, and online update of the Koopman matrix parameters.

[0009] Furthermore, the central controller performs parameter estimation and updates on the Koopman matrix parameters based on the most recent time-domain data window, and adaptively adjusts the weight coefficients in the rolling optimization according to the load changes obtained by the sensing and monitoring unit.

[0010] Furthermore, the drive assembly adopts a modular design, and the Mecanum wheel drive module and the lifting mechanism are connected to the central controller through a unified electrical and communication interface to switch between independent control and collaborative control.

[0011] Furthermore, it also includes a power management module, which is used to perform power distribution and voltage monitoring when multiple motors work together, so as to support energy scheduling and undervoltage protection under parallel operation of omnidirectional movement and lifting.

[0012] Furthermore, this solution discloses a jacking cooperative predictive control method, applied to the aforementioned Mecanum wheel jacking integrated drive assembly, comprising: The system collects load or execution pressure data provided by the encoder, the height sensor, and the sensing and monitoring unit to form a system state that includes chassis pose-related quantities and lifting height. The system state is encoded and mapped to the Koopman state space using a multilayer perceptron. Multi-step prediction is performed within the Koopman state space, and the rolling optimization is executed, while simultaneously solving for the target rotational speed and lifting target amount of each Mecanum wheel; When a disturbance caused by chassis movement is detected, the aforementioned feedforward compensation is added to the lifting control quantity; During the lifting operation, the speed constraint is applied to the chassis movement speed; The parameters of the Koopman matrix are updated online and executed in a closed loop according to the control cycle.

[0013] Furthermore, when the load estimate obtained by the sensing and monitoring unit changes beyond a threshold, the Koopman matrix parameters are updated, and the weights of the objective function for rolling optimization are adjusted simultaneously to balance pose error and lifting height error.

[0014] Compared with existing technologies, the Mecanum wheel lifting integrated drive assembly and its motion and lifting coordinated predictive control method described in this invention have the following advantages: (1) The Mecanum wheel jacking integrated drive assembly and its motion and jacking cooperative predictive control method described in this invention connects four Mecanum wheel drive modules, jacking execution unit, encoder / height / attitude or pressure and other sensing quantities to the central controller through a unified interface, and combines power management to realize power scheduling and undervoltage protection during parallel operation, thereby reducing coupling uncertainty from the structure and energy path. (2) The Mecanum wheel lifting integrated drive assembly and its motion and lifting coordinated predictive control method described in this invention introduces Koopman state space modeling with a multilayer perceptron as the encoder at the control level. Combined with multi-step prediction and rolling optimization, the wheel speed and lifting execution amount are solved in the same control cycle. Feedforward compensation is implemented based on attitude or acceleration signals, and the upper limit of the moving speed is set to be adaptively adjusted according to the lifting state. At the same time, the Koopman matrix parameters are updated online to adapt to load and working condition changes, thereby improving the synchronization of movement and lifting, the stability of pose and height tracking, and the robustness in complex scenarios. Attached Figure Description

[0015] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a schematic diagram of the motion and lifting coordinated predictive control method according to an embodiment of the present invention. Detailed Implementation

[0016] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other.

[0017] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first," "second," etc., may explicitly or implicitly include one or more of that feature. In the description of this invention, unless otherwise stated, "a plurality of" means two or more.

[0018] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art will understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0019] The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0020] Example 1 This embodiment provides an integrated Mecanum wheel lifting drive assembly and its motion and lifting coordinated predictive control method. The assembly includes a chassis, four independent Mecanum wheel drive modules, a lifting mechanism rigidly connected to the chassis, a sensing and monitoring unit, and a central controller; a power management module provides controlled power to each actuator and monitors power and voltage. Each Mecanum wheel drive module consists of a motor, a reduction gear, a mounting bracket, and an encoder for acquiring wheel speed; the lifting mechanism uses an actuator (e.g., a motor-driven linear lifting mechanism) fixedly connected to the chassis, with its displacement end connected to the platform; height sensors are arranged along the lifting direction to measure the current lifting height. The sensing and monitoring unit includes encoder signals from each wheel, height sensor signals, and sensing signals reflecting load or actuation pressure; attitude or acceleration sensors can also be configured as needed to characterize chassis tilt and vibration states. The central controller establishes communication connections with each motor driver, the electronic control unit of the lifting mechanism, and the sensing and monitoring unit. To facilitate unified modeling and control, the system state is defined as x, y, θ, h, and load-related observations, further incorporating wheel speed, attitude, or acceleration as inputs and disturbances. This structural layout, along with the coupled modeling of the "moving subsystem" and the "lifting subsystem" within the same state framework, forms the basis for achieving coordinated control.

[0021] At the control level, the central controller operates according to a fixed control cycle. In this embodiment, a cycle on the order of 100ms is selected to achieve a rolling closed loop of prediction-optimization-execution under the bus and computing power conditions of the industrial field. To handle the nonlinear coupling between movement and lifting while ensuring real-time performance, a multilayer perceptron (MLP) is used as the encoder to map the aforementioned nonlinear system state to the Koopman state space. In this space, the system evolution can be approximated as linear dynamics, facilitating multi-step state prediction and rolling optimization. The encoder output layer imposes explicit corresponding constraints on key physical quantities x, y, θ, and h to improve the stability and interpretability of the prediction. The Koopman matrix parameters are updated online in each control cycle based on the most recent time window data to adapt to load disturbances and changes in operating conditions. Through the combination of "MLP encoding - Koopman state space prediction - online update," the coupling effect of movement and lifting is explicitly incorporated into the prediction model, thereby collaboratively solving the target speed of the four wheels and the target lifting execution amount within a single control cycle.

[0022] The objective function of the rolling optimization problem takes into account both pose and height errors, and imposes a weighted penalty on the control increment. The feasible regions for wheel speed, angular velocity, and lifting speed / pressure are given in the constraints. To suppress coupling interference, when the controller detects attitude or acceleration disturbances caused by chassis movement, it adds a feedforward compensation term to the lifting control quantity to offset pressure fluctuations and height lag caused by movement vibrations. During lifting execution, the controller adaptively converges the upper limit of chassis movement speed according to the lifting state (e.g., current lifting speed or height error), thereby reducing slippage and attitude fluctuations caused by center of gravity changes and ensuring the synchronous stability of movement and lifting. The target four-wheel speeds and target lifting execution quantities obtained from the optimization solution are sent out in a closed loop through the driver and execution unit. The next control cycle repeats the process of "acquisition—encoding—prediction—optimization—execution—online update".

[0023] From a modeling perspective, the system is decomposed into two subsystems—movement and lifting—and then recoupled in a high-dimensional linear space. The movement subsystem takes the four-wheel speed and chassis tilt angle as inputs and the pose change as output; the lifting subsystem takes the execution pressure, load mass, and lifting speed as inputs and the height change as output. Common coupling phenomena in the field include: chassis vibration during movement causing pressure fluctuations in the lifting mechanism (e.g., hydraulic side), leading to unstable height control; conversely, lifting action causes center of gravity shift and normal load changes, resulting in instantaneous slippage of the Mecanum wheels and accumulation of pose errors. Through the aforementioned Koopman state-space prediction and constrained rolling optimization, this embodiment simultaneously handles two types of coupling within the same optimization framework, improving execution stability in complex terrain or multi-layered scenarios.

[0024] To facilitate engineering implementation, this embodiment adopts a modular design at the hardware and interface levels. Four Mecanum wheel drive modules and the lifting mechanism are connected to the central controller via a unified electrical and communication interface. The interface definition includes motor target speed / current commands, execution unit target displacement / pressure commands, various sensor data acquisition, and status and fault reporting. The power management module allocates power in multi-motor concurrent operation scenarios, monitors the bus voltage in real time, and executes power limiting or speed limiting strategies when undervoltage or current over-limit is detected. The strategy changes are then fed back as constraint parameters to the rolling optimizer, thereby achieving a closed-loop "energy-control" integration.

[0025] In a typical application scenario, the equipment needs to move omnidirectionally to a designated storage location within the warehouse aisle and perform several small-amplitude lifts. The central controller first completes online identification of the system status and initialization of the Koopman matrix parameters within several control cycles, and then enters a stable operation phase: every 100ms, wheel speed, attitude or acceleration, lifting height, load or execution pressure are collected and mapped to the Koopman state space via an MLP encoder; multi-step state prediction is performed based on a linear prediction model to solve for the target four-wheel speed and target lifting execution amount for the next control cycle; if the lifting command is detected to be in execution state, the upper limit of the moving speed is tightened according to the current lifting speed or height error; if the attitude or acceleration disturbance is detected to exceed the limit, pressure feedforward compensation for the lifting control amount is generated. Under different load or ground friction conditions, the Koopman matrix parameters are continuously updated with time window data to maintain prediction accuracy and closed-loop robustness.

[0026] Considering the differences in various working conditions, this embodiment provides several optional implementations. The lifting execution unit can be a motor-driven linear lifting mechanism, a hydraulic cylinder, or a pneumatic cylinder, all of which can operate within the above framework: when using hydraulic or pneumatic methods, the sensing and constraint of "load or execution pressure" focuses more on the closed loop on the pressure side; when using an electric linear lifting mechanism, the change in "load estimation" is mainly obtained through observation of motor current and speed, and parameter adaptation can still be performed through the same Koopman prediction and rolling optimization mechanism. Regarding sensor configuration, when space or cost is limited, only an encoder and a height sensor can be configured, and the MLP encoder can estimate some unobserved quantities during state reconstruction; in situations with high requirements for dynamic disturbance rejection, it is recommended to configure attitude or acceleration sensors to improve the accuracy and timeliness of feedforward compensation.

[0027] To improve engineering usability, this embodiment recommends completing the following process during the assembly and calibration phases: In the static calibration phase, establish the zero point and scale factor of the altitude sensor, and complete the pulse / speed calibration of the four-wheel encoder; in the dynamic calibration phase, collect a period of time-series data under a combination of low-speed omnidirectional cruising and small-amplitude lifting to initialize the Koopman matrix parameters; subsequently, enter the online update phase, performing parameter recursion or least-squares updates according to a 100ms control cycle. This process can coexist with different hardware selections without changing the limitations of the claims.

[0028] From the perspective of the overall system effect, the integrated assembly and collaborative predictive control method have changed the traditional mode of "separation of movement and lifting, and difficulty in synchronization". Under the same external dimensions, the system reduces the uncertainty caused by the coupling between components through unified interface and algorithm collaboration; it solves the movement and lifting commands simultaneously within the same control cycle, and achieves the synchronization of millimeter-level height control and smooth pose control with speed constraints and feedforward compensation; when there are sudden changes in load or ground adhesion conditions, the effectiveness of the predictive model is maintained by updating the Koopman matrix parameters online, thereby improving the operating efficiency and stability in complex terrain, confined spaces or multi-layer operation scenarios.

[0029] Example 2 Under the structural and control framework of Example 1, the lifting execution unit is replaced with a hydraulic cylinder, and a pressure sensor is placed at the cylinder inlet. The central controller introduces an upper limit for pressure and an upper limit for the rate of change of pressure in the rolling optimization constraint. To suppress hydraulic pressure fluctuations caused by movement vibration, the feedforward compensation module generates a pre-compensated pressure quantity by passing the attitude or acceleration signal through a low-pass filter and a proportional circuit, and superimposes it on the lifting command. During the lifting execution, the upper limit of speed automatically converges according to the height error to avoid impact at the end of the lifting process. This variation does not require changes to the Koopman state-space prediction and online update mechanism; only targeted adjustments are made to the parameters and constraints to achieve the same cooperative control performance as Example 1.

[0030] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A Mecanum wheel jacking integrated drive assembly, characterized by, Comprise: a chassis and a plurality of Mecanum wheel drive modules, each of which comprises a motor and an encoder for acquiring wheel speed; a jacking mechanism, which comprises an execution unit and a height sensor for acquiring jacking height; a sensing and monitoring unit, which at least comprises encoder signals, height signals of the height sensor, and sensing signals for reflecting load or execution pressure; a central controller, which is communicatively connected with the electric control units of the Mecanum wheel drive modules and the jacking mechanism, and operates according to a preset control period; wherein the central controller is configured to: encode and map the nonlinear system state including chassis pose-related quantities and jacking height to Koopman state space through multilayer perception; perform multi-step state prediction in the Koopman state space, and based on the prediction results, perform rolling optimization to simultaneously solve the target rotational speed of each Mecanum wheel and the target execution quantity of the jacking mechanism within the same control period; under the condition that chassis movement causes disturbance, add feedforward compensation to the jacking control quantity, and impose an acceleration constraint on the chassis movement speed during jacking execution; update the Koopman matrix parameters online according to the control period.

2. The integrated drive assembly of claim 1, wherein: The execution unit is at least one of a motor-driven linear lifting mechanism, a hydraulic cylinder, or a pneumatic cylinder, and the height sensor and the execution unit form a closed-loop control.

3. The integrated drive assembly of claim 1, wherein: The sensing and monitoring unit further comprises sensors for characterizing attitude or acceleration, and the feedforward compensation generates a compensation term based on the attitude or acceleration and the estimated pressure or load disturbance of the jacking speed and adds it to the jacking control quantity.

4. The integrated drive assembly of claim 1, wherein: The rolling optimization sets constraints on wheel speed, angular velocity, and jacking speed or jacking pressure; when it is detected that the jacking instruction is in the execution state, the speed constraint implements adaptive attenuation on the upper limit of the chassis movement speed according to the jacking speed or height error.

5. The integrated drive assembly of claim 1, wherein: The central controller operates according to the control period, and in each control period, it sequentially completes: system state acquisition and encoding, multi-step prediction in the Koopman state space, rolling optimization solution, control quantity issuance, and online update of the Koopman matrix parameters.

6. The integrated drive assembly of claim 1, wherein: The central controller updates the Koopman matrix parameters according to the parameter estimation of the recent time domain data window, and adaptively adjusts the weight coefficient in the rolling optimization according to the load change obtained by the sensing and monitoring unit.

7. The integrated drive assembly of claim 1, wherein: The drive assembly adopts a modular design, and the Mecanum wheel drive modules and the jacking mechanism are connected with the central controller through a unified electrical and communication interface to switch between independent control and collaborative control.

8. The integrated drive assembly of claim 1, wherein: It also includes a power management module for power distribution and voltage monitoring when multiple motors work collaboratively to support energy scheduling and under-voltage protection under the parallel operation of omnidirectional movement and jacking.

9. A jacking cooperative predictive control method applied to the Mecanum wheel jacking integrated drive assembly of claim 1, characterized in that, Comprise: acquire load or execution pressure data provided by the encoder, the height sensor, and the sensing and monitoring unit to form a system state including chassis pose-related quantities and jacking height; encoding the system state into a Koopman state space via a multilayer perceptron; performing multi-step prediction within the Koopman state space and executing the receding horizon optimization while solving for each target speed of the Mecanum wheels and target actuation of the jacks; adding the feedforward compensation to the jacks' control when disturbance caused by chassis movement is detected; applying the speed constraint to the chassis movement speed during jacks' actuation; updating the Koopman matrix parameters on-line and closed-loop execution according to the control period.

10. The method of claim 9, wherein: updating the Koopman matrix parameters and synchronously adjusting the objective function weights of the receding horizon optimization to balance the pose error and jacks' height error when the load estimation from the sensing and monitoring unit changes beyond a threshold.

Images