A method and controller for motion control of a ball-oriented robot

Abstract

The application discloses a motion control method and controller for a ball robot, and the method comprises the following steps: acquiring initial position and initial speed information of a dynamic target through a visual sensor; calculating an ideal motion path and an expected arrival time of the dynamic target based on a projectile kinematics model, and sending a motion instruction to each joint of the robot to make the robot move to an expected interaction position in advance; collecting the current position of the dynamic target in real time through the visual sensor, comparing the current position with the expected motion path, and correcting the motion trajectory and arrival point of the dynamic target online; generating motion instructions of each joint of the robot through inverse kinematics calculation according to the corrected motion trajectory; all the motion instructions are sent to the joint servo actuator of the robot in parallel through a USB-CAN link and an Ethernet-CAN link, the network layer parameters of the main link are monitored in real time, and the backup link is switched to when the main link fails, so that the dynamic target is operated.

Description

Technical Field

[0001] This invention relates to the field of motion control technology for robots, and in particular to a motion control method and controller for ball robots. Background Technology

[0002] Dynamic targets (such as ping-pong balls) typically fly in the air for only tens to hundreds of milliseconds. This requires robots to process and analyze the dynamic target data collected by vision sensors within an extremely short time, predict the target's trajectory, and plan the robot's motion to achieve target control. However, traditional robot motion controllers are mostly designed for static or pre-programmed trajectory tasks, lacking specific configuration for high-speed dynamic tasks, and are also costly.

[0003] In the prior art, CN103389738A discloses a method and apparatus for predicting the trajectory of a ping-pong ball for a ping-pong robot. This technical solution uses a position detection module composed of a laser emitter and a photodiode. It can calculate the flight speed of the ping-pong ball based on the time difference and position information of the ping-pong ball passing through the two modules. Then, it predicts the flight trajectory of the ping-pong ball based on aerodynamic models and elastic collision mechanics models. Based on the prediction results, it determines whether the ping-pong ball has crossed the net, gone out of bounds, or is within the robot's hitting range. Finally, it sends the decision information to the ping-pong robot to perform the hitting action.

[0004] The aforementioned patent solution has several drawbacks. First, it has limited ability to perceive complex three-dimensional spatial trajectories. Second, the trajectory prediction model in this solution mainly relies on classical projectile physics and simplified aerodynamic models, which significantly reduces prediction accuracy. Third, the solution does not address the reliability design of the controller's underlying communication link. When facing high-speed dynamic tasks, if the communication link fails due to electromagnetic interference or physical faults, the entire system will be unable to function properly, lacking the necessary fault tolerance and redundancy mechanisms.

[0005] In addition to the aforementioned patented technologies, traditional robot motion controllers generally suffer from the following problems when facing high-speed dynamic target interaction tasks: First, sensor data processing delays lead to target positioning lags, making it difficult to meet the real-time requirements of millisecond-level response; Second, the trajectory prediction algorithm lacks deep coupling with the underlying execution system, making it impossible to quickly adjust the control strategy when the target trajectory changes abruptly; Third, controllers often use a single communication link to connect with the underlying servo actuator, resulting in insufficient anti-interference and fault tolerance capabilities of the system; Fourth, the highly integrated hardware architecture of the controller makes it difficult to upgrade computing power and adapt to the computing resource requirements of more complex algorithms in the future.

[0006] It is worth noting that while dual-line redundancy schemes based on homogeneous networks have been applied in the field of industrial control, a low-cost solution for simultaneously converting two heterogeneous networks with vastly different transmission mechanisms into a CAN bus and forming parallel hot standby is lacking. In robot control, if an effective switching strategy is not adopted when switching between different links, the servo motor will receive step pulse commands, resulting in severe jitter, and in severe cases, even system collapse. Therefore, how to achieve seamless switching of heterogeneous communication links on a low-cost, general-purpose hardware platform is a key technical problem that urgently needs to be solved in this field. Summary of the Invention

[0007] Purpose of the invention: In order to overcome the shortcomings of the existing technology, the present invention provides a motion control method and controller for ball robots that solves the problems of lag in tracking, inaccurate prediction and slow response of robots to high-speed moving targets.

[0008] Technical solution: To achieve the above objectives, the present invention provides a motion control method for ball robots, the method comprising:

[0009] The initial position and initial velocity information of dynamic targets are obtained through visual sensors;

[0010] The ideal motion path and expected arrival time of the dynamic target are calculated based on the projectile kinematics model, and motion commands are sent to each joint of the robot so that the robot moves to the expected interaction position in advance to wait.

[0011] The current position of a dynamic target is collected in real time by a visual sensor and compared with the expected movement path to correct the movement trajectory and destination of the dynamic target online.

[0012] Based on the corrected motion trajectory, motion commands for each joint of the robot are generated through inverse kinematics calculations.

[0013] All motion commands are sent to the robot joint servo actuators in parallel via the USB-CAN and Ethernet-CAN links. The network layer parameters of the primary link are monitored in real time, and the system switches to the backup link when the primary link fails, enabling operation of dynamic targets. Of the USB-CAN and Ethernet-CAN links, one is the primary link and the other is the backup link; generally, the USB-CAN link is the primary link.

[0014] Furthermore, the method of sending motion commands to the robot joint servo actuator in parallel via USB-CAN and Ethernet-CAN links, and monitoring the network layer parameters of the primary link in real time, switching to the backup link when the primary link fails, specifically includes:

[0015] Real-time detection of transmission delay characteristics of USB-CAN and Ethernet-CAN links, and calculation of phase time difference between the two heterogeneous links;

[0016] The motion commands of the two links are time-aligned based on hardware timestamps to eliminate clock drift caused by protocol differences and ensure that the CAN bus commands output by the two links are synchronized on the time axis.

[0017] During the transmission of motion commands, the network layer parameters of the primary link are monitored in real time. When the number of consecutive heartbeat packet loss exceeds a preset threshold or the cumulative upper limit of the underlying CAN communication error frames overflows, the link switching logic is triggered to transfer system control from the primary link to the backup link. During the link switching, the heterogeneous network phase time difference buffer queue alignment algorithm is called to perform micro-interpolation compensation on the command stream of the backup link.

[0018] Furthermore, the backup link has a circular sliding window buffer that continuously stores a backup instruction sequence with timestamps; the step of invoking a heterogeneous network phase time difference buffer queue alignment algorithm during link switching to perform micro-interpolation compensation on the instruction stream of the backup link includes:

[0019] When a path switch is triggered, the timestamp of the last successfully sent instruction before the primary link fails is extracted as the breakpoint reference value, and the backup instruction with the closest time span to the breakpoint reference value is retrieved in the circular sliding window buffer as the starting connection point to complete the physical phase alignment between the primary link and the backup link.

[0020] The joint position, instantaneous velocity, and acceleration corresponding to the initial connection point are extracted as dynamic boundary constraints.

[0021] Based on the boundary constraints, an interpolation function that meets the acceleration continuity constraint is constructed, and within the transition time window of the switching link, at least one set of transition completion instructions for filling the time delay gap is generated according to the servo sampling frequency of the control system.

[0022] The transition completion instruction is concatenated with the subsequent instruction sequence after the starting connection point in the circular sliding window buffer to form a continuous instruction stream without step transitions, which is then sent to the robot.

[0023] Furthermore, the step of acquiring the current position of the dynamic target in real time through a visual sensor, comparing it with the expected movement path, and correcting the trajectory and arrival point of the dynamic target online includes:

[0024] Based on location data and timestamp information collected by visual sensors, a time-stamped location sequence is formed;

[0025] A nonlinear filtering algorithm is used to reduce noise in the position sequence. Combined with the air resistance compensation model and Magnus effect correction parameters, a high-precision trajectory estimate is output.

[0026] The high-precision trajectory estimate is fused with the ideal motion path to obtain a corrected trajectory, and the arrival point is corrected accordingly. During fusion, the residual between the high-precision trajectory estimate and the ideal motion path is calculated, and the fusion weights are dynamically adjusted based on the confidence level of the visual sensor. Specifically, when the sphere is flying at high speed or affected by lighting, the position information captured by the visual sensor may contain significant noise or even errors. In this case, the weight of the high-precision trajectory estimate is reduced, while when the sphere's outline is clear, the weight of the high-precision trajectory estimate is increased.

[0027] Furthermore, the method also includes an emergency trajectory mutation handling step:

[0028] When a sudden change in the trajectory slope of a dynamic target exceeding a preset threshold is detected due to collision or external interference, a high-priority hardware interrupt is triggered.

[0029] Forcefully suspend the currently executing motion planning task, switch to emergency planning mode, and regenerate robot motion instructions based on the new motion trajectory;

[0030] New motion commands are sent to the robot joints via a dual-line redundant communication unit consisting of a USB-CAN link and an Ethernet-CAN link.

[0031] Furthermore, the method also includes an environment adaptive feedforward compensation step driven by historical data, specifically including:

[0032] Record and store the cumulative deviation residual between the current position of the dynamic target and the ideal motion path over multiple consecutive operation cycles, as well as the historical correction amount during the online correction process, to construct a historical correction feature database;

[0033] Based on the analysis of the historical correction feature database, the systematic deviation trend caused by the current working environment is identified, and real-time environmental compensation parameters are generated by reverse mapping accordingly; the real-time environmental compensation parameters include at least one of the following: aerodynamic coefficient correction value and environmental wind force vector estimate value;

[0034] The real-time environmental compensation parameters are fed forward as input to the projectile kinematics model to pre-correct the initially calculated ideal motion path, so that the robot moves to the expected interaction position after environmental compensation.

[0035] A motion controller for ball robots, comprising:

[0036] The core processing module adopts a detachable design, integrates a main processor and memory, and runs a Linux-based real-time operating system to implement the above-mentioned motion control method for ball robots.

[0037] The baseboard interface module connects to the core processing module via a standardized interface and integrates multi-protocol communication interfaces, including a serial communication interface, a fieldbus interface, a high-speed transmission interface, and a vision sensor interface; and

[0038] A dual-line redundant communication unit, connected to the core processing module, is configured to simultaneously enable the USB-CAN link and the Ethernet-CAN link as parallel heterogeneous communication paths to send control commands to the robot joint servo actuator.

[0039] Furthermore, the multi-protocol communication interface includes:

[0040] Serial communication interfaces, including RS485 and RS232 interfaces, supporting the Modbus protocol;

[0041] Fieldbus interfaces, including CAN interface, compatible with CANopen protocol;

[0042] High-speed transmission interfaces, including a USB 3.0 interface and a gigabit Ethernet interface, wherein the Ethernet interface supports TCP / IP protocol and EtherCAT protocol;

[0043] Display output interfaces, including an HDMI interface, are used to connect to a host computer for real-time monitoring or virtual simulation.

[0044] Beneficial effects: The motion control method and controller for ball robots of the present invention have the following beneficial effects:

[0045] (1) This invention combines visual feedback with projectile kinematics model to construct a two-layer control logic that combines pre-positioning and online correction, realizes closed-loop tracking of dynamic targets, and ensures the reliability of control command transmission by introducing dual-link parallel transmission and real-time monitoring switching mechanism, reduces the risk of control interruption caused by fluctuation or failure of a single communication link, and improves the overall stability of the system.

[0046] (2) To address the physical transmission delay and protocol differences between heterogeneous links, this feature achieves synchronization of primary and backup link signals on the time axis by aligning delay detection with hardware timestamps. Combined with phase difference buffer queues and micro-interpolation compensation, it solves the problem of instruction stream interruption caused by phase inconsistency between heterogeneous links, which helps maintain the continuity of instruction reception by the servo actuator during link switching and reduces sudden jumps in control instructions. Attached Figure Description

[0047] Figure 1 A schematic diagram of a motion controller for a ball robot;

[0048] Figure 2 This is an outline drawing of the motion controller;

[0049] Figure 3 This is a flowchart illustrating a motion control method for ball robots. Detailed Implementation

[0050] The invention will now be further described with reference to the accompanying drawings.

[0051] like Figure 1 The diagram shown is a schematic of the motion controller for a ball robot. Figure 2 The diagram shown is its external shape. This controller can be used for motion control of robots such as ball-throwing and catching robots and table tennis robots, including:

[0052] The core processing module adopts a detachable design, integrates a main processor and memory, and runs a Linux-based real-time operating system to implement the motion control method for ball robots of this invention.

[0053] The baseboard interface module connects to the core processing module via a standardized interface and integrates multi-protocol communication interfaces, including a serial communication interface, a fieldbus interface, a high-speed transmission interface, and a vision sensor interface; and

[0054] A dual-line redundant communication unit, connected to the core processing module, is configured to simultaneously enable the USB-CAN link and the Ethernet-CAN link as parallel heterogeneous communication paths to send control commands to the robot joint servo actuator.

[0055] Preferably, the multi-protocol communication interface includes:

[0056] Serial communication interfaces, including RS485 and RS232 interfaces, supporting the Modbus protocol;

[0057] Fieldbus interfaces, including CAN interface, compatible with CANopen protocol;

[0058] High-speed transmission interfaces, including a USB 3.0 interface and a gigabit Ethernet interface, wherein the Ethernet interface supports TCP / IP protocol and EtherCAT protocol;

[0059] Display output interfaces, including an HDMI interface, are used to connect to a host computer for real-time monitoring or virtual simulation.

[0060] like Figure 3The motion control method for the ball robot shown is implemented by the aforementioned motion controller, and the method includes the following steps S101-S105:

[0061] Step S101: Obtain the initial position and initial velocity information of the dynamic target through a visual sensor;

[0062] Step S102: Calculate the ideal motion path and expected arrival time of the dynamic target based on the projectile kinematics model, and send motion commands to each joint of the robot so that the robot moves to the expected interaction position in advance to wait.

[0063] Step S103: The current position of the dynamic target is collected in real time by the visual sensor and compared with the expected movement path to correct the movement trajectory and arrival point of the dynamic target online.

[0064] Step S104: Based on the corrected motion trajectory, generate motion commands for each joint of the robot through inverse kinematics calculation;

[0065] In step S105, all motion commands are sent to the robot joint servo actuator in parallel via the USB-CAN link and the Ethernet-CAN link. The network layer parameters of the primary link are monitored in real time, and the system switches to the backup link when the primary link fails, thus enabling operation on the dynamic target. Of the USB-CAN and Ethernet-CAN links, one is the primary link and the other is the backup link; generally, the USB-CAN link is the primary link.

[0066] This invention combines visual feedback with a projectile kinematics model to construct a two-layer control logic that combines pre-positioning and online correction, achieving closed-loop tracking of dynamic targets. By introducing a dual-link parallel transmission and real-time monitoring switching mechanism, the reliability of control command transmission is ensured, the risk of control interruption caused by fluctuations or failures of a single communication link is reduced, and the overall stability of the system is improved.

[0067] Preferably, since the USB-CAN link and the Ethernet-CAN link have fundamental differences in their underlying clock sources and protocols, the control signals output by the two heterogeneous links inevitably have a phase time difference. Therefore, the step S105 above, which involves sending motion commands to the robot joint servo actuator in parallel via the USB-CAN link and the Ethernet-CAN link, and monitoring the network layer parameters of the primary link in real time, and switching to the backup link when the primary link fails, specifically includes the following steps S201-S203:

[0068] Step S201: Real-time detection of the transmission delay characteristics of the USB-CAN link and the Ethernet-CAN link, and calculation of the phase time difference between the two heterogeneous links;

[0069] Step S202: Time axis alignment of motion commands on the two links is performed based on hardware timestamps to eliminate clock drift caused by protocol differences and ensure that the CAN bus commands output by the two links are synchronized on the time axis.

[0070] In step S203, during the transmission of motion commands, the network layer parameters of the primary link are monitored in real time. When the number of consecutive heartbeat packet loss exceeds a preset threshold or the cumulative upper limit of the underlying CAN communication error frames overflows, the link switching logic is triggered to transfer system control from the primary link to the backup link. During the link switching, the heterogeneous network phase time difference buffer queue alignment algorithm is called to perform micro-interpolation compensation on the command stream of the backup link to ensure that the servo command stream received by the robot actuator remains continuous and stepless at the moment of switching.

[0071] Preferably, the backup link has a circular sliding window buffer that continuously stores a backup instruction sequence with timestamps; the step S203 above, which calls the heterogeneous network phase time difference buffer queue alignment algorithm to perform micro-interpolation compensation on the instruction stream of the backup link during link switching, includes the following steps S301-S304:

[0072] Step S301: When path switching is triggered, extract the timestamp of the last successfully sent instruction before the primary link fails as the breakpoint reference value, and retrieve the backup instruction with the closest time span to the breakpoint reference value in the circular sliding window buffer as the starting connection point to complete the physical phase alignment between the primary link and the backup link.

[0073] Step S302: Extract the joint position, instantaneous velocity, and acceleration corresponding to the starting connection point as dynamic boundary constraints;

[0074] Step S303: Based on the boundary constraints, construct an interpolation function that meets the acceleration continuity constraints, and generate at least one set of transition completion instructions to fill the time delay gaps within the transition time window of the switching link, according to the servo sampling frequency of the control system.

[0075] Step S304: The transition completion instruction is concatenated with the subsequent instruction sequence after the starting connection point in the cyclic sliding window buffer to form a continuous instruction stream without step jumps, which is then sent to the robot to eliminate motion impact caused by the difference in transmission time due to heterogeneous networks.

[0076] To address the physical transmission delays and protocol differences between heterogeneous links, this feature achieves synchronization of primary and backup link signals on the time axis through delay detection and hardware timestamp alignment. Combined with phase difference buffer queues and micro-interpolation compensation, it solves the problem of command stream interruptions caused by phase inconsistencies in heterogeneous links, helping to maintain the continuity of command reception by the servo actuator during link switching and reducing sudden jumps in control commands.

[0077] Preferably, the step S103 above, which involves acquiring the current position of the dynamic target in real time using a visual sensor, comparing it with the expected motion path, and correcting the motion trajectory and arrival point of the dynamic target online, includes the following steps S401-S403:

[0078] Step S401: Based on the location data and timestamp information collected by the visual sensor, a time-stamped location sequence is formed;

[0079] Step S402: The position sequence is denoised using a nonlinear filtering algorithm, and high-precision trajectory estimates are output by combining the air resistance compensation model and the Magnus effect correction parameters.

[0080] Step S403: The high-precision trajectory estimate is fused with the ideal motion path to obtain a corrected motion trajectory, and the arrival point is corrected accordingly. During fusion, the residual between the high-precision trajectory estimate and the ideal motion path is calculated, and the fusion weight is dynamically adjusted based on the confidence level of the visual sensor. Specifically, when the sphere is flying at high speed or affected by lighting, the position information captured by the visual sensor may contain significant noise or even errors. In this case, the weight of the high-precision trajectory estimate is reduced, while when the sphere's outline is clear, the weight of the high-precision trajectory estimate is increased.

[0081] The above correction method fully considers the influence of air resistance and ball spin on the ball's trajectory, and also takes into account the confidence level of the visual sensor, which can effectively correct the trajectory and arrival point.

[0082] Preferably, the method further includes an emergency trajectory mutation handling step, comprising the following steps S501-S503:

[0083] Step S501: When a sudden change in the trajectory slope of a dynamic target exceeding a preset threshold is detected due to a collision or external interference, a high-priority hardware interrupt is triggered.

[0084] Step S502: Forcefully suspend the currently executing motion planning task, switch to emergency planning mode, and regenerate robot motion instructions based on the new motion trajectory;

[0085] In step S503, new motion commands are sent to the robot joints via a dual-line redundant communication unit consisting of a USB-CAN link and an Ethernet-CAN link.

[0086] Preferably, the method further includes an environment adaptive feedforward compensation step driven by historical data, specifically including the following steps S601-S603:

[0087] Step S601: Record and store the cumulative deviation residual between the current position of the dynamic target and the ideal motion path in multiple consecutive operation cycles, as well as the historical record of the correction amount in the online correction process, and construct a historical correction feature database.

[0088] Step S602: Based on the historical correction feature database, analyze the systemic deviation trend caused by the current working environment, and generate real-time environmental compensation parameters by reverse mapping; the real-time environmental compensation parameters include at least one of the following: aerodynamic coefficient correction value and environmental wind force vector estimation value;

[0089] Step S603: The real-time environmental compensation parameters are fed forward as a feedforward input to the projectile kinematics model to pre-correct the initially calculated ideal motion path, so that the robot moves to the expected interaction position after environmental compensation.

[0090] The aforementioned feedforward compensation mechanism reduces the adjustment pressure on subsequent online correction stages, helps improve the probability of robot actions being completed in one step, and achieves environmental adaptive optimization.

[0091] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A motion control method for ball robots, characterized in that, The methods include: The initial position and initial velocity information of dynamic targets are obtained through visual sensors; The ideal motion path and expected arrival time of the dynamic target are calculated based on the projectile kinematics model, and motion commands are sent to each joint of the robot so that the robot moves to the expected interaction position in advance to wait. The current position of a dynamic target is collected in real time by a visual sensor and compared with the expected movement path to correct the movement trajectory and destination of the dynamic target online. Based on the corrected motion trajectory, motion commands for each joint of the robot are generated through inverse kinematics calculations. All motion commands are sent to the robot joint servo actuators in parallel via USB-CAN and Ethernet-CAN links. The network layer parameters of the primary link are monitored in real time, and the system switches to the backup link when the primary link fails, thus enabling the operation of dynamic targets.

2. The motion control method for ball robots according to claim 1, characterized in that, The process of sending motion commands to the robot joint servo actuators in parallel via USB-CAN and Ethernet-CAN links, and monitoring the network layer parameters of the primary link in real time, switching to the backup link when the primary link fails, specifically includes: Real-time detection of transmission delay characteristics of USB-CAN and Ethernet-CAN links, and calculation of phase time difference between the two heterogeneous links; Time axis alignment of motion commands between two links is performed based on hardware timestamps; During the transmission of motion commands, the network layer parameters of the primary link are monitored in real time. When the number of consecutive heartbeat packet loss exceeds a preset threshold or the cumulative upper limit of the underlying CAN communication error frames overflows, the link switching logic is triggered to transfer system control from the primary link to the backup link. During the link switching, the heterogeneous network phase time difference buffer queue alignment algorithm is called to perform micro-interpolation compensation on the command stream of the backup link.

3. The motion control method for ball robots according to claim 2, characterized in that, The backup link has a circular sliding window buffer that continuously stores a backup instruction sequence with timestamps; during link switching, the heterogeneous network phase time difference buffer queue alignment algorithm is invoked to perform micro-interpolation compensation on the instruction stream of the backup link, including: When a path switch is triggered, the timestamp of the last successfully sent instruction before the primary link fails is extracted as the breakpoint reference value, and the backup instruction with the closest time span to the breakpoint reference value is retrieved in the circular sliding window buffer as the starting connection point to complete the physical phase alignment between the primary link and the backup link. The joint position, instantaneous velocity, and acceleration corresponding to the initial connection point are extracted as dynamic boundary constraints. Based on the boundary constraints, an interpolation function that meets the acceleration continuity constraint is constructed, and within the transition time window of the switching link, at least one set of transition completion instructions for filling the time delay gap is generated according to the servo sampling frequency of the control system. The transition completion instruction is concatenated with the subsequent instruction sequence after the starting connection point in the circular sliding window buffer to form a continuous instruction stream without step transitions, which is then sent to the robot.

4. The motion control method for ball robots according to claim 1, characterized in that, The step of acquiring the current position of a dynamic target in real time using a visual sensor, comparing it with the expected movement path, and correcting the target's trajectory and arrival point online includes: Based on location data and timestamp information collected by visual sensors, a time-stamped location sequence is formed; A nonlinear filtering algorithm is used to reduce noise in the position sequence. Combined with the air resistance compensation model and Magnus effect correction parameters, a high-precision trajectory estimate is output. The high-precision trajectory estimate is fused with the ideal motion path to obtain the corrected motion trajectory, and the arrival point is corrected accordingly.

5. The motion control method for ball robots according to claim 1, characterized in that, The method also includes an emergency trajectory mutation handling step: When a sudden change in the trajectory slope of a dynamic target exceeding a preset threshold is detected due to collision or external interference, a high-priority hardware interrupt is triggered. Forcefully suspend the currently executing motion planning task, switch to emergency planning mode, and regenerate robot motion instructions based on the new motion trajectory; New motion commands are sent to the robot joints via a dual-line redundant communication unit consisting of a USB-CAN link and an Ethernet-CAN link.

6. The motion control method for ball robots according to claim 1, characterized in that, The method also includes an environment adaptive feedforward compensation step driven by historical data, specifically including: Record and store the cumulative deviation residual between the current position of the dynamic target and the ideal motion path over multiple consecutive operation cycles, as well as the historical correction amount during the online correction process, to construct a historical correction feature database; Based on the analysis of the historical correction feature database, the systematic deviation trend caused by the current working environment is identified, and real-time environmental compensation parameters are generated by reverse mapping accordingly. The real-time environmental compensation parameters are fed forward as input to the projectile kinematics model to pre-correct the initially calculated ideal motion path, so that the robot moves to the expected interaction position after environmental compensation.

7. A motion controller for ball robots, characterized in that, include: The core processing module adopts a detachable design and integrates a main processor and memory, and is used to implement the motion control method for ball robots as described in any one of claims 1-6; The baseboard interface module is connected to the core processing module through a standardized interface and integrates a multi-protocol communication interface. as well as A dual-line redundant communication unit, connected to the core processing module, is configured to simultaneously enable the USB-CAN link and the Ethernet-CAN link as parallel heterogeneous communication paths to send control commands to the robot joint servo actuator.

8. The motion controller for ball robots according to claim 7, characterized in that, The multi-protocol communication interface includes: Serial communication interfaces, including RS485 and RS232 interfaces, supporting the Modbus protocol; Fieldbus interfaces, including CAN interface, compatible with CANopen protocol; High-speed transmission interfaces, including a USB 3.0 interface and a gigabit Ethernet interface, wherein the Ethernet interface supports TCP / IP protocol and EtherCAT protocol; Display output interfaces, including an HDMI interface, are used to connect to a host computer.

Images