Casting cleaning robot anti-collision control method based on multi-sensor fusion

Abstract

The invention discloses an anti-collision control method for a casting cleaning robot based on multi-sensor fusion. The anti-collision control method comprises a non-contact acquisition module and a contact acquisition module based on multi-sensor construction data; performing fusion processing on the data acquired by the non-contact acquisition module to acquire a robust fusion distance; a non-contact operation mechanism used for anti-collision operation of the casting cleaning robot is constructed, and the speed of an actuator at the tail end of the casting cleaning robot is dynamically adjusted based on the robust fusion distance; the casting cleaning robot is divided into a safety area, an early warning area and a danger area; and a contact operation mechanism used for adjusting the contact force of the casting cleaning robot is constructed, and based on a non-contact operation mechanism, dynamic adjustment of the contact force of the casting cleaning robot in the casting polishing stage is guaranteed according to data acquired by the contact acquisition module. Through a multi-sensor fusion mode, comprehensive perception and analysis of multi-dimensional information are realized, so that an excellent anti-collision effect can be achieved in cutting and grinding scenes.

Description

Technical Field

[0001] The invention relates to the technical field of casting cleaning and robot anti-collision protection, and in particular to an anti-collision control method of a casting cleaning robot based on multi-sensor fusion. Background Art

[0002] Industrial robots often face potential collision risks during the cutting and grinding of castings due to their complex shapes and changing working environments. Collisions between robot actuators and castings may not only damage the castings, resulting in an increase in the defective rate, but may also cause irreversible damage to the robot equipment and even cause production line shutdowns, resulting in economic losses. Therefore, how to effectively prevent and control collisions has become a key issue in improving casting processing efficiency and safety.

[0003] Currently, many anti-collision systems rely on a single type of sensor to detect the environment or control the robot's movements. However, in casting processing scenarios, the presence of strong light, dust, and noise makes a single sensor have great limitations. For example, strong light can cause signal distortion of infrared sensors, and noise can affect the measurement accuracy of ultrasonic sensors. At the same time, a single sensor often detects anomalies only when a collision is about to occur, and the reaction time is insufficient, making it impossible to achieve early protection. In addition, during the contact process between the robot and the casting, if there is a lack of precise control of the contact force, it may cause excessive or insufficient force during the processing process, affecting the processing quality or further causing a collision. In summary, the protection method of a single sensor is difficult to meet the requirements of safety and reliability in complex casting cleaning scenarios. Summary of the invention

[0004] The present invention provides a collision avoidance control method for a casting cleaning robot based on multi-sensor fusion, which realizes comprehensive perception and analysis of multi-dimensional information of complex casting cleaning scenes by means of multi-sensor data fusion, so as to ensure dynamic adjustment of contact force during the casting grinding stage under the premise of safe collision avoidance operation of the casting cleaning robot.

[0005] In order to achieve the above object, the technical solution of the present invention is:

[0006] A method for anti-collision control of a casting cleaning robot based on multi-sensor fusion specifically comprises the following steps:

[0007] S1: Non-contact data acquisition module and contact data acquisition module based on multiple types of sensors;

[0008] The non-contact acquisition module includes a first sensor module disposed at one side of the end of the casting cleaning robot, and used to detect and obtain the sampling distance between the end of the cleaning robot and the casting when the tool is clamped at the end of the robot to cut the riser of the casting;

[0009] The contact acquisition module includes a second sensor module disposed at a tool flange at the end of the casting cleaning robot and used to detect and obtain the contact force during the casting grinding stage;

[0010] S2: Perform fusion processing on the data acquired by the non-contact acquisition module to obtain a robust fusion distance;

[0011] S3: Construct a non-contact operation mechanism for the anti-collision operation area of ​​the casting cleaning robot, and dynamically adjust the actuator speed at the end of the casting cleaning robot based on the robust fusion distance to ensure that the anti-collision operation area of ​​the casting cleaning robot is in a safe zone;

[0012] The anti-collision operation area of ​​the casting cleaning robot includes a safety area, a warning area and a danger area;

[0013] S4: Construct a contact operation mechanism for adjusting the contact force of the casting cleaning robot, and based on the safety zone of the robot's anti-collision operation area maintained in S3 and the data obtained by the contact acquisition module, realize dynamic adjustment of the contact force of the casting cleaning robot during the casting grinding stage.

[0014] Furthermore, the data acquired by the non-contact acquisition module is fused and processed as described in S2, specifically including the following steps:

[0015] S21: constructing a sensor calibration compensation model to compensate the sampled data of the first sensor module, and the expression of the sensor calibration compensation model is:

[0016] D real =c*D measured +d

[0017] Where: D real represents the output of the sensor calibration compensation model, i.e., the expected actual distance; D measured represents the measured distance value of the sensor; c and d represent the fitting parameters used to compensate for the measurement error;

[0018] S22: Acquire continuous sampling distance data of the first sensor module based on a set sampling period;

[0019] And the first sensor module includes an ultrasonic sensor and an infrared photoelectric sensor;

[0020] And perform mean filtering on the continuously sampled distance data to obtain mean filtering data;

[0021] The expression for performing mean filtering is:

[0022]

[0023] Where: D filterdrepresents mean filter data; n represents the number of consecutive sampling times; D i Indicates the distance data collected in the i-th sampling period, i.e. D reeal ;

[0024] S23: Perform data fusion on the mean filter data to obtain a robust fusion distance, and the expression for performing data fusion is:

[0025] D fused =w ir *D ir +w ul *D ul

[0026] Where: D ir Represents the distance data of the ultrasonic sensor after mean filtering; D ul Indicates the distance data of the infrared photoelectric sensor after mean filtering; w ir With w ul Respectively represent the weights of ultrasonic sensor and infrared photoelectric sensor; D fused represents the robust fusion distance.

[0027] Furthermore, the S22 further includes performing distance continuity constraint processing on the continuously sampled distance data acquired by the first sensor module;

[0028] The distance continuity constraint processing is specifically as follows:

[0029] S221: Obtain the continuous sampling distance data D of the current sampling period i The continuous sampling distance data D from the last sampling cycle i-1 and determining whether the distance difference ΔD exceeds a preset distance threshold Threshold;

[0030] If the distance difference ΔD exceeds the preset distance threshold Threshold, the continuous sampling distance data D of the current sampling period is i Remove as outliers;

[0031] Otherwise, no outlier removal operation is performed;

[0032] S222: Acquire continuous sampling distance data after outliers are removed;

[0033] The neighbor interpolation method is used to linearly interpolate the missing sampling period data in the continuous sampling distance data after the outliers are removed. The expression is:

[0034]

[0035] Where: D new Indicates linear interpolation data; Dprev The data of the previous sampling period expressing the data of the missing sampling period; D next Expresses the data of the next sampling period after the data of the missing sampling period.

[0036] Furthermore, the non-contact operation mechanism for the anti-collision operation of the casting cleaning robot is constructed in S3, and its expression is:

[0037]

[0038] Where: V target V represents the target running speed of the casting cleaning robot; low Indicates the minimum speed, that is, the safe speed when the end of the casting cleaning robot approaches the casting; V high Indicates the maximum speed, that is, the high speed when the end of the casting cleaning robot is away from the casting; D warn Indicates the lower limit of the warning distance; D safe Indicates the upper limit of the safe distance.

[0039] Further, the contact operation mechanism for adjusting the contact force of the casting cleaning robot constructed in S4 is specifically as follows:

[0040] Acquire the current contact force F of the casting cleaning robot through the second sensor module;

[0041] And set the sensor contact force threshold range F l ~F h , and set the expected contact force F of the current casting cleaning robot according to the empirical value target , and obtain the contact operation strategy according to the expected contact force and the contact force threshold range;

[0042] And the contact operation strategy is

[0043] Determine the relationship between the current contact force F and the contact force threshold range;

[0044] The contact force threshold range includes a first threshold range, a second threshold range and a third threshold range; and the first threshold range is greater than or equal to the minimum contact force F l and is less than the expected contact force F target ;

[0045] The second threshold range is greater than the expected contact force F target And less than or equal to the maximum contact force F h ;

[0046] The third threshold range is equal to the expected contact force F target ;

[0047] If it is confirmed that the current contact force F is greater than or equal to the minimum contact force F land is less than the expected contact force F target , the contact force between the casting cleaning robot and the casting is increased;

[0048] If it is confirmed that the current contact force F is greater than the expected contact force F target And less than or equal to the maximum contact force F h , the contact force between the casting cleaning robot and the casting is reduced;

[0049] If it is confirmed that the current contact force F is equal to the expected contact force F target , the contact force between the casting cleaning robot and the casting remains unchanged.

[0050] Further, in S221, confirming to execute the outlier removal operation also includes

[0051] When it is confirmed that the distance difference ΔD of any sensor in the first sensor module exceeds the preset distance threshold Threshold, it is also necessary to determine the continuous sampling distance data D of the current sampling period as the abnormal value to be removed. i , the determination distance difference between the continuous sampling distance data obtained by another sensor in the same sampling period;

[0052] If the distance difference does not exceed the preset threshold, the outlier is retained;

[0053] Otherwise, the outlier to be eliminated is eliminated.

[0054] Beneficial effects: The present invention provides a collision avoidance control method for a casting cleaning robot based on multi-sensor fusion. A non-contact acquisition module and a contact acquisition module are constructed based on multi-sensor data. The method realizes comprehensive perception and analysis of multi-dimensional information by means of multi-sensor fusion, thereby solving the problem that the collision avoidance detection capability of the casting cleaning robot cannot be effectively realized due to the low detection accuracy of a single sensor. The data acquired by the non-contact acquisition module is fused to obtain a robust fusion distance. The fused data can adapt to a more complex on-site environment of the casting cleaning robot, thereby greatly improving the robustness and applicability of the collision avoidance of the casting cleaning robot. A hierarchical control of the casting cleaning robot is formed by constructing a non-contact operation mechanism and a contact operation mechanism. The operation speed of the casting cleaning robot is effectively controlled by the non-contact operation mechanism to ensure the safety performance of the casting operation. The contact force for operating the casting is dynamically realized by the contact operation mechanism to ensure the effective polishing of the casting and thus improve the product quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0056] Figure 1 This is a flow chart of the anti-collision control method of a casting cleaning robot based on multi-sensor fusion of the present invention. DETAILED DESCRIPTION

[0057] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0058] This embodiment provides a method for anti-collision control of a casting cleaning robot based on multi-sensor fusion. Figure 1 As shown, the specific steps include:

[0059] S1: Non-contact data acquisition module and contact data acquisition module based on multiple types of sensors;

[0060] The non-contact acquisition module includes a first sensor module disposed at one side of the end of the casting cleaning robot, which is used to detect and obtain the sampling distance between the end of the cleaning robot and the casting when the tool is clamped at the end of the robot to cut the riser of the casting; wherein the first sensor module includes an ultrasonic sensor and an infrared photoelectric sensor;

[0061] The contact acquisition module includes a second sensor module arranged at the tool flange at the end of the casting cleaning robot for detecting and acquiring the contact force during the casting grinding stage; the second sensor module is a force sensor;

[0062] In this embodiment, ultrasonic sensors, infrared photoelectric sensors and force sensors are arranged at the end of the robot. Among them, the ultrasonic sensor and the infrared photoelectric sensor are installed on one side of the end of the robot, and are mainly used to detect workpieces or obstacles at a long distance in the cutting scene (the end of the robot clamps the tool to cut the riser of the casting); the force sensor is installed at the tool flange at the end of the robot, and is mainly used to detect the contact force in the grinding stage. The infrared photoelectric sensor measures the distance by emitting infrared light and detecting the time difference or intensity change of the light signal after it is reflected from the surface of the object; similarly, the ultrasonic sensor calculates the distance by emitting an ultrasonic signal and measuring the time it takes for the signal to be reflected from the object; and the force sensor measures the size and direction of the contact force by detecting the mechanical stress or strain applied to it, and usually uses a piezoelectric or strain gauge sensor;

[0063] S2: Perform fusion processing on the data acquired by the non-contact acquisition module to obtain a robust fusion distance;

[0064] In a specific embodiment, the data acquired by the non-contact acquisition module is fused and processed as described in S2, specifically including the following steps:

[0065] S21: Calibrate the infrared sensor and the ultrasonic sensor according to the reflection characteristics of the surface material of the casting workpiece (such as a smooth surface and a rough surface), and construct a sensor calibration compensation model to compensate the sampling data of the first sensor module;

[0066] And the expression of the sensor calibration compensation model is

[0067] D real =c*D measured +d

[0068] Where: D real represents the output of the sensor calibration compensation model, i.e., the expected actual distance; D measured represents the measured distance value of the sensor; c and d represent the fitting parameters used to compensate for the measurement error;

[0069] S22: Acquire continuous sampling distance data of the first sensor module based on a set sampling period;

[0070] And the first sensor module includes an ultrasonic sensor and an infrared photoelectric sensor;

[0071] And perform mean filtering on the continuously sampled distance data to obtain mean filtering data;

[0072] The expression for performing mean filtering is:

[0073]

[0074] Where: D filtered represents mean filter data; n represents the number of consecutive sampling times; D i Indicates the distance data collected in the i-th sampling period, i.e. D real ;

[0075] In this embodiment, the continuously sampled signals are averaged by means of mean filtering to reduce the impact of sudden external interference during the sensor sampling process;

[0076] In a specific embodiment, the S22 further includes performing distance continuity constraint processing on the continuously sampled distance data of the first sensor module;

[0077] The distance continuity constraint processing is specifically as follows:

[0078] S221: Obtain the continuous sampling distance data D of the current sampling period i The continuous sampling distance data D from the last sampling cycle i-1 and determining whether the distance difference ΔD exceeds a preset distance threshold Threshold;

[0079] If the distance difference ΔD exceeds the preset distance threshold Threshold, the continuous sampling distance data D of the current sampling period is i Remove as outliers;

[0080] Otherwise, no outlier removal operation is performed;

[0081] The principle of distance continuity constraint processing in this embodiment is: in dynamic data collection, the distance change is usually continuous. If the distance change between two consecutive measurements exceeds the set threshold (such as 5% or other values ​​adjusted according to the actual environment), the data is considered abnormal and is discarded. Assume that the current measurement value is D i , the measured value at the last moment is D i-1 , then: Δd=|D i -D i-1 |If Δd>Threshold, determine D i The distance continuity constraint processing utilizes the continuity constraint of distance changes in physical scenes to avoid misjudging normal data as abnormal. By introducing distance continuity constraint processing and multi-sensor joint judgment, the accuracy and flexibility of abnormal data removal are significantly improved.

[0082] In a specific embodiment, the step S221 confirms that an outlier removal operation is to be performed, and further includes:

[0083] When it is confirmed that the distance difference ΔD of any sensor in the first sensor module exceeds the preset distance threshold Threshold, it is also necessary to determine the continuous sampling distance data D of the current sampling period as the abnormal value to be removed. i , the determination distance difference between the continuous sampling distance data obtained by another sensor in the same sampling period;

[0084] If the distance difference does not exceed the preset threshold, the outlier is retained;

[0085] Otherwise, the outlier to be eliminated is eliminated;

[0086] In this embodiment, the abnormal data elimination does not rely on a single sensor alone, but also combines the output of multiple sensors for comprehensive judgment: for example, when the infrared sensor detects an abnormal value (such as a sudden change), the ultrasonic sensor output is verified to be consistent; if the two results are significantly different, the environmental factors (such as light intensity, noise) are further analyzed to determine whether the abnormal value should be eliminated. This reduces the misjudgment caused by the error of a single sensor and enhances the accuracy of data elimination;

[0087] S222: Acquire continuous sampling distance data after outliers are removed;

[0088] The neighbor interpolation method is used to linearly interpolate the missing sampling period data in the continuous sampling distance data after the outliers are removed. The expression is:

[0089]

[0090] Where: D new Indicates linear interpolation data; D prev The data of the previous sampling period expressing the data of the missing sampling period; D next Express the data of the next sampling period after the data of the missing sampling period;

[0091] In this embodiment, after the abnormal data is removed, the neighbor interpolation method is used to interpolate and repair using the normal data before and after the moment to ensure the smoothness of the data flow. When multiple consecutive data are abnormal, high-order interpolation or curve fitting can be used for repair to avoid fusion errors caused by missing data after removing the abnormal data.

[0092] S23: Perform data fusion on the mean filter data to obtain a robust fusion distance, and the expression for performing data fusion is:

[0093] D fused =w ir *D ir +w ul *D ul

[0094] Where: D irRepresents the distance data of the ultrasonic sensor after mean filtering; D ul Indicates the distance data of the infrared photoelectric sensor after mean filtering; w ir With w ul They represent the weights of the ultrasonic sensor and the infrared photoelectric sensor respectively. The values ​​of the two weights are determined by the environmental conditions. For example, when cutting generates strong light, the weight of the ultrasonic sensor is 1 and the weight of the infrared sensor is 0 to eliminate the interference of light. When a large amount of dust is generated, the weight of the infrared sensor is increased to reduce the error of ultrasonic ranging caused by solid media. fused represents the robust fusion distance;

[0095] In this embodiment, filtering and noise removal of the collected raw data can effectively improve the signal-to-noise ratio of the data and ensure the accuracy of the measured value; at the same time, the data fusion of the infrared sensor and the ultrasonic sensor adopts the weighted idea, assigns different weights to data from different sources, and changes the weight values ​​of each under different conditions to adapt to the complex and changeable field environment;

[0096] Before formal data collection, this embodiment needs to calibrate the infrared sensor and ultrasonic sensor. After the data is collected, the sensor data is preprocessed using mean filtering to eliminate sudden noise signals. At the same time, the distance continuity constraint is used to remove abnormal data and improve the signal-to-noise ratio. After data preprocessing, the infrared sensor and ultrasonic sensor data are fused together based on weighted average and rule reasoning according to different scenarios and environmental conditions to obtain a robust fusion distance.

[0097] S3: Construct a non-contact operation mechanism for the collision-avoidance operation of the casting cleaning robot, and dynamically adjust the actuator speed at the end of the casting cleaning robot based on the robust fusion distance;

[0098] The casting cleaning robot is divided into a safety zone, a warning zone and a danger zone;

[0099] In a specific embodiment, the non-contact operation mechanism for the anti-collision operation of the casting cleaning robot is constructed in S3, and its expression is:

[0100]

[0101] Where: V targer V represents the target running speed of the casting cleaning robot; low Indicates the minimum speed, that is, the safe speed when the end of the casting cleaning robot approaches the casting; V high Indicates the maximum speed, that is, the high speed when the end of the casting cleaning robot is away from the casting; D warn Indicates the lower limit of the warning distance; D safe Indicates the upper limit of the safe distance;

[0102] In this embodiment, according to the robust fusion distance D fused The value of the warning distance range D warn In comparison, the speed V of the robot end effector is dynamically adjusted to divide the robot operation area into the following three intervals:

[0103] (1) Safe Zone (D fused >D safe )

[0104] Definition: The current distance is greater than the safe distance D safe , the robot is far away from the casting and no special protection is required.

[0105] The control strategy is: the casting cleaning robot maintains the preset high speed V high , no additional speed adjustment is performed.

[0106] (2) Warning area (D warn <D fused ≤D safe )

[0107] Definition: The current distance enters the warning range and the robot gradually approaches the casting.

[0108] Control strategy: According to the distance D fused Calculate the target speed V target , so that the robot movement gradually slows down, according to the sensor feedback D fused value, call the dynamic adjustment formula to calculate the target speed V target ; In the warning area, the speed is based on D fused Gradually decreases, changing linearly or nonlinearly. And dynamically adjust the formula:

[0109]

[0110] (3) Danger zone (D fused ≤D warn )

[0111] Definition: The current distance is less than or equal to the warning distance. The robot is very close to the casting and requires special protection.

[0112] Control strategy: Immediately switch to low speed operation V low , or directly trigger the protection mechanism (such as emergency stop or evacuation action);

[0113] S4: Construct a contact operation mechanism for adjusting the contact force of the casting cleaning robot, and based on the non-contact operation mechanism, realize dynamic adjustment of the contact force of the casting cleaning robot during the casting grinding stage according to the data obtained by the contact acquisition module;

[0114] In a specific embodiment, the contact operation mechanism for adjusting the contact force of the casting cleaning robot constructed in S4 is specifically as follows:

[0115] Acquire the current contact force F of the casting cleaning robot through the second sensor module;

[0116] And set the sensor contact force threshold range F l ~F h , and set the expected contact force F of the current casting cleaning robot according to the empirical value target , and obtain the contact operation strategy according to the expected contact force and the contact force threshold range;

[0117] And the contact operation strategy is

[0118] Determine the relationship between the current contact force F and the contact force threshold range;

[0119] The contact force threshold range includes a first threshold range, a second threshold range and a third threshold range; and the first threshold range is greater than or equal to the minimum contact force F l and is less than the expected contact force F target ;

[0120] The second threshold range is greater than the expected contact force F target And less than or equal to the maximum contact force F h ;

[0121] The third threshold range is equal to the expected contact force F target ;

[0122] If it is confirmed that the current contact force F is greater than or equal to the minimum contact force F l and is less than the expected contact force F target , the contact force between the casting cleaning robot and the casting is increased;

[0123] If it is confirmed that the current contact force F is greater than the expected contact force F target And less than or equal to the maximum contact force F h , the contact force between the casting cleaning robot and the casting is reduced;

[0124] If it is confirmed that the current contact force F is equal to the expected contact force F target, the contact force between the casting cleaning robot and the casting is kept unchanged. Among them, the way to increase or decrease the contact force between the casting cleaning robot and the casting is, strictly speaking, to increase or decrease the contact force, and the grinding tool and the casting are always in contact during this process. The increase or decrease of the contact force is determined by controlling the distance between the end of the grinding tool and the contact surface (that is, the height of the robot end in the normal direction of the grinding plane); that is, considering that the sanding belt used for grinding is tough and uneven, as the distance increases or decreases, the contact area between the sanding belt and the casting changes, and the magnitude of the grinding force changes accordingly. And the control program or method for controlling the distance between the end of the grinding tool and the contact surface is an existing well-known technology, which will not be elaborated here.

[0125] The robot anti-collision method in the polishing task of the casting cleaning robot in this embodiment adopts a hierarchical control method. The first level is the early warning control before contact (non-contact operation mechanism), that is, before contacting the casting, infrared and ultrasonic sensors are used to determine whether it has entered the early warning distance range, and dynamically adjust the robot's movement speed. The second level is the fine control after contact (contact operation mechanism), that is, setting the threshold of the contact force during the polishing process to ensure efficient polishing under the premise of safety. The protection mechanism is divided into a contact stage and a non-contact stage for design according to whether the casting cleaning robot is in contact with the casting. In the non-contact stage, whether the protection mechanism is triggered is determined according to the safety distance; in the contact stage, whether the protection mechanism is triggered is determined according to the size of the contact force. The robot protection mechanism can effectively cover various processes of casting cleaning and ensure the safety performance of the robot throughout the casting cleaning process.

[0126] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for anti-collision control of a casting cleaning robot based on multi-sensor fusion, characterized in that: The specific steps include: S1: Non-contact data acquisition module and contact data acquisition module based on multiple types of sensors; The non-contact acquisition module includes a first sensor module disposed at one side of the end of the casting cleaning robot, and used to detect and obtain the sampling distance between the end of the cleaning robot and the casting when the tool is clamped at the end of the robot to cut the riser of the casting; The contact acquisition module includes a second sensor module disposed at a tool flange at the end of the casting cleaning robot and used to detect and obtain the contact force during the casting grinding stage; S2: Perform fusion processing on the data acquired by the non-contact acquisition module to obtain a robust fusion distance; S3: Construct a non-contact operation mechanism for the anti-collision operation area of ​​the casting cleaning robot, and dynamically adjust the actuator speed at the end of the casting cleaning robot based on the robust fusion distance to ensure that the anti-collision operation area of ​​the casting cleaning robot is in a safe zone; The anti-collision operation area of ​​the casting cleaning robot includes a safety area, a warning area and a danger area; S4: Construct a contact operation mechanism for adjusting the contact force of the casting cleaning robot, and combine it with the safety zone of the robot's anti-collision operation area maintained in S3 to achieve dynamic adjustment of the contact force of the casting cleaning robot during the casting grinding stage based on the data obtained by the contact acquisition module.

2. The anti-collision control method of a casting cleaning robot based on multi-sensor fusion according to claim 1 is characterized in that: The data acquired by the non-contact acquisition module is fused and processed as described in S2, specifically including the following steps S21: constructing a sensor calibration compensation model to compensate the sampled data of the first sensor module, and the expression of the sensor calibration compensation model is: D real =c*D measured +d Where: D real represents the output of the sensor calibration compensation model, i.e., the expected actual distance; D measured represents the measured distance value of the sensor; c and d represent the fitting parameters used to compensate for the measurement error; S22: Acquire continuous sampling distance data of the first sensor module based on a set sampling period; And the first sensor module includes an ultrasonic sensor and an infrared photoelectric sensor; And perform mean filtering on the continuously sampled distance data to obtain mean filtering data; The expression for performing mean filtering is: Where: D filtered represents mean filter data; n represents the number of consecutive sampling times; D i Indicates the distance data collected in the i-th sampling period, i.e. D real ; S23: Perform data fusion on the mean filter data to obtain a robust fusion distance, and the expression for performing data fusion is: D fused =w ir *D ir +w ul *D ul Where: D ir Represents the distance data of the ultrasonic sensor after mean filtering; D ul Indicates the distance data of the infrared photoelectric sensor after mean filtering; w ir With w ul Respectively represent the weights of ultrasonic sensor and infrared photoelectric sensor; D fused represents the robust fusion distance.

3. The anti-collision control method of a casting cleaning robot based on multi-sensor fusion according to claim 2 is characterized in that: The step S22 further includes performing distance continuity constraint processing on the continuously sampled distance data of the first sensor module; The distance continuity constraint processing is specifically as follows: S221: Obtain the continuous sampling distance data D of the current sampling period i The continuous sampling distance data D of the last sampling cycle i-1 and determining whether the distance difference ΔD exceeds a preset distance threshold Threshold; If the distance difference ΔD exceeds the preset distance threshold Threshold, the continuous sampling distance data D of the current sampling period is i Remove as outliers; Otherwise, no outlier removal operation is performed; S222: Acquire continuous sampling distance data after outliers are removed; The neighbor interpolation method is used to linearly interpolate the missing sampling period data in the continuous sampling distance data after the outliers are removed. The expression is: Where: D new Indicates linear interpolation data; D prev The data of the previous sampling period expressing the data of the missing sampling period; D next Expresses the data of the next sampling period after the data of the missing sampling period.

4. The anti-collision control method of a casting cleaning robot based on multi-sensor fusion according to claim 2 is characterized in that: The non-contact operation mechanism for the anti-collision operation of the casting cleaning robot is constructed in S3, and its expression is: Where: V target V represents the target running speed of the casting cleaning robot; low Indicates the minimum speed, that is, the safe speed when the end of the casting cleaning robot approaches the casting; V high Indicates the maximum speed, that is, the high speed when the end of the casting cleaning robot is away from the casting; D warn Indicates the lower limit of the warning distance; D safe Indicates the upper limit of the safe distance.

5. The anti-collision control method of a casting cleaning robot based on multi-sensor fusion according to claim 2 is characterized in that: The contact operation mechanism for adjusting the contact force of the casting cleaning robot constructed in S4 is specifically as follows: Acquire the current contact force F of the casting cleaning robot through the second sensor module; And set the sensor contact force threshold range F l ~F h , and set the expected contact force F of the current casting cleaning robot according to the empirical value target , and obtain the contact operation strategy according to the expected contact force and the contact force threshold range; And the contact operation strategy is Determine the relationship between the current contact force F and the contact force threshold range; The contact force threshold range includes a first threshold range, a second threshold range and a third threshold range; and the first threshold range is greater than or equal to the minimum contact force F l and is less than the expected contact force F target ; The second threshold range is greater than the expected contact force F target And less than or equal to the maximum contact force F h ; The third threshold range is equal to the expected contact force F target ; If it is confirmed that the current contact force F is greater than or equal to the minimum contact force F l and is less than the expected contact force F target , the contact force between the casting cleaning robot and the casting is increased; If it is confirmed that the current contact force F is greater than the expected contact force F target And less than or equal to the maximum contact force F h , the contact force between the casting cleaning robot and the casting is reduced; If it is confirmed that the current contact force F is equal to the expected contact force F target , the contact force between the casting cleaning robot and the casting remains unchanged.

6. The anti-collision control method of a casting cleaning robot based on multi-sensor fusion according to claim 3 is characterized in that: S221 confirms that the outlier removal operation is performed, and also includes When it is confirmed that the distance difference ΔD of any sensor in the first sensor module exceeds the preset distance threshold Threshold, it is also necessary to determine the continuous sampling distance data D of the current sampling period as the abnormal value to be removed. i , the determination distance difference between the continuous sampling distance data obtained by another sensor in the same sampling period; If the distance difference does not exceed the preset threshold, the outlier is retained; Otherwise, the outlier to be eliminated is eliminated.

Images