CN112578675B - High-dynamic vision control system and task allocation and multi-core implementation method thereof - Google Patents
High-dynamic vision control system and task allocation and multi-core implementation method thereof Download PDFInfo
- Publication number
- CN112578675B CN112578675B CN202110211030.7A CN202110211030A CN112578675B CN 112578675 B CN112578675 B CN 112578675B CN 202110211030 A CN202110211030 A CN 202110211030A CN 112578675 B CN112578675 B CN 112578675B
- Authority
- CN
- China
- Prior art keywords
- control quantity
- visual
- control
- dynamic
- core
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
- G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
Abstract
The application discloses a high dynamic vision control system and a task allocation and multi-core implementation method thereof, which comprise the following steps: the multi-core processor is used for receiving and processing the image frames and outputting a control quantity; the multi-core processor includes: the visual measurement process module is positioned in the first kernel and used for outputting a visual measurement value according to the sampling period and the calculation time delay; the nominal control algorithm module is positioned in the second kernel and used for generating a first control quantity according to the visual measurement value so as to ensure that the system stably operates; the active performance recovery algorithm module is positioned in the other cores and used for automatically updating the parameters on the cores in real time and generating a second control quantity in each sampling period so as to enable the system to be adjusted in a self-adaptive manner; the controlled variable is the sum of the first controlled variable and the second controlled variable. Therefore, the functional segmentation and multi-core configuration of the visual measurement process, the nominal control algorithm and the active performance recovery algorithm can be realized, the algorithm maintenance process is simplified, and the system performance is improved.
Description
Technical Field
The invention relates to the technical field of vision control, in particular to a high-dynamic vision control system and a task allocation and multi-core implementation method thereof.
Background
In a high dynamic vision control system, high performance control is increasingly becoming an important requirement in the fields of vision high precision servo, precision machining and the like. Highly dynamic vision control systems tend to suffer from reduced control performance due to factors including: when disturbance is not modeled, system components are maintained/replaced, and similar systems are deployed in a large range, the control algorithm is difficult to adapt to all the systems in a high-performance mode, and the control algorithm is required to be reconfigured/designed under the three conditions.
However, the existing high-dynamic vision control system is still in the scope of common vision control, so that the control algorithm is difficult to process the control performance reduction caused by the three factors, the algorithm maintenance structure is not concise, and the algorithm maintenance and deployment efficiency is low in the large-range deployment of the similar system. In addition, the way of increasing the computing power by using a high-performance processor is to select a processor, and the increase of the control performance reaches the upper limit.
Therefore, how to achieve high performance control of a high dynamic vision control system is a technical problem to be solved urgently by those skilled in the art.
Disclosure of Invention
In view of the above, the present invention provides a high dynamic vision control system and a task allocation and multi-core implementation method thereof, which achieve improvement of control performance from task allocation and multi-core implementation of a vision measurement process and a control algorithm. The specific scheme is as follows:
a high dynamic vision control system, comprising: the system comprises a camera for acquiring real-time image frames of a high-dynamic controlled object, and a multi-core processor for receiving and processing the image frames and outputting a control quantity to the high-dynamic controlled object; the multi-core processor includes:
the visual measurement process module is positioned in the first kernel and used for outputting a visual measurement value according to the sampling period and the calculation time delay;
the nominal control algorithm module is positioned in the second inner core and used for generating a first control quantity according to the visual measurement value output by the visual measurement process module so as to ensure that the system stably operates;
the active performance recovery algorithm module is positioned in the other cores except the first core and the second core and used for automatically updating the parameters on the cores in real time and generating a second control quantity in each sampling period so as to enable the system to be adjusted in a self-adaptive manner; and the control quantity output by the multi-core processor is the sum of the first control quantity generated by the nominal control algorithm module and the second control quantity generated by the active performance recovery algorithm module.
Preferably, in the above-mentioned high dynamic vision control system provided by the embodiment of the present invention, the nominal control algorithm module includes:
the pre-filter is used for processing the input reference signal and outputting a tracking instruction at the current moment;
the feedback controller is used for generating a visual feedback control quantity at the current moment according to the control quantity at the previous moment and the visual measurement value at the previous moment;
and the output unit is used for generating a first control quantity at the current moment according to the tracking instruction output by the prefilter and the visual feedback control quantity output by the feedback controller.
Preferably, in the above-mentioned high dynamic vision control system provided by an embodiment of the present invention, the feedback controller includes:
the single step register is used for receiving the control quantity of the current moment and outputting the control quantity of the previous moment to the visual observer;
the visual observer is used for outputting a state estimation result of the high-dynamic controlled object at the current moment according to the control quantity at the previous moment and the visual measurement value at the previous moment;
and the feedback gain unit is used for receiving the output result of the visual observer and generating the visual feedback control quantity at the current moment according to the result and the feedback gain.
Preferably, in the above-mentioned high dynamic vision control system provided by the embodiment of the present invention, the active performance recovery algorithm module includes:
the attenuation observer is used for generating a residual signal according to the visual measurement value output by the visual measurement process module;
the dynamic feedback system is positioned in the same inner core as the attenuation observer and is used for outputting a second control quantity at the current moment according to a residual error signal at the previous moment generated by the attenuation observer;
the first parameter updating algorithm unit is used for updating the parameters of the dynamic feedback system in real time;
and the second parameter updating algorithm unit is used for updating the parameters of the attenuation observer in real time.
Preferably, in the high dynamic vision control system provided in the embodiment of the present invention, the first parameter updating algorithm unit is specifically configured to determine the number of required kernels according to the selected order of the dynamic feedback system, the dimension of the vision measurement value, and the dimension of the control quantity, and perform synchronous real-time updating on the parameter of the dynamic feedback system by using a synchronous clock for each kernel;
the second parameter updating algorithm unit is specifically configured to, when an identification instruction is received, obtain a last line vector of a kernel space of the high-dynamic vision control system by online identification using a subspace identification method, and update the parameters of the attenuation observer in real time by using elements of the last line vector of the kernel space of the high-dynamic vision control system.
The embodiment of the present invention further provides a task allocation and multi-core implementation method for the high dynamic vision control system provided by the embodiment of the present invention, including:
outputting a visual measurement value according to the sampling period and the calculation time delay through a visual measurement process module; the vision measurement process module is positioned in a first core of the multi-core processor;
generating a first control quantity according to the visual measurement value output by the visual measurement process module through a nominal control algorithm module so as to enable the system to operate stably; the nominal control algorithm module is located in a second core of the multi-core processor;
automatically updating the parameters on the kernel in real time through an active performance recovery algorithm module, and generating a second control quantity in each sampling period so as to enable the system to be adjusted in a self-adaptive manner; the active performance recovery algorithm module is located in a plurality of other cores of the multi-core processor except the first core and the second core; and the control quantity output by the multi-core processor is the sum of the first control quantity generated by the nominal control algorithm module and the second control quantity generated by the active performance recovery algorithm module.
Preferably, in the task allocation and multi-core implementation method provided in the embodiment of the present invention, the generating, by the nominal control algorithm module, the first control quantity according to the visual measurement value output by the visual measurement process module specifically includes:
processing the input reference signal through a pre-filter, and outputting a tracking instruction at the current moment;
generating a visual feedback control quantity at the current moment according to the control quantity at the previous moment and the visual measurement value at the previous moment through a feedback controller;
and generating a first control quantity at the current moment through an output unit according to the tracking instruction output by the prefilter and the visual feedback control quantity output by the feedback controller.
Preferably, in the task allocation and multi-core implementation method provided in the embodiment of the present invention, the generating, by the feedback controller, the visual feedback control quantity at the current time according to the control quantity at the previous time and the visual measurement value at the previous time specifically includes:
receiving the control quantity of the current moment through a single step register, and outputting the control quantity of the previous moment to the visual observer;
outputting a state estimation result of the high-dynamic controlled object at the current moment according to the control quantity at the previous moment and the visual measurement value at the previous moment by a visual observer;
and receiving the output result of the visual observer through a feedback gain unit, and generating a visual feedback control quantity at the current moment according to the result and the feedback gain.
Preferably, in the method for task allocation and multi-core implementation provided in the embodiment of the present invention, the parameter on the kernel where the active performance recovery algorithm module is located is automatically updated in real time, and the second control quantity is generated in each sampling period, which specifically includes:
generating a residual signal according to the visual measurement value output by the visual measurement process module through an attenuation observer;
outputting a second control quantity at the current moment according to a residual error signal at the previous moment generated by the attenuation observer through a dynamic feedback system which is positioned in the same kernel as the attenuation observer;
updating the parameters of the dynamic feedback system in real time through a first parameter updating algorithm unit;
and updating the parameters of the attenuation observer in real time through a second parameter updating algorithm unit.
Preferably, in the method for task allocation and multi-core implementation provided in the embodiment of the present invention, the updating the parameter of the dynamic feedback system in real time by using the first parameter updating algorithm unit specifically includes:
determining the number of required kernels according to the selected order of the dynamic feedback system, the dimension of a visual measurement value and the dimension of a controlled variable through the first parameter updating algorithm unit, and synchronously updating the parameters of the dynamic feedback system in real time for each kernel by adopting a synchronous clock;
updating the parameters of the attenuation observer in real time through a second parameter updating algorithm unit, which specifically comprises:
and when the second parameter updating algorithm unit receives the identification instruction, the last line vector of the kernel space of the high-dynamic vision control system is obtained by online identification by adopting a subspace identification method, and each element of the last line vector of the kernel space of the high-dynamic vision control system is used for updating the parameters of the attenuation observer in real time.
From the above technical solution, it can be seen that the high dynamic vision control system provided by the present invention includes: the system comprises a camera for acquiring real-time image frames of a high-dynamic controlled object and a multi-core processor for receiving and processing the image frames and outputting a control quantity to the high-dynamic controlled object; the multi-core processor includes: the visual measurement process module is positioned in the first kernel and used for outputting a visual measurement value according to the sampling period and the calculation time delay; the nominal control algorithm module is positioned in the second inner core and used for generating a first control quantity according to the visual measurement value output by the visual measurement process module so as to ensure that the system stably operates; the active performance recovery algorithm module is positioned in the other cores except the first core and the second core and used for automatically updating the parameters on the cores in real time and generating a second control quantity in each sampling period so as to enable the system to be adjusted in a self-adaptive manner; the control quantity output by the multi-core processor is the sum of the first control quantity generated by the nominal control algorithm module and the second control quantity generated by the active performance recovery algorithm module.
The high-dynamic vision control system provided by the invention realizes the functional segmentation and multi-core configuration of the vision measurement process, the nominal control algorithm and the active performance recovery algorithm, can actively recover various unmodeled disturbances, system component performance attenuation and the like which may occur in the whole operation period only by automatically updating the kernel where the active performance recovery algorithm is located under the condition that the nominal control algorithm keeps operating to maintain the stability of the system, does not influence the work of the kernel where the vision measurement process and the nominal control algorithm are located, does not influence the stability of a high-dynamic controlled object, does not need system shutdown, further simplifies the maintenance process of the control algorithm, improves the deployment and maintenance efficiency of the control algorithm when the same system is deployed in a large range, realizes the balanced configuration of computing resources, and improves the real-time performance of the system. In addition, the invention also provides a corresponding task allocation and multi-core implementation method for the high dynamic vision control system, so that the system has higher practicability and has corresponding advantages.
Drawings
In order to more clearly illustrate the embodiments of the present invention or technical solutions in related arts, the drawings used in the description of the embodiments or related arts will be briefly introduced below, it is obvious that the drawings in the following description are only embodiments of the present invention, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
Fig. 1 is a schematic structural diagram of a high dynamic vision control system according to an embodiment of the present invention;
fig. 2 is a block diagram illustrating internal tasks of a multi-core processor in the high dynamic vision control system according to an embodiment of the present invention;
FIG. 3 is a schematic diagram of inter-core and intra-core distribution of a first parameter update algorithm according to an embodiment of the present invention;
FIG. 4 is a diagram illustrating inter-core and intra-core distribution of a second parameter update algorithm according to an embodiment of the present invention;
fig. 5 is a flowchart of a task allocation and multi-core implementation method of a high dynamic vision control system according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The present invention provides a high dynamic vision control system, as shown in fig. 1, comprising: the system comprises a camera 10 for acquiring real-time image frames of a high-dynamic controlled object, and a multi-core processor 20 for receiving and processing the image frames and outputting a control quantity to the high-dynamic controlled object; the multi-core processor 20 includes:
the visual measurement process module 21 is located in the first kernel 1 and used for outputting a visual measurement value according to the sampling period and the calculation delay;
the nominal control algorithm module 22 is located in the second kernel 2 and is used for generating a first control quantity according to the visual measurement value output by the visual measurement process module so as to enable the system to operate stably;
an active performance recovery algorithm module 23 is located in each of the plurality of cores 3, 4, …,N+4, used to update the parameter of the kernel automatically in real time, and generate the second control quantity in each sampling period, so as to make the system self-adaptively adjust; the control quantity output by the multi-core processor is the sum of the first control quantity generated by the nominal control algorithm module and the second control quantity generated by the active performance recovery algorithm module.
It should be noted that the high dynamic controlled object refers to a type of controlled object whose control cycle is often at millisecond level in order to ensure the stability and control performance of such controlled object; the camera is an area-array camera with a sampling period ofTThe output image frame can be a gray scale image or a color image, and the resolution (such as 640 × 480 pixels), the frame rate (such as 60 frames/second) and the channel width (such as 10 bits/12 bits) are not limited;
the vision measurement process is a process of acquiring motion information of a highly dynamic controlled object from an image frame, for example: and obtaining the position and deflection angle of the high-dynamic controlled object through an image processing algorithm, projection and coordinate transformation. The vision measurement process module is separately arranged in the first kernel 1 and has a sampling period ofTThe sampling period is used as the control quantity of the multi-core processor
The output period is calculated by the following formula:
wherein the content of the first and second substances,
a matrix of structures determined for the camera,
for the high dynamic state quantity of the controlled object,
in order to obtain a value for the visual measurement,
calculating a time delay for the vision measurement process, the time delay being of a magnitude satisfying
，
Is the current time (i.e. the first
The time at which one sampling period begins).
As shown in FIG. 2, the kernel in which the vision measurement process module is located outputs the vision measurement value
A nominal control algorithm module is given; the input of the multi-core processor is an image frame (input by the second interface) and the output is a control quantity
(input by the first interface). The third interface is an external interactive interface and is used for configuring the multi-core processor and interacting the processor with other upper computers.
In the high dynamic vision control system provided by the embodiment of the invention, the functional segmentation and multi-core configuration of the vision measurement process, the nominal control algorithm and the active performance recovery algorithm are realized, the high dynamic system can actively recover various unmodeled disturbances, system component performance attenuation and the like which may occur in the whole operation period only by automatically updating the kernel where the active performance recovery algorithm is located under the condition that the nominal control algorithm keeps operating to maintain the stability of the system, does not affect the work of the kernel where the vision measurement process and the nominal control algorithm are positioned, does not affect the stability of the high-dynamic controlled object, does not need the system to be stopped, and then the maintenance process of the control algorithm is simplified, the deployment and maintenance efficiency of the control algorithm in the large-range deployment of the similar system is improved, the balanced configuration of computing resources is realized, and the real-time performance of the system is improved.
Specifically, for the adjustment of the controller in the large-scale deployment of the similar system, the invention realizes the efficient deployment mode: namely, all the systems adopt a nominal control algorithm with the same configuration, and self-adaptive adjustment is carried out on the same type but different systems through an active performance recovery algorithm. This adaptation may be performed by other cores 3, 4, …,Nthe +4 algorithm running in parallel is realized, and the work of the first kernel 1 and the second kernel 2 is not influenced, so that the controller tuning is changed from the traditional 'parameter adjustment-online running-halt-parameter adjustment-online running-until excellent control performance is obtained' into 'design nominal control algorithm-online running-active performance recovery algorithm parallel running-until excellent control performance is obtained', and the deployment efficiency is greatly improved when similar systems are deployed in a large range.
In practical implementation, in the above-mentioned high dynamic vision control system provided by the embodiment of the present invention, as shown in fig. 2, the nominal control algorithm module 22 may include: the device comprises a pre-filter, a feedback controller and an output unit;
a pre-filter for the input reference signal
Processing and outputting the tracking command of the current time
(ii) a The pre-filter is designed off-line, can ensure the tracking performance and has the following form:
wherein the content of the first and second substances,
is the state vector of the pre-filter,
a tracking instruction of the current time output by the pre-filter of the current time,
as a reference signal, the reference signal is,
、
、
、
the four parameter matrixes are designed off line and the parameters are solidified for the parameter matrix of the pre-filter.
A feedback controller for controlling the amount of control according to the previous time
The vision measurement value at the last moment
Generating a visual feedback control quantity at the current time
(ii) a The feedback controller is designed off-line, can ensure the feedback performance, and has the following form:
wherein the content of the first and second substances,
is the state vector of the feedback controller,
the output quantity of the feedback controller at the current moment, namely the visual feedback control quantity at the current moment,
is the value of the visual measurement at the previous moment,
is a system matrix of a high-dynamic controlled object,
is an input matrix of a high-dynamic controlled object,
a structural matrix determined for the camera,
Is a gain matrix and satisfies
Are all located within the unit circle,
is a feedback gain matrix and satisfies
Are all located within the unit circle. Second core 2 output
To the first interface, output
And
to the active performance recovery algorithm module (cores 3, 4, …,N+ 4) output period ofT. The second core 2 has the highest priority, each sampling periodTThere must be an output that cannot be blocked by any other task.
An output unit for outputting the tracking instruction according to the output of the pre-filter
And the visual feedback control quantity output by the feedback controller
A first control amount at the current time is generated.
Further, in practical implementation, in the above-mentioned high dynamic vision control system provided by the embodiment of the present invention, as shown in fig. 2, the feedback controller may include: the single-step register, the visual observer and the feedback gain unit;
single step register, usingReceiving the control quantity at the current moment
Outputting the control quantity of the last time to the visual observer
；
A visual observer for observing the control amount according to the previous time
And the visual measurement value at the previous moment
And outputting the state estimation result of the high-dynamic controlled object at the current moment
；
A feedback gain unit for receiving output result of the visual observer
And based on the result
And feedback gainFGenerating a visual feedback control quantity at the current time
。
In practical implementation, in the above-mentioned high dynamic vision control system provided by the embodiment of the present invention, as shown in fig. 2, the active performance recovery algorithm module 23 may include: the system comprises an attenuation observer, a dynamic feedback system, a first parameter updating algorithm unit and a second parameter updating algorithm unit;
an attenuation observer for generating a residual signal according to the vision measurement value output by the vision measurement process module
(ii) a The attenuation observer has the following structure:
wherein the content of the first and second substances,
in order to attenuate the observer gain,
respectively a system matrix of the attenuation observer, an input matrix of the attenuation observer, an output matrix of the attenuation observer, an input matrix of the attenuation observer, and an input matrix of the attenuation observer,
is the state vector of the attenuation observer. When the prior model of the highly dynamic visual controlled object is known,
can be determined according to a system prior model and written in off-line; when the prior model of the high-dynamic controlled object is unknown or parameter uncertainty caused by component performance attenuation occurs,
the identification is carried out on line by adopting a system identification method, and the identification process is realized by updating an algorithm (kernel 4, …,N) After completion of identification
Containing the parameters to be determined, in total
Each, all are stored in one
And the vector is the last line vector of the kernel space of the high dynamic vision control system obtained by identification.
A dynamic feedback system, located in the same kernel as the attenuation observer, for generating a residual signal at the last moment according to the attenuation observer
Outputting the second control quantity of the current time
(ii) a The dynamic feedback system has the following structure:
wherein the subscript
Refers to
The dimension (c) of (a) is,
refers to
The dimension (c) of (a) is,
for the state vector of the dynamic feedback system,
the output quantity at the present time thereof, i.e. the second control quantity at the present time,
is the residual signal at the previous time instant.
As mentioned above, the output period of the dynamic feedback system isTEach sampling periodTThere must be an output that can be blocked; the output cycle output after blocking
Setting zero; parameter matrix thereof
、
、
The update period of (A) is an integral multiple of the output period, defined as
Named parameter update period.
The parameters that need to be determined in advance during the initialization phase include:
is the order of the dynamic feedback system;
in total
Number of which it satisfies
。
The parameter vector is a column vector, the symbol represents matrix transposition, and the two parameter vectors contain
And (5) designing parameters to be online.
Parameter matrix
、
、
Is updated byOver-paired parameter vector
Is implemented.
A first parameter update algorithm unit for updating the parameters of the dynamic feedback system in real time, using the kernel 4, …,N+3；
a second parameter updating algorithm unit for updating the parameters of the attenuation observer in real time by using the kernelN +4。
Further, when implemented, in the above-mentioned high dynamic vision control system provided by the embodiment of the present invention, the first parameter updating algorithm unit is specifically configured to update the algorithm unit according to the selected order of the dynamic feedback system
And visual measurement values
Dimension and control quantity of
Determining the number of required kernels, and respectively aligning the parameters of the dynamic feedback system to each kernel by adopting a synchronous clock
Performing synchronous real-time updating;
specifically, the input to each core is
、
、
Output is
. FIG. 3 shows the distribution of time between cores and the distribution of time inside the cores where the first parameter updating algorithm unit is located
A core, in the figure
. Wherein the content of the first and second substances,
refers to the period of one parameter update of the dynamic feedback system, which is the sampling periodTThe integral multiple of the total number is set manually;
referring to a dynamic feedback system to perform one-time parameter targeting
The window width of the data to be acquired (i.e. the number of sampling cycles of data acquisition 2) is calculated,
is determined according to the amount of data required by the parameter calculation algorithm. Data acquisition 1 is not used for targeting parameters
And (4) calculating. Data acquisition 1 and data acquisition 2 are realized by FIFO with the length of FIFO
One sampling periodTThe length of data involved.
Synchronous clock for synchronously updating parameter vectors
To the dynamic feedback system parameter matrix
、
、
. Due to the fact that the core 4, …,N+3 is calculated to
Are difficult to be completely consistent, and parameter vectors in a dynamic feedback system are ensured
The elements in the system are updated synchronously, and the data will be synchronized by the synchronous clock in the cores 4, …,Nand +3 gives a synchronous updating signal after finishing the calculation. After the synchronous updating signal comes, the parameter vector is output to a dynamic feedback system (kernel 3).
The second parameter updating algorithm unit is specifically used for generating the effect caused by the performance attenuation of components when the prior model of the high-dynamic controlled object is unknownAfter the parameters are uncertain, or when the control algorithm is deployed in a large range in the similar system, namely when an identification instruction is received, a vector is obtained by online identification by adopting a subspace identification method
And using the vector
The parameters of the attenuation observer are updated in real time by the elements of (1).
In particular, the vector
Each element of (1)
Corresponding attenuation observer
The corresponding parameters in the matrix. The second parameter update algorithm module has as input
、
、
Output is
. Fig. 4 shows a time allocation manner inside a core where the second parameter updating algorithm unit is located, and the operation process is as follows: inner coreN+4 no vector
When calculating (2), the data is in the state of collecting data, and the data enters FIFO (the FIFO length is
One sampling periodTThe length of data involved) and awaits a recognition instruction. After receiving the identification instruction, taking the current data in the FIFO for calculation, and outputting the output vector after the calculation is finished corresponding to the second parameter updating algorithm of FIG. 4
To the attenuation observer, the FIFO is emptied and the process is repeated. The identification instruction is input by a third interface of the multi-core processor.
Based on the same inventive concept, the embodiment of the invention also provides a task allocation and multi-core implementation method of the high dynamic vision control system, and as the principle of solving the problems of the method is similar to that of the high dynamic vision control system, the implementation of the method can refer to the implementation of the high dynamic vision control system, and repeated parts are not repeated.
In specific implementation, the task allocation and multi-core implementation method of the high dynamic vision control system provided by the embodiment of the present invention, as shown in fig. 5, specifically includes the following steps:
s501, outputting a visual measurement value through a visual measurement process module according to a sampling period and a calculation time delay; the vision measurement process module is positioned in a first core of the multi-core processor;
s502, generating a first control quantity according to the visual measurement value output by the visual measurement process module through a nominal control algorithm module so as to enable the system to operate stably; the nominal control algorithm module is positioned in a second core of the multi-core processor;
s503, automatically updating the parameters on the kernel in real time through the active performance recovery algorithm module, and generating a second control quantity in each sampling period so as to enable the system to be adjusted in a self-adaptive manner; the active performance recovery algorithm module is positioned in a plurality of other cores except the first core and the second core in the multi-core processor; the control quantity output by the multi-core processor is the sum of the first control quantity generated by the nominal control algorithm module and the second control quantity generated by the active performance recovery algorithm module.
In the task allocation and multi-core implementation method of the high dynamic vision control system provided by the embodiment of the invention, under the condition that the nominal control algorithm keeps running to maintain the stability of the system, the restoration of the attenuation of the control performance caused by the fact that the control algorithm is difficult to adapt to all systems when the same type of system is deployed in a large range without influencing the vision measurement process and the operation of the kernel where the nominal control algorithm is located can be realized only by automatically updating the kernel where the active performance recovery algorithm is located, so that the deployment and maintenance efficiency of the control algorithm when the same type of system is deployed in a large range is improved, the balanced configuration of computing resources can be realized, and the real-time performance of the system is improved.
In specific implementation, in the method for task allocation and multi-core implementation of a high dynamic vision control system according to the embodiment of the present invention, step S502 generates a first control quantity according to a vision measurement value output by a vision measurement process module through a nominal control algorithm module, which may specifically include: processing the input reference signal through a pre-filter, and outputting a tracking instruction at the current moment; generating a visual feedback control quantity at the current moment according to the control quantity at the previous moment and the visual measurement value at the previous moment through a feedback controller; and generating a first control quantity at the current moment through an output unit according to the tracking instruction output by the prefilter and the visual feedback control quantity output by the feedback controller.
Further, in a specific implementation, in the method for task allocation and multi-core implementation of a high dynamic vision control system according to an embodiment of the present invention, the generating, by the feedback controller, a visual feedback control quantity at a current time according to a control quantity at a previous time and a visual measurement value at the previous time may specifically include: receiving the control quantity of the current moment through the single-step register, and outputting the control quantity of the previous moment to the visual observer; outputting a state estimation result of the high-dynamic controlled object at the current moment according to the control quantity at the previous moment and the visual measurement value at the previous moment by a visual observer; and receiving the output result of the visual observer through a feedback gain unit, and generating the visual feedback control quantity at the current moment according to the result and the feedback gain.
In specific implementation, in the method for task allocation and multi-core implementation of a high dynamic visual control system according to the embodiment of the present invention, step S503 is to automatically update parameters of a kernel in real time through an active performance recovery algorithm module, and generate a second control quantity in each sampling period, which may specifically include: generating a residual signal according to the visual measurement value output by the visual measurement process module through an attenuation observer; outputting a second control quantity at the current moment according to a residual error signal at the previous moment generated by the attenuation observer through a dynamic feedback system which is positioned in the same kernel as the attenuation observer; updating the parameters of the dynamic feedback system in real time through a first parameter updating algorithm unit; and updating the parameters of the attenuation observer in real time through a second parameter updating algorithm unit.
In the foregoing steps, the updating the parameter of the dynamic feedback system in real time by the first parameter updating algorithm unit may specifically include: and determining the number of required kernels according to the selected order of the dynamic feedback system, the dimension of the visual measurement value and the dimension of the control quantity through a first parameter updating algorithm unit, and synchronously updating the parameters of the dynamic feedback system in real time by adopting a synchronous clock to each kernel.
In the foregoing step, the updating the parameter of the attenuation observer in real time by using the second parameter updating algorithm unit may specifically include: and when the second parameter updating algorithm unit receives the identification instruction, the last line vector of the kernel space of the high-dynamic vision control system is obtained by online identification through a subspace identification method, and the parameters of the attenuation observer are updated in real time by using each element of the last line vector of the kernel space of the high-dynamic vision control system.
For more specific working processes of the above steps, reference may be made to corresponding contents disclosed in the foregoing embodiments, and details are not repeated here.
The embodiments are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same or similar parts among the embodiments are referred to each other. The device, the equipment and the storage medium disclosed by the embodiment correspond to the method disclosed by the embodiment, so that the description is relatively simple, and the relevant points can be referred to the method part for description.
Those of skill would further appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative components and steps have been described above generally in terms of their functionality in order to clearly illustrate this interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present application.
The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. A software module may reside in Random Access Memory (RAM), memory, Read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.
The embodiment of the invention provides a high dynamic vision control system, which comprises: the system comprises a camera for acquiring real-time image frames of a high-dynamic controlled object and a multi-core processor for receiving and processing the image frames and outputting a control quantity to the high-dynamic controlled object; the multi-core processor includes: the visual measurement process module is positioned in the first kernel and used for outputting a visual measurement value according to the sampling period and the calculation time delay; the nominal control algorithm module is positioned in the second inner core and used for generating a first control quantity according to the visual measurement value output by the visual measurement process module so as to ensure that the system stably operates; the active performance recovery algorithm module is positioned in the other cores except the first core and the second core and used for automatically updating the parameters on the cores in real time and generating a second control quantity in each sampling period so as to enable the system to be adjusted in a self-adaptive manner; the control quantity output by the multi-core processor is the sum of the first control quantity generated by the nominal control algorithm module and the second control quantity generated by the active performance recovery algorithm module. The high-dynamic vision control system realizes the functional segmentation and multi-core configuration of a vision measurement process, a nominal control algorithm and an active performance recovery algorithm, can actively recover various unmodeled disturbances, system component performance attenuation and the like which may occur in the whole operation period by the high-dynamic system only through the automatic updating of the kernel where the active performance recovery algorithm is located under the condition that the nominal control algorithm keeps operating to maintain the stability of the system, does not influence the work of the kernel where the vision measurement process and the nominal control algorithm are located, does not influence the stability of a high-dynamic controlled object, does not need system shutdown, further simplifies the maintenance process of the control algorithm, improves the deployment and maintenance efficiency of the control algorithm when the same system is deployed in a large range, realizes the balanced configuration of computing resources, and improves the real-time performance of the system. In addition, the invention also provides a corresponding task allocation and multi-core implementation method for the high dynamic vision control system, so that the system has higher practicability and has corresponding advantages.
Finally, it should also be noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
The high dynamic vision control system and the task allocation and multi-core implementation method thereof provided by the invention are described in detail above, and a specific example is applied in the text to explain the principle and the implementation mode of the invention, and the description of the above embodiment is only used to help understanding the method and the core idea thereof; meanwhile, for a person skilled in the art, according to the idea of the present invention, there may be variations in the specific embodiments and the application scope, and in summary, the content of the present specification should not be construed as a limitation to the present invention.
Claims (10)
1. A high dynamic vision control system, comprising: the system comprises a camera for acquiring real-time image frames of a high-dynamic controlled object, and a multi-core processor for receiving and processing the image frames and outputting a control quantity to the high-dynamic controlled object; wherein the multi-core processor comprises:
the visual measurement process module is positioned in the first kernel and used for outputting a visual measurement value according to the sampling period and the calculation time delay;
the nominal control algorithm module is positioned in the second inner core and used for generating a first control quantity according to an input reference signal and a visual measurement value output by the visual measurement process module so as to ensure that the system stably runs and tracks the reference signal;
the active performance recovery algorithm module is positioned in the other cores except the first core and the second core and used for automatically updating the parameters on the cores in real time and generating a second control quantity in each sampling period so as to enable the system to be adjusted in a self-adaptive manner; and the control quantity output by the multi-core processor is the sum of the first control quantity generated by the nominal control algorithm module and the second control quantity generated by the active performance recovery algorithm module.
2. The high dynamic vision control system of claim 1, wherein the nominal control algorithm module comprises:
the pre-filter is used for processing the input reference signal and outputting a tracking instruction at the current moment;
the feedback controller is used for generating a visual feedback control quantity at the current moment according to the control quantity at the previous moment and the visual measurement value at the previous moment;
and the output unit is used for generating a first control quantity at the current moment according to the tracking instruction output by the prefilter and the visual feedback control quantity output by the feedback controller.
3. The high dynamic vision control system of claim 2, wherein the feedback controller comprises:
the single step register is used for receiving the control quantity of the current moment and outputting the control quantity of the previous moment to the visual observer;
the visual observer is used for outputting a state estimation result of the high-dynamic controlled object at the current moment according to the control quantity at the previous moment and the visual measurement value at the previous moment;
and the feedback gain unit is used for receiving the output result of the visual observer and generating the visual feedback control quantity at the current moment according to the result and the feedback gain.
4. The highly dynamic vision control system of claim 3, wherein the active performance recovery algorithm module comprises:
the attenuation observer is used for generating a residual signal according to the visual measurement value output by the visual measurement process module;
the dynamic feedback system is positioned in the same inner core as the attenuation observer and is used for outputting a second control quantity at the current moment according to a residual error signal at the previous moment generated by the attenuation observer;
the first parameter updating algorithm unit is used for updating the parameters of the dynamic feedback system in real time;
and the second parameter updating algorithm unit is used for updating the parameters of the attenuation observer in real time.
5. The high-dynamic vision control system according to claim 4, wherein the first parameter updating algorithm unit is specifically configured to determine the number of required kernels according to the selected order of the dynamic feedback system, the dimension of the vision measurement value, and the dimension of the control quantity, and perform synchronous real-time updating on the parameters of the dynamic feedback system at each kernel by using a synchronous clock;
the second parameter updating algorithm unit is specifically configured to, when an identification instruction is received, obtain a last line vector of a kernel space of the high-dynamic vision control system by online identification using a subspace identification method, and update the parameters of the attenuation observer in real time by using elements of the last line vector of the kernel space of the high-dynamic vision control system.
6. A task allocation and multi-core implementation method of the high-dynamic vision control system according to any one of claims 1 to 5, comprising:
outputting a visual measurement value according to the sampling period and the calculation time delay through a visual measurement process module; the vision measurement process module is positioned in a first core of the multi-core processor;
generating a first control quantity according to an input reference signal and a visual measurement value output by the visual measurement process module through a nominal control algorithm module so as to enable a system to stably operate and track the reference signal; the nominal control algorithm module is located in a second core of the multi-core processor;
automatically updating the parameters on the kernel in real time through an active performance recovery algorithm module, and generating a second control quantity in each sampling period so as to enable the system to be adjusted in a self-adaptive manner; the active performance recovery algorithm module is located in a plurality of other cores of the multi-core processor except the first core and the second core; and the control quantity output by the multi-core processor is the sum of the first control quantity generated by the nominal control algorithm module and the second control quantity generated by the active performance recovery algorithm module.
7. The task allocation and multi-core implementation method according to claim 6, wherein the generating a first control quantity by a nominal control algorithm module according to the input reference signal and the visual measurement value output by the visual measurement process module specifically comprises:
processing the input reference signal through a pre-filter, and outputting a tracking instruction at the current moment;
generating a visual feedback control quantity at the current moment according to the control quantity at the previous moment and the visual measurement value at the previous moment through a feedback controller;
and generating a first control quantity at the current moment through an output unit according to the tracking instruction output by the prefilter and the visual feedback control quantity output by the feedback controller.
8. The task allocation and multi-core implementation method according to claim 7, wherein generating the visual feedback control quantity at the current time by the feedback controller according to the control quantity at the previous time and the visual measurement value at the previous time specifically includes:
receiving the control quantity of the current moment through the single-step register, and outputting the control quantity of the previous moment to the visual observer;
outputting a state estimation result of the high-dynamic controlled object at the current moment according to the control quantity at the previous moment and the visual measurement value at the previous moment by the visual observer;
and receiving the output result of the visual observer through a feedback gain unit, and generating a visual feedback control quantity at the current moment according to the result and the feedback gain.
9. The task allocation and multi-core implementation method according to claim 8, wherein the active performance recovery algorithm module automatically updates the parameters of the kernel in real time, and generates the second control quantity in each sampling period, specifically comprising:
generating a residual signal according to the visual measurement value output by the visual measurement process module through an attenuation observer;
outputting a second control quantity at the current moment according to a residual error signal at the previous moment generated by the attenuation observer through a dynamic feedback system which is positioned in the same kernel as the attenuation observer;
updating the parameters of the dynamic feedback system in real time through a first parameter updating algorithm unit;
and updating the parameters of the attenuation observer in real time through a second parameter updating algorithm unit.
10. The task allocation and multi-core implementation method according to claim 9, wherein the updating the parameter of the dynamic feedback system in real time by the first parameter updating algorithm unit specifically comprises:
determining the number of required kernels according to the selected order of the dynamic feedback system, the dimension of a visual measurement value and the dimension of a controlled variable through the first parameter updating algorithm unit, and synchronously updating the parameters of the dynamic feedback system in real time at each kernel by adopting a synchronous clock;
updating the parameters of the attenuation observer in real time through a second parameter updating algorithm unit, which specifically comprises:
and when the second parameter updating algorithm unit receives the identification instruction, the last line vector of the kernel space of the high-dynamic vision control system is obtained by online identification by adopting a subspace identification method, and each element of the last line vector of the kernel space of the high-dynamic vision control system is used for updating the parameters of the attenuation observer in real time.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110211030.7A CN112578675B (en) | 2021-02-25 | 2021-02-25 | High-dynamic vision control system and task allocation and multi-core implementation method thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110211030.7A CN112578675B (en) | 2021-02-25 | 2021-02-25 | High-dynamic vision control system and task allocation and multi-core implementation method thereof |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112578675A CN112578675A (en) | 2021-03-30 |
| CN112578675B true CN112578675B (en) | 2021-05-25 |
Family
ID=75113985
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110211030.7A Active CN112578675B (en) | 2021-02-25 | 2021-02-25 | High-dynamic vision control system and task allocation and multi-core implementation method thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112578675B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113253653A (en) * | 2021-05-31 | 2021-08-13 | 中国人民解放军国防科技大学 | Optimization method for visual real-time control performance of high-dynamic system and related device |
| CN113506320B (en) * | 2021-07-15 | 2024-04-12 | 清华大学 | Image processing method and device, electronic equipment and storage medium |
Family Cites Families (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2006040387A (en) * | 2004-07-27 | 2006-02-09 | Funai Electric Co Ltd | Dvd device |
| US20100119109A1 (en) * | 2008-11-11 | 2010-05-13 | Electronics And Telecommunications Research Institute Of Daejeon | Multi-core multi-thread based kanade-lucas-tomasi feature tracking method and apparatus |
| CN101789128B (en) * | 2010-03-09 | 2012-01-18 | 成都三泰电子实业股份有限公司 | Target detection and tracking method based on DSP and digital image processing system |
| US8988531B2 (en) * | 2010-07-08 | 2015-03-24 | Texas Instruments Incorporated | Method and apparatus for sub-picture based raster scanning coding order |
| CN102096925A (en) * | 2010-11-26 | 2011-06-15 | 中国科学院上海技术物理研究所 | Real-time closed loop predictive tracking method of maneuvering target |
| CN107424178A (en) * | 2017-02-24 | 2017-12-01 | 西安电子科技大学 | A kind of Target Tracking System implementation method based on Cortex series polycaryon processors |
| CN107368785A (en) * | 2017-06-16 | 2017-11-21 | 浙江万里学院 | The video target tracking method of multinuclear local restriction |
| CN108769550B (en) * | 2018-05-16 | 2020-07-07 | 中国人民解放军军事科学院军事医学研究院 | Image significance analysis system and method based on DSP |
| CN110706256A (en) * | 2019-09-27 | 2020-01-17 | 杭州博雅鸿图视频技术有限公司 | Detection tracking algorithm optimization method based on multi-core heterogeneous platform |
| CN111462189B (en) * | 2020-04-16 | 2023-06-20 | 吉林大学 | Image locking and tracking system and method |
-
2021
- 2021-02-25 CN CN202110211030.7A patent/CN112578675B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN112578675A (en) | 2021-03-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112578675B (en) | High-dynamic vision control system and task allocation and multi-core implementation method thereof | |
| Ho et al. | Adaptive neural control for a class of nonlinearly parametric time-delay systems | |
| US10474951B2 (en) | Memory efficient scalable deep learning with model parallelization | |
| US10482379B2 (en) | Systems and methods to perform machine learning with feedback consistency | |
| JP6514257B2 (en) | Machine learning apparatus, servo control apparatus, servo control system, and machine learning method | |
| US9513623B2 (en) | Method for generating trajectory for numerical control process | |
| He et al. | Uncertainty and disturbance estimator-based distributed synchronization control for multiple marine surface vehicles with prescribed performance | |
| US20150279021A1 (en) | Video object tracking in traffic monitoring | |
| US9134713B2 (en) | System and method for fault prognostics enhanced MPC framework | |
| Tarel et al. | Using robust estimation algorithms for tracking explicit curves | |
| CN104778661A (en) | Skeleton smoothing method and skeleton smoothing equipment used in pose estimation | |
| CN108053444A (en) | Pupil positioning method and device, equipment and storage medium | |
| Huang et al. | Distributed output feedback consensus control of networked homogeneous systems with large unknown actuator and sensor delays | |
| Peñarrocha et al. | Estimation in multisensor networked systems with scarce measurements and time varying delays | |
| Shaghaghian et al. | Improving GPS/INS integration using FIKF‐filtered innovation Kalman filter | |
| CN109934775B (en) | Image processing, model training, method, device and storage medium | |
| JP7045857B2 (en) | Systems and methods for superior performance with respect to highest performance values in model predictive control applications | |
| JP6970827B2 (en) | Arithmetic processing unit | |
| Üstüntürk | Output feedback stabilization of nonlinear dual-rate sampled-data systems via an approximate discrete-time model | |
| CN110856014A (en) | Moving image generation method, moving image generation device, electronic device, and storage medium | |
| CN113361380B (en) | Human body key point detection model training method, detection method and device | |
| Smyshlyaev et al. | Output-feedback adaptive control for parabolic PDEs with spatially varying coefficients | |
| Voelker et al. | Simultaneous constrained moving horizon state estimation and model predictive control by multi-parametric programming | |
| US20210232947A1 (en) | Signal processing device, signal processing method, and computer program product | |
| Fiter et al. | State dependent sampling: an LMI based mapping approach |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |