Learning control device, learning control method, and program product utilizing feedback signals

1 . A learning control device comprising:

a learning memory configured to store a correction value to be used in a learning trial;

a hardware processor connected to the learning memory, the hardware processor being configured to:

output a coefficient;

generate a feedback signal on the basis of a tracking error for a target state of a control target and output the feedback signal, the control target operating in response to an input control signal and outputting an operation result state during the learning trial, the feedback signal being used for causing the operation result state of the control target to track the target state, the feedback signal being generated such that a control target waveform represented by transition of an output signal representing the operation result state becomes a waveform obtained by calculating the coefficient on the whole of a base control target waveform determined in advance;

output a calculation result obtained by calculating the coefficient on the correction value, to a feedback communication path by which the tracking error according to the operation result state of the control target is input; and

update the correction value stored in the learning memory in accordance with a signal communicated through the feedback communication path.

2 . The learning control device according to claim 1 , wherein the base control target waveform is the control target waveform obtained when the control target is controlled in accordance with a base learning control waveform represented by transition of the correction value read from the learning memory in the learning trial.

3 . The learning control device according to claim 2 , wherein

the operation result state is represented by a position,

the target state is represented by a target position, and

the hardware processor is configured to:

output a tracking error inverse calculation result obtained by calculating an inverse of the coefficient on the tracking error for the target position;

output a first target velocity corresponding to the tracking error matching the tracking error inverse calculation result in a base target velocity curve, the base target velocity curve representing the base control target waveform by a relation between the tracking error and the target velocity; and

output, as output target velocity, a calculation result obtained by calculating the coefficient on the first target velocity; and

generate, from the output target velocity, the input control signal relative to velocity of the control target and to output the input control signal as the feedback signal.

4 . The learning control device according to claim 1 , wherein the target state and the operation result state include at least one of a position of the control target and velocity of the control target.

5 . A learning control method comprising:

outputting a coefficient;

generating a feedback signal on the basis of a tracking error for a target state of a control target and outputting the feedback signal, the control target operating in response to an input control signal and outputting an operation result state during a learning trial, the feedback signal being used for causing the operation result state of the control target to track the target state, the feedback signal being generated such that a control target waveform represented by transition of an output signal representing the operation result state becomes a waveform obtained by calculating the coefficient on the whole of a base control target waveform determined in advance;

outputting a calculation result obtained by calculating the coefficient on a correction value, which is stored in a memory and used in the learning trial, to a feedback communication path by which the tracking error according to the operation result state of the control target is input; and

updating the correction value stored in the memory in accordance with a signal communicated through the feedback communication path.

6 . A computer program product comprising a non-transitory computer-readable recording medium on which programmed instructions are recorded, the programmed instructions causing a computer to execute processing of:

outputting a coefficient;

generating a feedback signal on the basis of a tracking error for a target state of a control target and outputting the feedback signal, the control target operating in response to an input control signal and outputting an operation result state during a learning trial, the feedback signal being used for causing the operation result state of the control target to track the target state, the feedback signal being generated such that a control target waveform represented by transition of an output signal representing the operation result state becomes a waveform obtained by calculating the coefficient on the whole of a base control target waveform determined in advance;

outputting a calculation result obtained by calculating the coefficient on a correction value, which is stored in a memory and used in the learning trial, to a feedback communication path by which the tracking error according to the operation result state of the control target is input; and

updating the correction value stored in the memory in accordance with a signal communicated through the feedback communication path.