JP2001142506A - Vehicle with operation support function - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a vehicle with operation support function, which realizes work similar to that when a skilled operator operates by learning and preserving the skill of the skilled operator, which is difficult to be made into formulation, and reproducing it when an unskilled operator operates the vehicle. SOLUTION: A controller which can learn a control rule, namely, which successively corrects and optimizes the function relation of input and output is installed between the operation tool of a vehicle and a driving part. The controller inputs an operation signal that an operation tool outputs and a sensor signal detecting a work environment and a situation in the vehicle and outputs an operation command sent to the driving part. The control rule is shown by the relation of the input and the output of the controller but it frequently has non-linearity in general and it is constituted by using a computer.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a vehicle operated by an operator, such as a transportation vehicle or a construction vehicle, and a control device therefor. The present invention relates to a vehicle having a function of assisting operation by using a control law taken when the vehicle operates.

[0002]

2. Description of the Related Art Conventionally, a transporting vehicle or a construction vehicle is operated by an operator getting on the vehicle and directly operating an operating tool to give an operation command to the vehicle. For example, in many construction machines, travel and arm operation are realized by directly controlling a hydraulic valve with an operation lever. Further, in order to reduce the force required for operating the operation lever, there is a method in which an electric signal is output by moving the operation lever, and this signal is amplified to drive the hydraulic valve.

[0003]

Since such a conventional vehicle does not have a function of judging an environment or the like or utilizing work experience, it is necessary to operate the vehicle in order to realize a highly difficult work. It required a trained and highly skilled operator. However, the number of such skilled operators is decreasing, and it is becoming difficult for companies and organizations to have sufficient skilled operators. Further, in fostering skilled operators, there is no way to effectively use the skills of existing skilled operators, and resources are wasted. This problem is caused by the fact that there is no work vehicle that can save the skill of a skilled operator and can be applied to subsequent work.

[0004] An object of the present invention is to learn and save the skills of a skilled operator who is difficult to formulate, and to reproduce the skills when an unskilled operator operates the vehicle, thereby performing operations equivalent to those performed by the skilled operator. An object of the present invention is to provide a vehicle with an operation support function that can be realized.

[0005]

The object of the present invention is to provide a controller capable of learning a control law between an operating tool and a drive unit of a vehicle, in other words, a controller capable of sequentially correcting and optimizing a functional relationship between input and output. It is realized by intervening. The controller receives an operation signal output from the operating tool, a sensor signal for detecting a work environment and a situation inside the vehicle, and outputs an operation command to be sent to the drive unit. The control law is expressed by the relationship between the input and output of the controller.
In general, nonlinearity is often strong, and it is realistic to configure using a computer.

[0006] Learning is realized by storing the relationship between input and output when operated by a skilled operator. It is not advisable to save all situations one by one as it is from the viewpoint of storage resources. It is better to find a function relationship between input and output and store the function or function parameter. After the learning is completed, by inputting the operation signal and the sensor signal to the controller, an operation command as an output can be obtained by using the obtained function or the function parameter (reproduction of the learning result).

[0007] Switching between learning and playback can be manually performed by the operators. However, when there is a significant difference in the handling of operating tools between a skilled operator and an unskilled operator, in other words, when there is a significant difference in the change over time of the input signal, the skilled Unskilled can be automatically determined and automatically switched.

As another method of reproducing the learning result,
There is a method of changing the weight of the operation signal of the operator and the output of the learning controller. When the operation is performed by a skilled operator, the weight of the learning controller is reduced, and excessive intervention of the learning controller is avoided, so that the operator can easily exercise his / her skill. When operated by an unskilled operator, the weight of the learning controller is increased, and the learning controller is actively interposed.
Utilize the skills of skilled operators. This weight change is
The operator can manually perform the adjustment, or, similarly to the case of the previous switching, can automatically determine the skill / unskill and automatically adjust.

[0009]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings.

(First Embodiment) FIG. 1 shows a hydraulic power shovel 1 as a working vehicle and a neural network as a learning controller 2, and an operation signal 4 of an operation lever 3 and an operation command 5 corresponding to various sensor signals. Is output to provide operation support. The learning controller 2 includes an input layer 18, a hidden layer 1,
9. A neural network consisting of three layers, that is, an output layer 20, is configured to store in advance parameters used by a skilled operator. As the sensors, external sensors such as a camera 6 and a laser range finder 7 and internal sensors such as a posture sensor 8 and a force sensor 9 are used. Each sensor is connected to the input of the learning control device 2 via the signal processing devices of the signal processing devices 10a to 10d. On the other hand, the output 5 of the learning control device 2 passes through the amplifiers 11a to 11f,
Connect with hydraulic excavator 1. The outputs of the amplifiers 11a to 11f are connected to the left and right actuators of the crawler 13, the base turning section actuator 14, the actuator 15 of the working arm forearm, the actuator 16 of the upper arm, and the actuator 17 of the bucket.

By using the learning controller as described above,
Even when an unskilled operator works in various situations, it is possible to appropriately drive each actuator according to the working environment and the situation inside the vehicle.

(Second Embodiment) FIG.
Can be directly input to the amplifier 33.
This is an example in which a switch 35 is provided on the way so that the input can be made via the learning controller 34. When used by a skilled operator, the operation signal 32 is directly input to the amplifier 33 so that the learning controller does not interfere with the operation. On the other hand, in the case of an unskilled operator, operation support is performed by using the learning controller 34.

(Third Embodiment) FIG. 3 shows an example in which the switch 35 is manually switched in the example of FIG. 2 and is automatically switched. The skilled operator and the unskilled operator each have a unique operation pattern, so that a difference occurs in the quality of work. This specific operation pattern is processed by the operator determination device 41 to determine whether the operator is an expert or an unskilled operator, and the switching unit 4
Switch 2.

(Fourth Embodiment) FIG. 4 shows an example of FIG. 2 in which the switching unit 35 is used for both skilled and unskilled operators. This is an example in which it is possible to respond to the operator. An operation command is obtained by multiplying an operation signal 45 from an operation lever 44 by a coefficient k1 by a multiplier 46 for each actuator,
The output of the learning controller 47 multiplied by the coefficient k2 by the multiplier 48 is obtained by adding by the adder 43. By changing the ratio between the coefficients k1 and k2, the weighting of the operation signal and the learning controller output can be changed. Therefore, it is possible to cope with operators of various skill levels.

(Fifth Embodiment) FIG. 5 shows an example in which the weighting of the operation signal and the output of the learning controller in the example of FIG. 4 is manually set, but is automatically set. As in the example of FIG. 3, the skill level of the operator is evaluated by the operator determination device 50, and the coefficient k1 is changed by the multiplier 51 and the coefficient k2 is changed by the multiplier 52 using the result.

(Sixth Embodiment) The learning parameters of these learning controllers and the discrimination information of skilled / unskilled persons are recorded on an external medium and stored, or reproduced by reading from the external medium. Is possible. As shown in FIG. 6, the learning data of each work machine stored in the external medium 61 is collected in the analysis device 62, integrated and processed, copied to a new external medium 63, and distributed to each work machine. The learning data collected separately by each work machine can be integrated and widely applied.

(Seventh Embodiment) As shown in FIG.
The difference between the operation signal 65 of the operation lever 64 and the output 67 of the learning controller 66 that has learned the skilled work is obtained by the comparator 68,
The processing can be evaluated by the evaluation device 69.

[0018]

According to the present invention, even a non-skilled operator can realize work of the same quality as a skilled operator in various situations. In addition, by changing the degree of involvement of the learning controller according to the operator's skill level, the skilled operator does not feel uncomfortable due to the presence of the learning controller, and the unskilled operator has sufficient operation support by the learning controller. You can work while receiving.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a signal flow according to a first embodiment.

FIG. 2 is a block diagram illustrating a signal flow according to a second embodiment.

FIG. 3 is a block diagram illustrating a signal flow according to a third embodiment.

FIG. 4 is a block diagram illustrating a signal flow according to a fourth embodiment.

FIG. 5 is a block diagram illustrating a signal flow according to a fifth embodiment.

FIG. 6 is a diagram showing a processing flow in a sixth embodiment.

FIG. 7 is a block diagram illustrating a signal flow according to a seventh embodiment.

[Explanation of symbols]

DESCRIPTION OF REFERENCE NUMERALS 1: excavator, 2, 34, 47, 66: learning controller, 3, 31, 44, 64: operating lever, 4, 32, 4
5.65: operation signal, 35, 42: switch, 41, 50
... operator judgment device, 43 ... adder, 46 / 48.5
1.52 ... multiplier, 68 ... comparator.

Claims (7)

[Claims] 1. A vehicle that operates a drive unit based on an operation command to perform a task, and one or more operation tools that output an operation signal when operated by an operator;
A sensor that recognizes the working environment, detects the internal state of the vehicle, etc. and outputs a signal, an input is an operation signal and a sensor signal, and an output is an operation command. A vehicle with an operation support function, comprising a controller that sequentially corrects and outputs an operation command corresponding to the input operation signal. 2. A vehicle that operates a drive unit based on an operation command to perform a task, and one or more operation tools that output an operation signal when operated by an operator;
A sensor that recognizes the working environment, detects the internal state of the vehicle, etc. and outputs a signal, an input is an operation signal and a sensor signal, and an output is an operation command. A controller that sequentially corrects and outputs an operation command corresponding to the input operation signal, and a switch that switches an operation command to the vehicle to either an operation signal from an operating tool or an output from the controller. Vehicle with operation support function. 3. The method according to claim 3, wherein the operator is determined to be an expert or an unskilled person by monitoring the operation signal.
Operation support characterized by automatically switching an operation signal from an operating tool as an operation command to a vehicle for an expert and an output from a controller for an unskilled operator. Vehicle with function. 4. A vehicle that operates a drive unit based on an operation command to perform a task, one or more operation tools that output an operation signal when operated by an operator,
A sensor that recognizes the working environment, detects the internal state of the vehicle, etc. and outputs a signal, an input is an operation signal and a sensor signal, and an output is an operation command. A controller that sequentially corrects and outputs an operation command corresponding to the input operation signal, an operation signal from the operation tool, and a multiplier that multiplies each of the outputs from the controller by a coefficient, and adds the multiplication result, A vehicle with an operation support function comprising an adder that generates an operation command to the vehicle. 5. The method according to claim 5, wherein the operation signal is monitored to determine whether the operator is a skilled person or an unskilled person.
If you are an expert, you can multiply the operation signal from the operation tool by
A vehicle with an operation support function, wherein a non-skilled person automatically adjusts a multiplier so as to increase a coefficient multiplied by an output from a controller. 6. The vehicle with an operation support function according to claim 1, wherein a functional relationship between an input and an output of a controller and discrimination information between a skilled person and an unskilled person can be recorded and reproduced on an external medium. Vehicle with operation support function. 7. An operating tool that outputs an operation signal when operated by an operator, a sensor that outputs a signal by recognizing a work environment, detecting an internal state of a vehicle, and the like, and an input. A controller that stores the functional relationship between input and output using the operation signal, sensor signal, and output as operation commands, and outputs an operation command corresponding to the input operation signal, and an operation signal from an operation tool and an output from the controller. A vehicle with an operation support function having a comparator for comparing the two.