JP6776738B2 - Information providing device and information providing method - Google Patents

Description

The present invention relates to an information providing device.

There is a technology to provide driving assistance to the driver based on the information acquired from the sensor mounted on the vehicle. For example, a system that gives a voice warning when a vehicle deviates from the lane and a technology that automatically corrects the steering angle when the warning is ignored are known.

When driving assistance is provided to the driver by an alarm, there is a problem that the driver becomes accustomed to the alarm and cannot take appropriate measures. In order to deal with this, in the inventions described in Patent Documents 1 and 2, attention is given by gradually changing the alarm sound. Further, in the invention described in Patent Document 3, it is sharpened by deciding whether or not to call attention based on whether or not the road on which the vehicle is traveling is a priority road.

JP-A-2015-018397 Japanese Unexamined Patent Publication No. 2010-186227 Japanese Unexamined Patent Publication No. 2012-014219 Japanese Unexamined Patent Publication No. 2005-062911

Risk compensation behavior is a problem in systems that provide driving assistance to drivers. Risk-compensating behavior refers to a person changing his or her behavior to the side where the risk increases when it becomes safe. For example, if there is a system that detects the approach to an obstacle and automatically applies the brakes with a probability of 95%, the driver may rely on the system and not apply the brakes by himself / herself. That is, the risk to the remaining 5% becomes extremely high.
In the conventional technology, there is a problem that it is difficult to prevent such behavior that unconsciously relies on the system.

The present invention has been made in consideration of the above problems, and an object of the present invention is to prevent inappropriate behavior of a driver due to habituation in a system for assisting a driver.

The information providing device according to the present invention is
An acquisition means for acquiring sensor information from one or more sensors mounted on a vehicle, a generation means for generating support information which is information to be provided to a user based on the sensor information, and a certainty level of the support information. Based on the above, the expression rate, which is the probability of providing the support information to the user, is calculated, and the information providing means for providing the support information is provided based on the expression rate.

The generation means is a means for generating information to be provided to the user based on one or more sensor information. The information provided to the user is typically information that calls attention to danger, but may be other information. For example, it may read a road sign and notify it, or it may give advice on driving.
In addition, the information providing means calculates the probability (expression rate) of providing the information to the user based on the certainty of the information to be provided. Conviction is the degree to which the information to be provided is correct.
According to such a configuration, the probability that the user receives the correct information can be adjusted to an arbitrary value. For example, it is possible to prevent habituation by intentionally suppressing and equalizing the probability that the user will receive the correct information.

Further, the information providing means determines the expression rate so that the probability that the user receives the correct information is substantially uniform through the plurality of support information when the generation means generates the support information. May be.

In conventional information providing devices, it is common to actively present the acquired information, but if the probability of presenting the information is unconditionally increased, the user becomes accustomed to the presented information and the presented information is incorrect. In that case, the user may take inappropriate actions.
In the present invention, in order to correspond to this, the expression rate is set low when the certainty of the information is high, and the expression rate is set high when the certainty of the information is low. By doing so, the probability of providing the correct information to the user can be made uniform.
For example, if the correct information is provided in succession and then the incorrect information is provided, the driver may believe that the information is correct and take inappropriate actions, but the correct information is provided. By deliberately suppressing and equalizing the probability of doing so, it is possible to prevent inappropriate behavior due to habituation.

Further, the information providing means may be characterized in that the certainty of the support information is calculated based on the accuracy of the sensor information obtained from the one or more sensors.

The information provided is generated by sensor information obtained from one or more sensors. Therefore, the certainty of the support information can be calculated based on the accuracy of the sensor information. As the accuracy of the sensor information, a design value may be used, or a value obtained by correcting the design value according to the environment may be used.

Further, the support information includes a first information type which is an information type whose conviction varies depending on whether or not it is in a specific situation, and the providing means is a case where the support information is the first information type. In addition, it may be characterized in that information indicating whether or not the patient is in the specific situation is acquired, and the expression rate is corrected based on the information.

For example, sensors that measure distance using radar are known to have reduced accuracy during rainfall. That is, when the distance between vehicles is measured using such a sensor, the certainty of the information "distance between vehicles" changes under the condition of "when it rains". Therefore, it is desirable to correct the expression rate after acquiring whether or not the patient is in a specific situation. In the case of the above-mentioned example, since the certainty is lowered when it rains, it is preferable to make a correction to increase the occurrence rate when it rains.

Further, the support information includes a second information type in which the conformance rate is lower than the first threshold value, and the providing means is categorical when the support information is the second information type. It may be characterized by providing the support information in an expression other than the above.

The information type with a low precision rate is an information type with few detection omissions but many false positives. Since this type of information has a high erroneous provision rate (probability that the information is erroneous when the information is provided), it is preferable to provide the information in a non-assertive expression. For example, provide information with the expression "Be careful of the right rear" instead of "There is a vehicle behind the right".

Further, the providing means is characterized in that when the support information is of the second information type, the expression rate is set lower than that of the case where the support information is other than the second information type. May be good.

In the case of the second information type, that is, the information type with few omissions but many false positives, setting the expression rate by the same method as other information increases the number of times false information is provided, and the user's information It may damage the trust. Therefore, when the target information is the second information type, the correction may be performed to intentionally lower the expression rate.

The present invention can be specified as an information providing device including at least a part of the above means. It can also be specified as an information providing method performed by the information providing device. The above processes and means can be freely combined and carried out as long as there is no technical contradiction.

According to the present invention, in a system that assists a driver, it is possible to prevent an inappropriate behavior of the driver due to habituation.

It is a system block diagram of the information provision system which concerns on 1st Embodiment. It is a flowchart explaining the process performed in 1st Embodiment. It is a figure explaining the provision probability in 1st Embodiment. It is a figure explaining the provision probability in 2nd Embodiment. It is a system block diagram of the information providing apparatus which concerns on 3rd Embodiment. It is a flowchart explaining the process performed in 3rd Embodiment. It is a flowchart explaining the process performed in 3rd Embodiment. It is a figure explaining the provision probability in 4th Embodiment.

(First Embodiment)
The information providing system according to the first embodiment will be described with reference to FIG. 1, which is a system configuration diagram. The information providing system according to the first embodiment estimates the presence or absence of a specific event based on the information collected from a plurality of sensors provided in the vehicle (hereinafter referred to as sensor information), and based on the estimation result, It is a system that gives advice on driving operations.
The specific event is, for example, the presence or absence of a pedestrian in front of the driving lane, the necessity of changing lanes, the presence or absence of a stopped vehicle, etc., but is limited to these as long as it can be estimated based on sensor information. Absent.

The information providing system according to the first embodiment includes a sensor 10 mounted on the vehicle and an information providing device 20 mounted on the vehicle. The sensor 10 and the information providing device 20 may each be configured by a plurality of units.

The information providing device 20 can be configured as an information processing device having a CPU, a main storage device, and an auxiliary storage device. When the program stored in the auxiliary storage device is loaded into the main storage device and executed by the CPU, each means shown in FIG. 1 functions. In addition, all or a part of the illustrated functions may be executed by using a circuit designed exclusively for them.

The sensor 10 is a group of a plurality of sensors mounted on a vehicle, and is a means for outputting sensor information. The sensor 10 may include, for example, a speed sensor, an acceleration sensor, a yaw rate sensor, a steering angle sensor, a radar, a camera, and the like. Further, the sensor information may be any information as long as it relates to the driving of the vehicle. For example, it may be information on speed, acceleration, angle, current position of own vehicle, position of obstacle. Further, it may be information on the biological information of the driver.
Further, the sensor information does not necessarily have to represent a physical quantity. For example, it may be an image acquired by a camera, a feature amount extracted from the image, or the like.

Next, the information providing device 20 will be described.
The information providing device 20 is a device that collects sensor information from the sensor 10, acquires information to be provided to the driver based on the sensor information (hereinafter referred to as support information), and outputs the information. The information providing device 20 may be an independent device mounted on the vehicle, or may be realized by software built in a car navigation device or the like. Further, it may be realized by software running on a small computer such as a smartphone, a mobile phone, a tablet terminal, a personal information terminal, or a wearable computer (smart watch or the like) used in a vehicle.

The sensor information acquisition unit 21 is a means for acquiring sensor information (a set of sensor information acquired from a plurality of sensors) output by the sensor 10.

The support information generation unit 22 is a means for generating support information based on the acquired sensor information. Assistance information is information provided to a driver to assist driving, typically information indicating the condition of the vehicle, information indicating the surrounding traffic conditions, information indicating the condition of the driver, and the like. However, it may be other than this.
Examples of information indicating the state of the vehicle include information on the maintenance state of the lane, information on the speed, and the like. In addition, as information indicating the surrounding traffic conditions, information on obstacles existing in front or on the side, information on objects to be noted (pedestrians, etc.), and the like can be mentioned. In addition, as information indicating the driver's state, information on an inattentive state, a drowsy state, a dozing state, and the like can be mentioned. Of course, other information may be used as support information.

The input / output unit 23 is a means for receiving an input operation performed by the user and presenting information to the user. Information may be presented by, for example, a screen or voice. Further, the input / output unit 23 may be an interface with an external device that inputs / outputs information. For example, it may be a means of communicating with a robot that interacts with the driver by voice.

Next, the process performed by the information providing device 10 will be described with reference to FIG. FIG. 2 is a flowchart showing a process in which the support information generation unit 22 generates support information (advice in driving operation) based on the collected sensor information and provides it to the driver.

First, in step S11, sensor information is collected. In the present embodiment, the sensor information acquisition unit 21 periodically collects the sensor information and transmits it to the support information generation unit 22 at a predetermined timing. The transmitted sensor information is temporarily accumulated in the support information generation unit 22.

Next, in step S12, the support information generation unit 22 generates support information based on the received sensor information. The support information generation unit 22 stores a plurality of models representing the relationship between the sensor information and the support information, and generates support information based on the model.
For example, when it is determined that the direction of the driver's line of sight acquired from the line-of-sight sensor deviates from the front for 3 seconds or more, caution information for inattentive driving is generated.
In another example, for example, pattern matching is performed between the image acquired by the camera and the road sign image stored in advance, and when they match, information indicating the existence of the road sign is generated.

Next, in step S13, the support information generation unit 22 calculates the certainty of the support information. FIG. 3 is a diagram illustrating the support information to be generated and the degree of certainty thereof.

Here, an example of paying attention to inattentive driving based on the information output by the line-of-sight sensor will be described. For example, based on the output of the line-of-sight sensor, it is possible to recognize that the driver's line of sight has deviated for 3 seconds or more with a probability of 80%. In this case, the certainty of the support information "Caution information for inattentive driving" is 80%.
Further, consider a case where the road sign is recognized based on the image acquired by the camera and the driver is notified by voice. Here, it is assumed that the road sign can be correctly recognized with a probability of 70%. That is, the certainty of the support information "road sign" is 70%.
Such certainty of each support information is stored in advance in the support information generation unit 22.

Here, an example is given in which the sensor information used is one type, but when the support information is generated using a plurality of sensor information, the information on the accuracy of the sensor information used is acquired and the contribution rate is obtained. Weighting may be performed according to.

Next, in step S14, the expression rate is calculated. The expression rate is the probability of providing the support information to the driver when the support information is generated (that is, when the target event occurs).
In the present embodiment, when the certainty of the support information expressed in parts per 100 is A and the expression rate is B, the expression rate is set so that A × B (hereinafter, correct answer probability) is 40%. The correct answer probability is the probability that the driver will receive the correct information when a trigger that generates support information occurs.

For example, when the certainty of the support information is 80%, the expression rate is 50%. In the case of this example, since the certainty (that is, the probability that the generated information is correct) is 80% and the expression rate is 50%, the correct information can be provided to the driver with a probability of 40%.
When the certainty of the support information is 100%, the expression rate is 40%. Similarly, in the case of this example, the correct information can be provided to the driver with a probability of 40%.

As can be seen from FIG. 3, in the information providing device according to the present embodiment, the expression rate is adjusted so that the probability that the correct information can be provided to the driver is always the same for all the support information.
In the first embodiment, since the conviction is a fixed value, a pre-calculated expression rate is used, but when the conviction fluctuates due to an external factor, the expression rate is calculated each time. May be good.

Next, in step S15, it is determined whether or not to provide the information to the driver based on the calculated correct answer probability. For example, a positive judgment or a negative judgment is made based on the generated random number and the correct answer probability. As a result, when an affirmative determination is made, the process proceeds to step S16, and the input / output unit 23 outputs support information. For example, provide the driver with voice guidance such as "Be careful when driving aside."

When the information provision is completed in step S16 or when a negative determination is made in step S15, the process returns to step S11.

As described above, according to the first embodiment, the probability of providing information (expression) so that the correct information can always be provided to the driver with the same probability based on the certainty of the information to be provided. Rate) is calculated. This ensures that the driver always receives the correct information with the same probability, regardless of the certainty of the support information.

If the confidence in the support information is different each time, for example, if the information with high (low) confidence is provided in succession and then the information with low (high) confidence is provided, it is due to familiarity. As a result, the driver may take inappropriate actions, but according to the information providing system according to the present embodiment, this can be suppressed. Further, since the driver does not have to think about the variation in the accuracy of the provided information, there is an advantage that the driver is not burdened excessively.

(Second embodiment)
In the first embodiment, the confidence level for each support information is set by using the information regarding the accuracy of the sensor. On the other hand, some types of sensors exhibit different accuracy in different environments. For example, if there is a sensor that detects an obstacle based on a visible light image acquired by a camera, its accuracy changes depending on the ambient illuminance. The second embodiment is an embodiment in which the accuracy of the sensor, which changes according to the environment, is corrected in order to cope with this.

In the second embodiment, as shown in FIG. 4, different accuracy is defined for the same sensor depending on the environmental conditions. In addition, "type: 1" in the figure means that it does not depend on the environmental condition, and "type: 2" means that it depends on the environmental condition.
In this example, two types of accuracy are defined depending on whether or not the illuminance around the vehicle exceeds the threshold value. That is, if the illuminance exceeds the threshold, the accuracy is 80% (that is, the certainty of the support information is 80%), and otherwise, the accuracy is 50% (that is, the certainty of the support information is 50%). ) Is treated as. It should be noted that which environmental condition the current environment applies to may be determined based on the sensor information acquired from another independent sensor (for example, the illuminance sensor installed outside the vehicle), or the sensor information already acquired. The judgment may be made based on (for example, a camera image).

According to the second embodiment, the certainty of the support information that changes due to changes in the environment can be appropriately corrected.

(Third embodiment)
In the first and second embodiments, the support information was generated directly from the sensor information. On the other hand, the third embodiment is an embodiment in which support information is generated by using machine learning.

FIG. 5 is a system configuration diagram of the information providing device 20 according to the third embodiment. The information providing device 20 according to the third embodiment has a control unit 24 and a classification unit 25 instead of the support information generation unit 22 described in the first embodiment.

The information providing device according to the third embodiment has (1) a phase in which the classifier is learned using the collected sensor information (hereinafter referred to as a learning phase), and (2) has already learned the collected sensor information. Classification is performed using a classifier, and a phase of providing information to the driver based on the classification result (hereinafter, information provision phase) is executed.

The control unit 24 is a means for controlling the entire information providing device. In the learning phase, the control unit 24 executes a process of transmitting the acquired sensor information to the classification unit 25 described later, and in the information providing phase, after transmitting the acquired sensor information to the classification unit 25, the control unit 24 executes the process. A process of receiving the classification result transmitted from the classification unit 25 and generating support information based on the result is executed.

The classification unit 25 has a plurality of two-class classifiers, and is a means for classifying based on the acquired sensor information. In the learning phase, the classification unit 25 performs learning based on the acquired sensor information and executes a process of generating a model (classifier) for classifying. Further, in the information providing phase, a process of classifying the class using the acquired sensor information and transmitting the classification result to the control unit 24 is executed.

Next, the details of the learning phase will be described. The learning phase is a process in which the classification unit 25 learns based on the collected sensor information and generates a classifier. FIG. 6 is a flowchart of processing executed by the classification unit 25 in the learning phase.

First, sensor information is collected from each sensor provided in the vehicle (step S21). In the present embodiment, the sensor information acquisition unit 21 periodically collects the sensor information and transmits it to the classification unit 25 at a predetermined timing. The transmitted sensor information is temporarily accumulated by the classification unit 25. The sensor information collected is a set of a plurality of data. For example, it may be a set of information representing the behavior of the vehicle (speed, front-rear acceleration, left-right acceleration, etc.), vehicle position information, and an image (or a feature amount obtained from the image) acquired by a camera. Further, the sensor information may be information in a time series format corresponding to a predetermined period.

Next, the accumulated sensor information is used as learning data to generate a classifier (step S22). Specifically, a model is generated by inputting sensor information (learning data). The class to be classified is, for example, "presence or absence of pedestrians", "necessity of lane change", "presence or absence of traffic jam", etc., but a binary class (presence or absence of a certain event) can be estimated based on sensor information. Anything can be used as long as it is a thing.

For example, using time-series data representing the steering angle, the presence or absence of blinkers, and position information, it is possible to generate a model for estimating "whether there is a place where lane changes frequently occur nearby". The teacher data may be given from the outside or may be estimated from the sensor information. In the case of the above-mentioned example, it is possible to determine whether or not there is a lane change from the steering angle and the presence or absence of the blinker, and use it as teacher data. In this case, by inputting the position information, it is possible to obtain a two-class classifier that outputs the necessity of changing lanes.

The learning phase may be performed only once or a plurality of times. If it is performed multiple times, the classifier may be regenerated or updated.

Next, the details of the information providing phase will be described with reference to FIG. 7, which is a flowchart showing the processing in the information providing phase. In the information providing phase, the classification unit 25 classifies the class based on the collected sensor information, and the information providing device 20 provides the driver with information (advice in driving operation) based on the result.

First, in step S31, the sensor information is received. In the present embodiment, the sensor information acquisition unit 21 periodically collects the sensor information and transmits it to the classification unit 25 at a predetermined timing. The transmitted sensor information is temporarily accumulated in the classification unit 25.
The collected sensor information is a set of a plurality of data as in the learning phase, but it does not have to match the items used in the learning phase. For example, in the learning phase, the position information of the vehicle, the steering angle, and the presence / absence of the blinker may be used, and in the information providing phase, only the position information of the vehicle may be used. Further, the sensor information may be information in a time series format corresponding to a predetermined period.

Next, in step S32, the classification unit 25 inputs the received sensor information into the learned classifier and acquires the result of class classification.
Next, in step S33, the classification unit 25 calculates the certainty of the support information. The certainty of the support information can be determined based on the likelihood (probability) output by the classifier. Although an example of using the likelihood output by the classifier is given here, the conviction may be calculated based on other information.
Since the processes of steps S34 to 36 are the same as the processes of steps S14 to 16, detailed description thereof will be omitted.

According to the third embodiment, the probability of correct answer can be determined more accurately. Moreover, more complicated events can be detected.

(Fourth Embodiment)
In the first to third embodiments, the expression rate was adjusted so that the probabilities of providing information to the driver were all the same. However, depending on the characteristics of the support information, this response may not be appropriate. For example, the type of support information is "there are few omissions in detection, but there are many false positives". That is, when an event to be detected occurs, support information is generated with a high probability, but the generated support information contains a large amount of noise (event not to be detected). It can be said that such a type of information has a low precision.
In such a case, if the provision probability is matched with other support information, the driver may be troubled by the large number of false alarms and the trust may be impaired.

In order to deal with this, the fourth embodiment is an embodiment in which, when the conformity rate of the support information is low, the correction is performed to suppress the expression rate to a lower level. FIG. 8 is a diagram showing the relationship between the expression rate and the provision probability in the fourth embodiment. In addition, "type: 3" in the figure means that the information has a low precision rate.
In this example, the expression rate is suppressed to 10% and the correct answer probability is set to 3% for the support information having an accuracy of 30%. In this way, by intentionally lowering the expression rate of information having a low precision rate, false alarms can be suppressed and reliability can be prevented from being impaired.

(Modification example)
The above embodiment is merely an example, and the present invention can be appropriately modified and implemented without departing from the gist thereof.
For example, in the third embodiment, a two-class classifier is used, but a multi-class classifier may be used. Similarly, in the description of the embodiment, the result of detecting the presence or absence of a certain event is used as the support information, but the support information may be other than this. Further, the support information may be generated by using a device other than the classifier.

Further, when the conformity rate of the support information is low, the information may be provided in a non-assertive expression as compared with the case where the conformity rate is high. For example, the expression "There are pedestrians. Please be careful" may be changed to the expression "There may be pedestrians."

In addition, the correct answer probability does not necessarily have to be the same throughout all the support information. For example, different values may be used depending on the category of support information.

10 ... Sensor 20 ... Information providing device 21 ... Sensor information acquisition unit 22 ... Support information generation unit 23 ... Input / output unit

Claims (6)

An acquisition means for acquiring sensor information from one or more sensors mounted on a vehicle,
A generation means for generating support information, which is information to be provided to a user, based on the sensor information.
An information providing means for calculating the expression rate, which is the probability of providing the support information to the user, based on the certainty of the support information, and providing the support information based on the expression rate.
Have a,
When the generation means generates support information, the information providing means sets the expression rate lower when the certainty of the support information is high than when the certainty of the support information is low. ,
Information providing device. The information providing means calculates the certainty of the support information based on the accuracy of the sensor information obtained from the one or more sensors.
The information providing device according to claim 1 . The support information includes a first information type, which is an information type whose conviction varies depending on whether or not it is in a specific situation.
When the support information is the first information type, the information providing means acquires information indicating whether or not the support information is in the specific situation, and determines the certainty degree based on the information.
The information providing device according to claim 1 or 2. The support information includes a second information type in which the precision rate is lower than the first threshold value.
The information providing means provides the support information in a non-conclusive expression when the support information is the second information type.
The information providing device according to any one of claims 1 to 3 . The information providing means sets the expression rate lower when the support information is of the second information type than when the support information is other than the second information type.
The information providing device according to claim 4 . An acquisition step to acquire sensor information from one or more sensors mounted on the vehicle,
A generation step of generating support information, which is information to be provided to the user, based on the sensor information.
Based on the certainty of the support information, the expression rate, which is the probability of providing the support information to the user, is calculated, and the information providing step of providing the support information based on the expression rate, and the information providing step.
Only including,
In the information providing step, when the support information is generated in the generation step, when the certainty of the support information is high, the expression rate is set lower than when the certainty of the support information is low. ,
Information provision method.

Images