DE19937490B4 - Method and device for inputting control commands for comfort devices, in particular in motor vehicles - Google Patents

Abstract

A method for inputting control commands for comfort devices by means of a controller connected to at least one database, an input unit and an output unit, wherein the user makes an abnormal command input via a first interaction, the controller (4) breaks the abnormal command input into command attributes, Searching database (5, 7) for the detected command attributes and checks the result for uniqueness, wherein the control unit (4) asks the user for any other command attributes until the search result is unambiguous, the order of input of the other command attributes by the user is set free, and generating a normalized command input from the search result for the respective comfort device.

Description

The invention relates to a method and a device for inputting control commands for comfort devices.

Electric comfort devices require control commands in a predefined manner and in a specific format. If an entered control command differs only slightly from the format, the device can not usually identify and execute the control command. However, this rigid structure contradicts the usual human communication, so that man has to adapt to the machine. This often causes problems when the user can not remember the formats because his alternative inputs are ignored by the machine. To reduce these difficulties, great efforts have been made in the area of user guidance. These user guides, for example, request successive inputs until the complete set of instructions is available, whereby the type of user guidance is again based primarily on the needs of the machine. An example where these problems are very disturbing is the destination input in navigation devices.

From the WO 98/27 531 A1 a method is known for identifying a name by means of a terminal, in particular for identifying a retrieved name of a start or destination of a route in a navigation terminal, based on a plurality of characters queried in the terminal as a string which are part of the name, wherein it is determined whether the retrieved character string corresponds to the beginning of a character string included in a record list, the latter character string being included in the plural names, and upon determining that the retrieved character string is included in an exception list, the user is prompted to input further characters of the name, while a determination that the retrieved string is not included in the exception list, a method for a one-to-one identification without entering other characters of the name is initiated.

From the DE 197 09 518 C1 A method is known for inputting speech to a destination address in a real-time route guidance system in which input speech utterances of a user are recognized by speech recognition means and classified according to their recognition likelihood and the speech uttering the largest recognition likelihood is identified as the input utterance, wherein at least one utterance is a valid one Speech command is, which activates the voice commands associated with this control functions of the route guidance system, all permissible speech utterances are stored on at least one database, at least one operating function of the route guidance system comprises at least one input dialog, wherein after activation of the at least one operating function of the route guidance system depending on the at least one Input dialogue from the permissible S stored on at least one database In real terms, at least one encyclopedia is generated in real time and then the at least one lexicon is loaded as a vocabulary in the speech recognition device. In addition to improved speech recognition, procedures for resolving ambiguities ("Neustadt" problem) are described. Furthermore, the system also offers the use of keywords under which a destination address can be stored and retrieved. However, this highly advanced method does not disclose any hints for improving the communication between man and machine.

The DE 196 15 693 C1 shows an apparatus and method for action determination. Based on digital action information passed by a user of the device, an action is first determined from a predeterminable number of actions using action hints comprising the action information. To determine at least one action to be performed from the action note, a first set of actions in which all determined action notes match at least a portion of the key terms associated with an action, and a second set of actions where at least a portion of the action notes have at least one portion the keywords associated with an action match. The first set of actions and / or the second set of actions is / are used as the basis for uniquely identifying one or more actions. From the action information, the action parameters are determined for each of the determined actions.

Goal of the revelation in the DE 44 32 632 A1 is to allow for an automatic inquiry system as natural as possible queries that are not subject to restrictions that would have to be communicated to the user beforehand. For this purpose, the system is divided into several sections, which have unique interfaces and, moreover, operate essentially independently of one another. A first section is used to recognize words and outputs a word graph as an interface. A second section forms from the word graph a concept graph in which the word graph is searched essentially only for those word sequences from which values for a database query can be derived, and such word sequences or individual words are replaced by concepts in the concept graph. In another section, the Concepts in the concept graph bridged and filled in gaps between the concepts. In doing so, a language model is taken into account. The determination of the concepts from the word graph and the determination of the meaning of the word sequences in the concepts is done by a stochastic attributed grammar. Since irrelevant filler words are neglected early in the spoken request, the computational effort is relatively low, although freely spoken queries are allowed.

• - Parsen des kombinierten multimodalen Inputs;
• - Semantische Interpretation;
• - Dialog, durch den der Nutzer Feedback erhält, inwieweit das System die Nutzeranfrage verstanden hat, wobei durch Interaktion mit dem Nutzer das Verständnis der Anfrage, insbesondere zu fehlender Information und zu Doppeldeutigkeiten, weiter verbessert werden soll;
• - Festlegung, welche Anwendung die Anfrage verarbeiten soll, API Codegenerierung;
• - Anzeige einer Antwort, soweit anwendbar.

The US 5,748,974 A shows a multimodal natural-language interface and associated method for a computer system that interprets user requests. In doing so, natural language input from the user is combined with selected information from a running application, and then the user request is sent in a rendered form to a responsible application for processing. The multimodal natural-language interface performs the following steps:

• - parsing the combined multimodal input;
• - semantic interpretation;
• - Dialogue, through which the user receives feedback, to what extent the system has understood the user request, whereby the interaction of the user should further improve the understanding of the request, in particular for missing information and ambiguity;
• - Determine which application should handle the request, API code generation;
• - Display of an answer, if applicable.

The aforementioned prior art has the disadvantage that the user can not freely determine the sequence of further inputs in case of queries of the respective system, but is guided by the respective system therein.

The invention is therefore based on the technical problem of providing a method and a device for the command input of control commands for comfort devices in motor vehicles, by means of which an improved man-machine communication is achieved.

The solution of the technical problem results from the objects with the features of claims 1 and 6. Further advantageous embodiments emerge from the dependent claims.

According to the invention, the user can make any abnormal command input. This abnormal command input is examined in the controller for command attributes. The database is then searched for the found command attributes. This search result is now checked for uniqueness by the ECU. If the examination result is unambiguous, the control unit forms a formatted control command and transmits it to the relevant comfort device. However, if the result is ambiguous, the user will be prompted to enter additional command attributes until the search result is unambiguous. What matters is that the user is not required to specify the order of the command attributes, i. the user largely specifies the search strategy, for which he preferably chooses command attributes known to him. As a result, almost a human communication between man and machine is achieved.

In another preferred embodiment, the command attributes are user-selectable, i. this is not only free in terms of order, but also in terms of search characteristics. For ambiguities, the approach of the system is dependent on the number, i. with low ambiguities the user can successively be offered the search result for selection, whereas with larger ambiguities further command attributes are requested. Since the database can only search for command attributes, which are also included in this database, the database is designed to be extensible, i. the user can assign additional attributes to individual records.

The invention will be explained in more detail below with reference to a preferred embodiment. The single figure shows a schematic block diagram of a device for command input of control commands for comfort devices in motor vehicles.

The device 1 includes an input unit 2 , an output unit 3 , a control unit 4 , an internal database 5 and a GSM module 6 via which the control unit 4 Access to an external database 7 Has. The input unit 2 , which is preferably designed as a speech recognition device, is connected to an input of the control unit 4 connected. The output side is the control unit 4 with the output unit 4 , which is formed for example as a speaker and / or as a display unit, and at least one to be controlled comfort device 8th connected. Furthermore, the controller has 4 Access to the internal database 5 which, for example, can be searched for command attributes using known search engines. The internal database 5 is designed for example as a CD-ROM. About the input unit 2 Now a user can give an unnormalized command. This command is in the controller 4 analyzed for command attributes. These detected command attributes serve as search criteria, ie the control unit searches the database 5 or, if applicable, the external database 7 ,

Using some exemplary dialogs for destination input for a navigation system, the abstract terms are clarified somewhat.

Example 1:

"I want to see Verena Zucker."

The control unit 4 recognizes the command attribute "Verena Zucker" and searches the database with this command attribute 5 , Since there is only one Verena Zucker, living in Berlin, in Germany, the search result is one-to-one and the assigned address can be transferred to the navigation system.

Example 2:

"I would like to become the inventor of the German patent application DE 197 XX XXX."

In this command input are two interdependent command attributes, namely "Patent Application DE 197 XX XXX" and "Inventor". First, it may happen that the system does not recognize the command attributes. In such cases, an expansion command must be entered into the controller so that the controller 4 recognize these command attributes in the future. Another problem is that the database 5 has an information base for this command attribute. There are two different solutions that can be used in parallel. On the one hand, the database can be extended manually or it can be set to a suitable external database 7 resorted. If, for example, a command attribute "conical-brother" is to be used, "conical-brother" can be inserted in a personal address directory behind the corresponding name, ie the user can freely choose command attributes and then adapts the data base of the database accordingly 5 at.

Example 3:

"I want to go to Franz Gerber."

• „Was wissen Sie von Franz Gerber?“

The control unit 4 captures the command attribute "Franz Gerber" and searches the database with this search term. The control unit detects 4 in that there are a multiplicity of possible destination addresses. Therefore, the user is prompted for additional command attributes, for example in the form:

• "What do you know about Franz Gerber?"

• „Zahnarzt“ oder „Beruf: Zahnarzt“.

The user can now think about what he still knows about "Franz Gerber", for example, that this dentist is. The user can then enter:

• "Dentist" or "Occupation: Dentist".

• „Franz Gerber, Zahnarzt, München;
• Franz Gerber Zahnarzt, Frankfurt;
• Franz Gerber, Zahnarzt, Osnabrück,"

The control unit 4 then perform a search with the conjunction "Franz Gerber" and "Dentist". If the result is unambiguous, the address is transferred to the navigation system. If, for example, three possible target persons remain open, they can be offered for selection:

• Franz Gerber, Dentist, Munich;
• Franz Gerber Dentist, Frankfurt;
• Franz Gerber, Dentist, Osnabrück, "

• „Franz Gerber, Zahnarzt, Osnabrück“
• auswählen. Ist die Mehrdeutigkeit noch größer, beispielsweise zehn Personen, so kann nach weiteren Befehlsattributen gefragt werden.

If the user knows that the wanted Franz Gerber lives somewhere in Lower Saxony, he knows how to do it

• "Franz Gerber, Dentist, Osnabrück"
• choose. If the ambiguity is even greater, for example ten people, you can ask for further command attributes.

In addition to the destination input for navigation devices, the device can also be used to operate a car phone, radio or air conditioning units.

Claims (10)

A method for inputting control commands for comfort devices by means of a controller connected to at least one database, an input unit and an output unit, wherein the user makes an abnormal command input via a first interaction, the controller (4) breaks the abnormal command input into command attributes, Searching database (5, 7) for the detected command attributes and checks the result for uniqueness, wherein the control unit (4) asks the user for any other command attributes until the search result is unambiguous, the order of input of the other command attributes by the user is set free, and generating a normalized command input from the search result for the respective comfort device. Method according to Claim 1 , characterized in that the method is used for comfort devices in motor vehicles and the command attributes are freely selectable by the user. Method according to Claim 1 or 2 , characterized in that in case of ambiguity depending on the number of ambiguities, a successive query is performed or an additional command attribute is required. Method according to one of the preceding claims, characterized in that the Database (5, 7) with freely selectable command attributes can be expanded. Method according to one of the preceding claims, characterized in that the method is used for the destination input of a navigation system. Device for the command input of control commands for comfort devices (8) comprising a control unit (4) with a speech analysis algorithm, at least one input unit (2) and an output unit (3), which are each connected to the control unit (4), and at least one database (5 , 7) accessible to the controller (4), the controller (4) being arranged to decompose abnormal command inputs into command attributes, the database (5, 7) being scanned by the controller (4) for the command attributes detected and the result is checked by the controller (4) for uniqueness, the controller (4) asking the user for any further command attributes until the search result is unambiguous, the order of entry of the further command attributes being freely determined by the user is, wherein the control unit (4) a normalized command input from the search result for the respective Komfortger ät (8) generated. Device after Claim 6 , characterized in that the device is provided for comfort devices in motor vehicles and the control device (4) via a GSM module (6) with an external database (7) is connected. Device after Claim 6 or 7 , characterized in that the input unit (2) is designed as a speech recognition device. Device according to one of Claims 6 to 8th , characterized in that the output unit (3) is designed as a voice output device and / or as a display unit. Device after Claim 9 , characterized in that the display unit is designed as a touch screen.

Images