A method for controlling a lighting system according to a lighting scene

Abstract

A method for controlling a lighting system according to a lighting scene. The method comprises receiving a first user query, said first user query for controlling the lighting system and generating a first lighting scene based on the first user query. The method further comprises controlling the lighting system according to the first lighting scene. The method further comprises receiving a second user query and determining a semantic similarity between the first and the second user query that is above a similarity threshold based on a comparison of the first with the second user query. The method further comprises determining a time interval between the receiving of the first and the second user query and determining, based on the determined time interval, whether to control the lighting system according to the first lighting scene generated for the first user query or generate a second lighting scene, different from the first lighting scene, based on the second user query and control the lighting system according to the second lighting scene.

Description

[0001] A method for controlling a lighting system according to a lighting scene.

[0002] FIELD OF THE INVENTION

[0003] The invention relates to a method for controlling a lighting system according to a lighting scene. The invention further relates to a control system for controlling a lighting system according to a generated lighting scene, said lighting scene generated based on a user query. The invention further relates to computer program product for controlling a lighting according to a lighting scene.

[0004] BACKGROUND OF THE INVENTION

[0005] Lighting control systems have been in use for many years to create different lighting scenes for different situations, such as work, relaxation, or entertainment. These systems have traditionally relied on pre-defined rules or manual adjustments to generate the desired lighting scene. However, with recent advances in artificial intelligence (Al) and machine learning, it has become possible to generate lighting scenes using Gen Al models. These models are trained on vast amounts of data and can generate lighting scenes based on a varity of factors, such as user preferences and environmental conditions.

[0006] However, one of the main challenges with Gen Al generated scenes is the variability of outputs generated by the model. Due to the characteristics of large language models, different lighting scenes can be generated for the same prompt.

[0007] This variability can lead to inconsistencies in user experience and frustration for users. For example, a user might expect a certain lighting scene to be generated based on their input, but instead get a different scene that does not meet their expectations. On the other hand, in some cases users may expect or prefer a different lighting scene. For example, they may not be satisfied by the scene generated or they may prefer to be surprised by a different scene.

[0008] SUMMARY OF THE INVENTION

[0009] The inventors have realized that while at some cases it is desired to generate consistent outputs for the same command, in other cases, a user repeating a command may expect a different lighting scene. For example, the user may not be satisfied with the lighting scene previously generated.

[0010] It is therefore an object to provide a system that can provide improved user experience by providing appropriate lighting scenes based on user preferences.

[0011] According to a first aspect, the object is achieved by a method for controlling a lighting system based on a lighting scene, said lighting scene generated based on a user query. The method comprises receiving a first user query for controlling the lighting system. A user query refers to a request or input made by a user to control the lighting system. The user query can be in the form of an audio utterance comprising one or more words, in the form of text utterance comprising one or more words, in the form of an image, a gesture, etc. The method further comprises generating a lighting scene based on the first user query. Several methods may be used to generate a lighting scene based on a user query. For example, generating a lighting scene may comprise searching a (predefined) knowledge base that maps keywords present in the user query to certain lighting properties, searching for images and finding lighting properties in these images, etc. In another example, generating a lighting scene may comprise using a generative model, such as a Large Language Model (LLM) to generate the lighting scene. The method further comprises controlling the lighting system according to the first lighting scene. That is, the method comprises setting the parameters of the lighting devices of the lighting system according to lighting control parameters defined in the lighting scene. The method further comprises receiving a second user query, said second user query for controlling the lighting system. The method further comprises determining a semantic similarity between said first and second user query based on a comparison of the first with the second user query.

[0012] If said semantic similarity (or similarity score) is above a (predetermined) similarity threshold, the method comprises determining a time interval between the receiving of the first and the second user query and determining based on the determined time interval, whether to control the lighting system according to the first lighting scene generated for the first user query or generate a second lighting scene, different from the first lighting scene, based on the second user query and control the lighting system according to the second lighting scene.

[0013] If said semantic similarity (or similarity score) is below the (predetermined) similarity threshold, the method may comprise generating the second lighting scene, different from the first light scene, based on the second user query and control the lighting system according to the second lighting scene. That is, for a new (unseen) user query, a new lighting scene is generated that captures the intent of the user based on the user query.

[0014] With the advent of large language models for lighting scene scene generation, inputs that are semantically identical may produce different outputs at different times. This can pose a challenge when users expect a consistent experience based on a specific command. While existing lighting control systems attempt to generate the same lighting scene for a given command to ensure consistency, in reality, users may desire different lighting experiences when repeating the same command. For example, a user may not be satisfied with the lighting scene generated from their previous command or may simply wish to enjoy a different experience. The proposed method aims to address this issue by determining whether a user who submits a repeated command is interested in seeing the same lighting scene as before or a completely different one. As a result, the user experience is enhanced, leading to greater satisfaction. Said method may be performed by software running on a programmable device. This software may be provided as a computer program product.

[0015] The method may comprise that if said time interval is above a predetermined time threshold, controlling the lighting system according to the first lighting scene generated for the first user query. If said time interval is below the predetermined time threshold, generating a second lighting scene, different from the first light scene, based on the second user query and controlling the lighting system according to the second lighting scene. For instance, consider a case where a user requests to be energized, but the system generates an orange lighting scene that fails to satisfy them. If the user repeats the same input immediately, the system should generate a different lighting scene, such as a fire scene. If the user is pleased with the fire scene, they may not provide any further inputs. However, if the user repeats the command "I want to be energized" after a predetermined time interval, they may expect the same fire scene. Thus, the same repetitive inputs can result in different lighting control commands.

[0016] The determination of whether to control the lighting system according to the first lighting scene generated for the first user query or generate a second lighting scene, different from the first light scene, based on the second user query and control the lighting system according to the second lighting scene may further be based on a time of day of the reception of the first and / or the second user query. For example, consider a first user query 'I want to be energized' received in the morning, and the corresponding lighting scene generated. A second user query is received in the evening that is semantically similar to the first one (above the predetermined similarity threshold). Since the 'energize' experience is associated with the time of day, a new (second) lighting scene should be generated, and the corresponding lighting control command should be tailored to the evening setting instead of using the previously generated first lighting scene. This applies even if a sufficient amount of time has elapsed since the reception of the first query.

[0017] The method may further comprise determining the reception of a further user query during the time interval and the step of determining of whether to control the lighting system according to the first lighting scene generated for the first user query or generate a second lighting scene, different from the first light scene, based on the second user query and control the lighting system according to the second lighting scene may be further based on the reception of the further user query. For example, if there are more user queries in between the first and the second user query, it may be assumed that the user expects the previous lighting scene to be generated. Thus, in that case, the method may comprise controlling the lighting system according to the first lighting scene generated for the first user query.

[0018] The method may further comprise analyzing the second user query to determine an emotional state of the user. The determination of whether to control the lighting system according to the first lighting scene generated for the first user query or generate a second lighting scene, different from the first lighting scene, based on the second user query and control the lighting system according to the second lighting scene may be further based on the determined emotional state. For example, if the determined emotional state of the user is a state indicative of a negative emotion e.g., “angry” or “negative” tone, the method may comprise generating a second lighting scene and controlling the lighting system accordingly. The negative emotional state may be indicative that the user is not satisfied with the first generated user scene. On the other hand, if the determined emotional state of the user is a state indicative of a positive / neutral emotion e.g., “happy” or “neutral” tone, the method may comprise controlling the lighting system according to the first lighting scene as the positive state may be indicative that the user was satisfied with the first generated lighting scene.

[0019] The first and the second lighting scenes may be generated by a generative machine learning model, for example an LLM. Said generative model may run locally on a computing (edge) device or on the cloud. The method may comprise generating a prompt as input for the generative machine learning model. The prompt comprises at least said user querry (first or second respectively) and optionally further data indicative of one or more control characteristics of the lighting system, such as device types, device control parameters, LLM output format, relevant sensor information, information about the user (e.g., elderly, teenager, special medical conditions, etc.), preferences and habits of this specific user, etc. The prompt is transmitted to said generative machine learning model causing said generative model machine learning model to generate a lighting scene.

[0020] The method may further comprise storing the first lighting scene in a scene database comprising user queries and their corresponding generated lighting scenes. The system may access the scene database comprising user queries and corresponding generated lighting scenes. The method may further comprise accessing the scene database to obtain a lighting scene from the generated lighting scenes in the scene database for which a semantic similarity between the first user query and the user query corresponding to the lighting scene is above the similarity threshold and using the obtained lighting scene to control the lighting system. For example, every new user query received may be compared against the user queries stored in the scene database. If a semantic similarity between the new (second) user query and at least one of the user queries stored in the scene database is determined to be above the similarity threshold, the method may comprise using the lighting scene corresponding to the at least one user query to control the lighting system.

[0021] According to a second aspect, the object is achieved by a control system for controlling a lighting system based on a generated lighting scene. The lighting scene is generated based on a user query. The control system comprises: a user interface and one or more processors or controllers. The one or more processors are configured to receive a first user query, via the user input interface, said first user query for controlling the lighting system. The one or more processors are further configured to receive, via the input interface, a lighting scene generated based on the first user query. The lighting scene may be generated by a further system, such as a generative machine learning model operating in the cloud or by the one or more processors of the control system. That is, the one or more controllers are configured to cause the generation of the first lighting scene (generated locally or externally by a further system) and receive the generated lighting scene. The one or more processors are further configured to control the lighting system according to a first lighting scene generated based on the first user query. The one or more processors are configured to receive, via the input interface, a second user query for controlling the lighting system, compare the first with the second user query and determine, based on the comparison, a semantic similarity between the first and the second user query. If the semantic similarity is below a (predetermined) similarity threshold, the one or more processors may be configured to receive a second lighting scene generated based on the second user query. The second lighting scene may be generated by a further system, such as a generative model operating on the cloud, or by the one or more processors of the control system. That is, the one or more controllers are configured to cause the generation of the second lighting scene (generated locally or externally by a further system) and receive the generated lighting scene. If the semantic similarity is above the (predetermined) similarity threshold, the one or more processors are configured to determine a time interval between the receiving of the first and the second user query and determine, based on the determined time interval, whether to control the lighting system according to the first lighting scene generated for the first user query, or, control the lighting system according to a second lighting scene, different from the first lighting scene, said second lighting scene generated based on the second user query.

[0022] According to a third aspect, the object is achieved by a computer program product for a computing device, the computer program product comprising computer program code to perform the methods described herein. A computer program may, for example, be downloaded by or uploaded to an existing device or be stored upon manufacturing of these systems.

[0023] A non-transitory computer-readable storage medium stores at least one software code portion, the software code portion, when executed or processed by a computer, being configured to perform executable operations for generating a lighting scene for a lighting system based on a user query.

[0024] The executable operations comprise receiving a first user query, said query for controlling the lighting system, generating a first lighting scene based on the first user query, controlling the lighting system according to the first lighting scene; receiving a second user query, determining a semantic similarity between the first and the second user query that is above a similarity threshold based on a comparison of the first with the second user query, determining a time interval between the receiving of the first and the second user query, and determining, based on the determined time interval, whether to control the lighting system according to the first lighting scene generated for the first user query or generate a second lighting scene, different from the first light scene, based on the second user query and control the lighting system according to the second lighting scene.

[0025] The method, system, and computer program product may have similar and / or identical embodiments and advantages.

[0026] BRIEF DESCRIPTION OF THE DRAWINGS

[0027] This and other aspects of the present invention will now be described in more detail, with reference to the appended drawings showing embodiment(s) of the invention. Fig. 1 illustrates an example of a control system for controlling a lighting system;

[0028] Fig. 2 shows schematically a method for controlling a lighting system;

[0029] Fig. 3 shows schematically a method for controlling a lighting system;

[0030] Fig. 4 shows schematically a method for controlling a lighting system.

[0031] All the figures are schematic, not necessarily to scale, and generally only show parts which are necessary in order to elucidate the invention, wherein other parts may be omitted or merely suggested.

[0032] DETAILED DESCRIPTION

[0033] Fig. 1 shows an example of a control system 100 for controlling a lighting system 150. The lighting system 150 may comprise one or lighting devices, each may be individually controllable, for example in terms of brightness levels and / or colours to produce indirect, narrow spot, broad or diffuse effects, and possibly of different hues. The overall effect is often called a light or lighting scene or profile, and can alter the atmosphere or "mood" of a space. That is, a lighting scene comprises the settings, i.e., the parameters (e.g., brightness level, hue, beam direction, etc.), that the lighting devices of the lighting system 150 are to be controlled to create a lighting effect, atmosphere, or mood in a space according to a user query.

[0034] The control system 100 comprises an input interface 102 and one or more processor or controllers 106. The system may further comprise at least one data repository or storage or memory 108 for storing computer program code instructions. The controller 106 may be communicatively coupled to the memory module 108. The controller 106 may further be communicatively coupled to the cloud 120. Said memory module 108 may be comprised in the cloud 120. The system 100 may be comprised in a single device, for example, a smart phone device or parts of the system 100 may be compirsed in multiple devices, e.g., a voice assistant device and a mobile phone device, etc.

[0035] The controller 106 is configured to receive, via the input interface 102, a first user query for controlling the lighting system 150. A user query refers to a request or input made by a user to control the lighting system. The user query can be in the form of an audio utterance comprising one or more words. For example, the user may use a voice-enabled personal assistant to command the lighting system to “Generate a sunset in the tropics lighting scene” or “Make me relax”. The user query may be in the form of a text utterance comprising one or more words, etc. For example, the user may type a command, for example, using a smart phone device, to control the lighting system. In another example, the user query can be in the form of an image. For example, the user may select an image of a sunset or a forest to indicate that (s)he wishes to create a lighting scene that mimics the colors and ambiance of the image.

[0036] The control er 106 is configured to receive, via the input interface 102, a first lighting scene generated based on the first user query. The lighting scene comprises (defines) lighting settings (or parameters) that the one or more devices of the lighting system are to be controled to generate a lighting atmosphere in a space according to the user query. The first lighting scene may be a dynamic lighting scene.

[0037] In an example, one or more processors or controllers 106 of the control system 100 may be configured to generate the first lighting scene based on the user query. For example, the controller 106 may use rule-based or internal machine-learning models to generate the first lighting scene. Several methods may be used to generate a lighting scene based on a user query. For example, generating a lighting scene may comprise searching a (predefined) knowledge base that maps keywords present in the user query to certain lighting properties (e.g., system maps “romatic” to a set of pre-defined lighting parameters for the one or more devices of the lighting system 150). In another example, generating a lighting scene may comprise searching for images (e.g., an image of a candle for “romantic”) and finding lighting properties in these images, or using a semantic web technology which enables a web crawler to find relations between a particular keyword present in the user utterance and particular colors, for example “forest-green”, “sea-blue”, “party-colorful”.

[0038] In another example, one or more processors or controllers 106 of the control system 100 may be configured to generate the first lighting scene by using a generative machine learning model. For examle, a large language model operating on the one or more processors 106.

[0039] In another embodiment, the control system 100 may be communicately coupled to a further system 550, for example a generative machine learning model such as a large language model, e.g., the further system 550 operating on the cloud 120. The controller 106 may be configured to cause the further system to generate a lighting scene by generating a prompt, said prompt comprising at least said first user query, and transmitting said prompt to the large language model 550. For example, the language model 550 may identify keywords like "romantic" and "lighting scene", and then use this information to generate a response (output a lighting scene) that includes specific lighting configurations or parameters for the lighting system. The controller 106 may receive the lighting scene generated by the further system 550.

[0040] The first lighting scene may be stored in a scene database, said scene database comprising user queries and their corresponding generated lighting scenes. That is, every (new) lighting scene that is generated is stored in the scene database together with its corresponding user query for later access and retrieval. Said database may be comprised in a memory, such as memory 108. For example, every row in the scene database comprises a user query and a corresponding lighting scene defining settings for the one or more devices of the lighting system 150. In another example, every row in the scene database may comprise a user query and a high level textual description of the corresponding lighting scene.

[0041] That controller 106 is further configured to control the lighting system 150 according to the first lighting scene. That is, the controller 106 is configured to control the one or more lighting devices of the system 150 according to the lighting settings (or parameters) defined in the first lighting scene.

[0042] The controller 106 is further configured to receive, via the input interface 102, a second user query for controlling the lighting system 150. The controller 106 is further configured to compare the first and second user query to determine a semantic similarity between the first and the second user query. Several methods and techniques are known in prior art for determing a semantic similarity between the user queries. For example, for textbased queries, methods such as Jaccard similarity, Latent Semantic Analysis (LSA), WordNet, and Levenshtein distance can be used to determine the semantic similarity between the first and second user queries. For image and video queries, methods such as cosine similarity, word embeddings, convolutional neural networks (CNNs), etc, may be used. If the semantic similarity between the first and second user query is below the (predetermined) similarity threshold, the controller 106 is configured to control the lighting system according to a second lighting scene, different from the first light scene, generated based on the second user query.

[0043] That is, for every user query that is received (second user query), the controller 106 is configured to first check if there is a “match” (semantic similarity above threshold) with a previous user query e.f. , from the scene database. If there is not such a match, a new lighting scene needs to be generated.

[0044] If there is a match, that is, the semantic similarity between the first and second user query is above the similarity threshold, the controller 106 is configured to determine whether the user expects to see a past lighting scene as generated for the first (similar) user query or a different (second) lighting scene generated for the second user query. That is, if the semantic similarity between the first and second user query is above a (pre-determined) similarity threshold, the controller 106 is configured to 1) determine a time interval between the receiving of the first and the second user query and 2) determine, based on the determined time interval, whether to control the lighting system according to the first lighting scene generated for the first user query or control the lighting system according to a second lighting scene, different from the first light scene, generated based on the second user query. The second lighting scene may be received by the controller 106 via the input interface 102. The second lighting scene generated for the second user query may be stored in the scene database comprising user queries and corresponding generated lighting scenes.

[0045] In an example, if the time interval is above a predetermined time threshold the controller 106 may be configured to control the lighting system according to the first lighting scene. The controller 106 may access the scene database, e.g., through memory 108, to obtain the first lighting scene corresponding to the first user query (which has a semantic similarity with the second) and control the lighting system accordingly. If the time interval is below the predetermined time threshold the controller 106 is configured to cause the generation of a second lighting scene based on the second user query.

[0046] The second lighting scene may be generated locally. For example, one or more processors or controllers 106 of the control system 100 may be configured to generate the second lighting scene based on the second user query. For example, the controller 106 may use rule-based or internal machine-learning models to generate the second lighting scene. In another example, one or more processors or controllers 106 of the control system 100 may be configured to generate the second lighting scene by using a generative machine learning model such as a large language model. The second lighting scene may be generated by a further system. For example, the control system 100 may be communicately coupled to a further system 550, e.g., a generative machine learning model such as a large language model. The further system may be operating on the cloud 120. The controller 106 may be configured to cause the generation of the second lighting scene by generating a prompt, said prompt comprising at least said second user query, and transmitting said prompt to the large language model 550. The second lighting scene may be generated by the large language model 550. The second lighting scene may be a dynamic lighting scene. The second lighting scene may be generated with a different menthod or by a different means from the first lighting scene. Fig. 2 shows an example of a method 200 for controlling a lighting system based on a user query. The method comprises the steps of receiving 202 a first user query, said first user query for controlling the lighting system and generating 204 a first lighting scene based on the first user query. The first lighting scene may be generated locally or on the cloud 120, e.g., using a generative large language model. The method further comprises controlling 206 the lighting system 150 according to the first lighting scene. The method 200 further comprises receiving 207 a second user query, said second user query received at a different (later) point in time from the first user query, said second user query for controlling the lighting system 150. The method 200 further comprises determining 208 a semantic similarity between the first and the second user query based on a comparison between the first and the second user query. If said semantic similarity is above a predetermined similarity threshold, the method comprises the steps of determining 210 a time interval between the receiving of the first and the second user query and determining 212, based on the determined time interval, whether to control the lighting system according to the first lighting scene generated for the first user query or generate a second lighting scene, different from the first lighting scene, based on the second user query and control the lighting system according to the second lighting scene. The second lighting scene may be generated locally or on the cloud 120, e.g., using a generative large language model.

[0047] Fig. 3 shows an example of a method 300 for controling lighting system based on a user query. The method 300 comprises all steps 202-212 of method 200, and step 212 of method 200 is implemented by steps 312-316 of method 300. The method 300 comprises the steps of receiving 302 a first user query, said query for controlling the lighting system and generating 304 a first lighting scene based on the first user query. The method further comprises controlling 306 the lighting system 150 according to the first lighting scene. The method 300 further comprises receiving 307 a second user query, said second user query received at a different (later) point in time from the first user query, said second user query for controlling the lighting system 150. The method 300 further comprises determining 308 a semantic similarity between the first and the second user query based on a comparison between the first and the second user query. If said semantic similarity is above a predetermined similarity threshold, the method comprises the steps of determining 310 a time interval between the receiving of the first and the second user query. If said time interval is above a predetermined time threshold, the method 300 comprises controlling 312 the lighting system 150 according to the first lighting scene generated for the first user query. If said time interval is below the predetermined time threshold, the method 300 comprises generating 314 a second lighting scene, different from the first lighting scene, based on the second user query and controlling 316 the lighting system according to the second lighting scene. Optionally, if said semantic similarity is below the predetermined similarity threshold, method 300 optionally comprises generating 314 a second lighting scene, different from the first light scene, based on the second user query and controlling 316 the lighting system according to the second lighting scene.

[0048] Fig. 4 shows an example of a method 400 for controlling a lighting system 150 based on a user query. The method 400 comprises all steps 302-316 of method 300. Method 400 further comprises determining 409 the receival of a further user query during the time interval between the first and the second user query. If it is determined that a further user query has been received between the first and the second user query, the method 400 comprises controlling 412 the lighting system according to the first lighting scene generated for the first user query independent on whether the time interval between the first and second user query is above or below the predetermined time threshold. If it is determined that there is no further user query received between the first and second queries, the method 400 comprises determining 410 the time interval between the receiving of the first and the second user query and determining whether to control the lighting system according to the first lighting scene or control the lighting system according to the second lighting scene based on the time interval, as shown in the flowchart of Fig. 4.

[0049] The methods of Figs. 2-4 may optionally comprise the step of analyzing the second user query to determine an emotional state of the user. Several method and techniques are available for analyzing a user query to determine an emotional state of the user. For example, speech prosody analysis may be used to analyze the tone, pitch, and rhythm of the user's speech to identify the emotional state. For example, a high-pitched, fast-paced speech may indicate excitement or anxiety, while a slow-paced, low-pitched speech may indicate sadness or depression. In another example, Natural Language Processing (NLP) or sentiment analysis may be used to analyze spoken or written language to understand user’s emotional state. This involves identifying the words, phrases, and patterns of language that indicate a particular emotional state. In an embodiment, if the determined emotional state of the user is a state indicative of a negative emotion e.g., “angry” or “negative” tone, the method may comprise generating a second lighting scene and controlling the lighting system accordingly. The negative emotion state may be indicative that the user is not satisfied with the first generated user scene. On the other hand, if the determined emotional state of the user is a state indicative of a positive / neutral emotion e.g., “happy” or “neutral” tone, the method may comprise controlling the lighting scene according to the first user query as the positive state may be indicative that the user was satisfied with the first generated scene.

[0050] It should be noted that the above-mentioned embodiments illustrate rather than limit the invention, and that those skilled in the art will be able to design many alternative embodiments without departing from the scope of the appended claims.

[0051] In the claims, any reference signs placed between parentheses shall not be construed as limiting the claim. Use of the verb "comprise", and its conjugations does not exclude the presence of elements or steps other than those stated in a claim. The article "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention may be implemented by means of hardware comprising several distinct elements, and by means of a suitably programmed computer or processing unit. In the device claim enumerating several means, several of these means may be embodied by one and the same item of hardware. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage.

[0052] Aspects of the invention may be implemented in a computer program product, which may be a collection of computer program instructions stored on a computer readable storage device which may be executed by a computer. The instructions of the present invention may be in any interpretable or executable code mechanism, including but not limited to scripts, interpretable programs, dynamic link libraries (DLLs) or Java classes. The instructions can be provided as complete executable programs, partial executable programs, as modifications to existing programs (e.g. updates) or extensions for existing programs (e.g. plugins). Moreover, parts of the processing of the present invention may be distributed over multiple computers or processors or even the ‘cloud’.

[0053] Storage media suitable for storing computer program instructions include all forms of nonvolatile memory, including but not limited to EPROM, EEPROM and flash memory devices, magnetic disks such as the internal and external hard disk drives, removable disks, and CD-ROM disks. The computer program product may be distributed on such a storage medium, or may be offered for download through HTTP, FTP, email or through a server connected to a network such as the Internet.

Claims

CLAIMS:

1. A method for controlling a lighting system according to a lighting scene, said method comprising: receiving a first user query, said first user query for controlling the lighting system; generating a first lighting scene based on the first user query; controlling the lighting system according to the first lighting scene; receiving a second user query, said second user query for controlling the lighting system; determining that a semantic similarity between the first and the second user query is above a similarity threshold based on a comparison of the first with the second user query; determining a time interval between the receiving of the first and the second user query; determining, based on the determined time interval, whether to control the lighting system according to the first lighting scene generated for the first user query or generating a second lighting scene, different from the first lighting scene, based on the second user query and control the lighting system according to the second lighting scene.

2. The method according to claim 1, wherein the step of determining, based on the determined time interval, whether to control the lighting system according to the first lighting scene generated for the first user query or generate a second lighting scene, different from the first lighting scene, based on the second user query and control the lighting system according to the second lighting scene comprises: if said time interval is above a predetermined time threshold, controlling the lighting system according to the first lighting scene generated for the first user query; if said time interval is below the predetermined time threshold, generating a second lighting scene, different from the first lighting scene, based on the second user query and controlling the lighting system according to the second lighting scene.

3. The method according to claim 1, wherein the determining of whether to control the lighting system according to the first lighting scene generated for the first user query or generate a second lighting scene, different from the first lighting scene, based on the second user query and control the lighting system according to the second lighting scene is further based on a time of day.

4. The method according to claim 1, wherein the method further comprises: determining the receival of a further user query during the time interval, and wherein, the determining of whether to control the lighting system according to the first lighting scene generated for the first user query or generate a second lighting scene, different from the first lighting scene, based on the second user query and control the lighting system according to the second lighting scene is further based on the receival of the further user query.

5. The method according to claim 1, wherein the method further comprises: analyzing the second user query to determine an emotional state of the user, and wherein, the determining of whether to control the lighting system according to the first lighting scene generated for the first user query or generate a second lighting scene, different from the first lighting scene, based on the second user query and control the lighting system according to the second lighting scene is further based on the determined emotional state of the user.

6. The method according to any preceding claim, wherein the first lighting scene is generated by a generative machine learning model.

7. The method according to any preceding claim, wherein the second lighting setting is generated by a generative machine learning model.

8. The method according to any preceding claim, wherein the first lighting scene is stored in a scene database comprising user queries and corresponding generated lighting scenes.

9. The method according to claim 8, wherein the second lighting scene is stored in the scene database comprising user queries and corresponding generated lighting scenes.

10. The method according to claim 8, wherein the controlling of the lighting system according to the first lighting scene comprises the steps of: accessing the scene database to obtain a lighting scene from the generated lighting scenes in the scene database for which a semantic similarity between the first user query and the user query corresponding to the lighting scene is above the similarity threshold; using the obtained lighting scene to control the lighting system.

11. A control system (100) for controlling a lighting system (150) according to a lighting scene, said control system comprising: an input interface (102); one or more processors (106) configured to: receive, via the input interface, a first user query for controlling the lighting system; generate a first lighting scene based on the first user query; control the lighting system according to a first lighting scene generated based on the first user query; receive, via the input interface, a second user query for controlling the lighting system; determine a semantic similarity between the first and the second user query that is above a similarity threshold based on a comparison of the first with the second user query; determine a time interval between the receiving of the first and the second user query; determine, based on the determined time interval, whether to control the lighting system according to the first lighting scene generated based on the first user query or generate a second lighting scene, different from the first lighting scene, based on the second user query and control the lighting system according to the second lighting scene.

12. A computer program product for a computing device, the computer program product comprising computer program code to perform the method of claim 1 when the computer program product is run on the one or more processor of claim 11.

Images