Destination recommendation system

Abstract

When recommending the same destination to multiple users, the system will consider factors such as congestion at the destination, congestion at nearby parking lots, and congestion along the route to the destination to recommend the most appropriate destination. [Solution] The destination recommendation system 1 comprises a plurality of recommendation agents 10a to 10n that recommend destinations to each user, and a coordination agent 20 that coordinates the plurality of recommendation agents. The coordination agent 20 determines whether or not to present recommendation content for each recommendation agent, or to select recommendation content from the candidate recommendation content, based on the congestion status of the destination, the congestion status of parking lots near the destination, or the congestion status of the route to the destination when the destination was recommended to each recommendation agent, and the current or past effectiveness of each recommendation agent. The coordination agent 20 includes AI.

Description

Technical Field

[0005] , , ,

[0001] The present disclosure relates to a destination recommendation system.

Background Art

[0002] In-vehicle navigation devices have disclosed technologies for recommending destinations or recommending POIs (Points of Interest) that are likely to attract the user's interest on the route to the destination.

[0003] Patent Document 1 discloses quantitatively defining the basic recommendation degree of a pleasure destination according to the total number of destination settings in a number of destination recommendation devices. Also, different main factors for each user are modeled as penalties related to travel time, penalties for tolerance of congestion, and penalties for whether the destination has already been visited. A destination recommendation device that can recommend a destination according to the situation and preference of each user is disclosed.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

[0006] A destination recommendation system in one aspect of the present disclosure comprises a plurality of recommendation agents that recommend destinations to each user, and a coordination agent that coordinates the plurality of recommendation agents, wherein the plurality of recommendation agents estimate user preferences for destinations based on at least one of destination search history, destination visit history, destination setting history, and response history to recommendation content; estimate the user's state and situation based on at least one of vehicle behavior, driving operations, external camera video, internal camera video, and user biometric data; and recommend candidate recommendation content, the utility of the candidate recommendation content, date and time, and user Each of the users' current locations is generated, and the generated recommended content candidates, utility, date and time, and the user's current location are provided to the adjustment agent. The adjustment agent, based on at least one of the following—the congestion status of the destination when the destination was recommended to each recommendation agent, the congestion status of parking lots near the destination, and the congestion status of the route to the destination—and the utility of each recommendation agent, whether or not to present recommended content, or to select recommended content from the recommended content candidates, is provided to each recommendation agent whether or not to present the selected recommended content, and the recommended content itself. Each of the multiple recommendation agents then presents the recommended content to each user based on the determined whether or not to present the recommended content and the recommended content itself. [Effects of the Invention]

[0007] According to this disclosure, when recommending the same destination to multiple users, it is possible to recommend an appropriate destination by taking into account the congestion status of the destination, the congestion status of parking lots near the destination, or the congestion status of the route to the destination. [Brief explanation of the drawing]

[0008] [Figure 1] This diagram shows the configuration of the destination recommendation system according to the embodiment. [Figure 2] This figure illustrates the details of the recommended content candidate generation unit according to the embodiment. [Modes for carrying out the invention]

[0009] Specific embodiments to which the present invention is applied will be described in detail below with reference to the drawings. However, the present invention is not limited to the following embodiments. Also, for clarity of explanation, the following description and drawings have been simplified as appropriate.

[0010] Figure 1 shows the configuration of a destination recommendation system according to an embodiment. The destination recommendation system 1 includes a plurality of recommendation agents 10a-10n and one or more coordination agents 20. Each of the plurality of recommendation agents 10a-10n recommends a destination to the user AN riding in each vehicle via a display unit or speaker.

[0011] Each recommendation agent 10a-10n includes a recommendation content candidate generation unit 11 and a recommendation execution unit 12. Each recommendation agent 10a-10n is implemented by a computer mounted in the vehicle or the central server of each vehicle control system, and has the function of executing various controls based on various programs stored in the memory unit. The vehicle is equipped with an in-vehicle navigation system. The recommendation agent can acquire various information (e.g., vehicle location information, map information) from the in-vehicle navigation system or provide various information to the in-vehicle navigation system via a wired or wireless network.

[0012] The recommended content candidate generation unit 11 generates recommended content candidates (or a list of recommended content candidates) and their utility (score) by considering the user's preferences estimated using the user's past destination search history, setting history, and destination visit history, as well as the user's current location and distance to the destination. The utility function may be pre-set or learned sequentially from the data. Each recommendation agent generates a "recommended content candidate," and the adjustment agent determines the final "recommended content" from among the multiple recommended content candidates.

[0013] Now, with reference to Figure 2, the details of the recommended content candidate generation unit 11 will be explained. Figure 2 is a diagram illustrating the details of the recommended content candidate generation unit according to the embodiment. The recommended content candidate generation unit 11 can perform (1) user preference estimation, (2) user state / situation estimation, and (3) recommended content candidate generation.

[0014] (1) User preference estimation The user preference estimation unit 111 is connected via a wired or wireless network to the in-vehicle navigation system's destination search history database (DB) 31, destination visit history DB 32, and destination setting history DB 33 (as well as the car navigation system, voice interaction agent, CAN, etc.). The user preference estimation unit 111 estimates the user's preference score for destination genres from past history such as destination search history, destination visit history, destination setting history, and response history to recommended content. The user preference score may be estimated sequentially online, or it may be estimated and updated periodically offline (in batches).

[0015] Destination search and setting history can be obtained from car navigation systems, voice interaction agents, smartphone apps, etc. Destination visit history can be obtained from car navigation systems, CAN (Controller Area Network) (vehicle behavior / driving operation) IG-off points, etc. Response history to recommended content can obtain user reactions to recommended content (e.g., approval, rejection, etc.). The preference estimation unit 111 stores the estimated preference data in the user preference DB 113.

[0016] (2) Estimation of user state / condition The user state / situation estimation unit 112 is connected to CAN (vehicle behavior / driving operation) 41, external camera video DB 42, internal camera video DB 43, biometric data DB 44, destination / route setting history DB 45, and weather DB 46. The user state / situation estimation unit 112 estimates the user's state / situation, vehicle usage / condition, and daily lifestyle habits from CAN (vehicle behavior / driving operation), external camera video, internal camera video, biometric data, etc. Biometric data of the user can be acquired from a smartwatch or wearable device. A facial image of the user may be acquired from an internal camera.

[0017] The system may use a combination of online estimation and updating of user status and conditions, and batch estimation and updating of user status and conditions, or either one alone. Examples of user status and conditions estimated and updated online include, but are not limited to, fatigue level, drowsiness, tension, vehicle warnings, slippage, refueling / charging, weather, etc. Examples of user status and conditions estimated and updated offline (batch) include, but are not limited to, frequently visited places and their days and times, frequently driven routes and their days and times. The status estimation unit 112 stores the estimated user status and conditions data in the user status database 114.

[0018] (3) Generation of recommended content candidates The recommended content candidate generation unit 115 is connected to the user preference DB 113 and the status / situation DB 114. Also, the recommended content candidate generation unit 115 is connected to the destination congestion situation DB 51, the parking lot congestion situation DB 52 near the destination, and the congestion situation DB 53 of the route to the destination.

[0019] The recommended content candidate generation unit 115 generates a list of recommended content candidates from the preference scores for the destination genre estimated in (1) and (2) above, the user status / situation, the current location, the day of the week, the time, etc. At that time, the utility (score) of the recommended content candidate (list) is also estimated. Also, the recommended content candidate includes at least one or all of the destination congestion situation, the parking lot congestion situation near the destination, and the congestion situation of the route to the destination. Note that the congestion situation of the route to the destination can be obtained from VICS (Vehicle Information and Communication System) (registered trademark). Also, the congestion situation of the route to the destination may be estimated from the vehicle speeds of each vehicle acquired via the connected car. The destination congestion situation may be obtained from a monitoring system or a seat reservation system at the destination (for example, a store, a restaurant). The parking lot congestion situation near the destination may be obtained from a monitoring camera or a occupancy / vacancy information system of the parking lot near the destination, or may be estimated from the ignition-off situation of the connected car. As shown in FIG. 2, the generated recommended content list 70 shows the content and the utility (score) corresponding to the user ID. The list 70 recommends different contents (C1, C2, C3) to the same user having the same user ID. The utility of content C1 is the highest, and the utility of content C1 is the lowest.

[0020] Next, the recommendation history management unit 21 of the adjustment agent 20 stores and manages the following information etc. presented from each recommendation agent 10 in a database. The information to be stored includes the recommended content (list), the utility (score), the date and time, the current location, the ID uniquely identifying the recommendation agent, the recommended content candidate (list), the utility (score), the date and time, the current location, etc.

[0021] In addition, the recommended content (list), the user's response, the utility (score) obtained by the user, the date and time, the ID uniquely identifying the recommendation agent, the recommended content (list), the user's response, the utility (score) obtained by the user, the date and time, etc. are also stored.

[0022] Returning again to FIG. 1 to continue the explanation. The recommendation execution unit 12 of the recommendation agent 10 presents the recommended content to be executed, which is transmitted from the adjustment agent 20, to the user. The media for presentation is, for example, a car navigation, a voice dialogue agent, a smartphone app, etc., but is not limited thereto.

[0023] One or more adjustment agents 20 include a recommendation history management unit 21 and a recommendation strategy determination unit 22. In FIG. 1, one adjustment agent 20 is arranged in the center server, but a plurality of adjustment agents may be arranged in each vehicle. The adjustment agent 20 may be a system incorporating AI (Artificial Intelligence), or may be an operator such as a call center, for example.

[0024] The recommendation history management unit 21 manages the date and time, location, content, utility (score), etc. of past recommendations made by each recommendation agent 10a-10n.

[0025] The recommendation strategy determination unit 22 decides whether to present recommended content or to recommend content from the recommended content list based on the current or past utility (score) of each recommendation agent 10a-10n. The adjustment agent 20 may use AI to decide whether to present recommended content or to recommend content from the recommended content list. The recommendation strategy determination unit 22 may decide whether to present recommended content or to recommend content from the recommended content list based on the congestion status of the destination when the destination was recommended to each agent, the congestion status of the destination's parking lot, or the congestion status of the route to the destination. Destination congestion status includes the capacity of the destination (e.g., store, facility) (e.g., number of seats, number of tables). The recommendation strategy determination unit 22 may give a higher priority to recommending a destination to each recommendation agent the higher their utility (score) is, so as not to exceed the capacity of the destination. For example, in the case of the recommended content list 70 in Figure 2, C1 is provided to the user (User ID: AAA) with the highest utility (score).

[0026] The recommendation strategy determination unit 22 can determine a recommendation strategy by considering the current destination's capacity and the congestion situation around the destination. The recommendation strategy determination unit 22 can determine a recommendation strategy using AI. When the recommendation strategy determination unit 22 recommends the same destination (e.g., a store or restaurant) to multiple recommendation agents, if the number of users (customers) at the predicted arrival time exceeds the destination's capacity (e.g., maximum number of seats), it may refrain from recommending the destination to some recommendation agents (users) or lower its priority. When the recommendation strategy determination unit 22 recommends the same destination to multiple recommendation agents, if the number of users exceeds the capacity of the parking lot near the destination (e.g., maximum number of parking spaces), it may refrain from recommending the destination to some recommendation agents or lower its priority. Furthermore, when the recommendation strategy determination unit 22 recommends the same destination to multiple recommendation agents, if the congestion situation on the route to the destination is significantly high for some recommendation agents, it may refrain from recommending the destination to those recommendation agents or lower its priority. These are merely examples, and various recommendation strategies can be determined as understood by those skilled in the art.

[0027] The recommendation strategy determination unit 22 may determine a harmonious recommendation strategy for multiple users when recommending destinations to multiple users multiple times over a predetermined period (for example, several months or one year). The recommendation strategy determination unit 22 may calculate the average utility (score) of destinations for each user over the predetermined period and give higher priority to recommending destinations to users with lower average values. When the recommendation strategy determination unit 22 recommends the same destination (for example, a store or restaurant) to multiple recommendation agents, if the capacity (for example, maximum number of seats) of the destination is exceeded, it may refrain from recommending the destination to some recommendation agents (for example, users who have visited more than a predetermined number of times in the past predetermined period) or lower its priority. The recommendation strategy determination unit 22 may give higher priority to recommending destinations to users who did not receive a recommendation to a destination last time. In addition, it may refrain from recommending a destination or lower its priority based on the user's response history of rejecting recommended content. The recommendation strategy decision unit 22 can use AI to determine a harmonious recommendation for multiple users as a whole. This allows for harmonious recommendations to be made for multiple users when providing destination recommendations to each user multiple times over a predetermined period.

[0028] The coordination agent 20 may sequentially determine a recommendation strategy for the recommended content candidates from the recommendation agents. Alternatively, the coordination agent 20 may collect recommended content candidates from each recommendation agent for a predetermined period (e.g., 30 seconds, 1 minute, 5 minutes), and then determine a recommendation strategy considering the collected recommended content candidates.

[0029] Refer to Figure 1 to explain the exchange of information between the recommendation agent and the coordination agent. The recommendation agent 10 transmits the generated recommendation content and its estimated utility (score), etc., to the adjustment agent 20 (S1). The date and time when the recommendation agent 10 generated the content and its current location may also be transmitted.

[0030] The coordination agent 20 then transmits to the recommendation agent 10 the content to perform the recommendation (S2). In some embodiments, the content to perform the recommendation may be presented to the recommendation agent (e.g., 10a) by the coordination agent 20 when the coordination agent (e.g., 20) rejects a recommendation from the recommendation agent (e.g., 10a) (i.e., decides not to perform the recommendation), by offering other recommendation agents (e.g., any of 10b to n) with whom negotiation is possible. This allows the recommendation agent (e.g., 10a) to negotiate with the other recommendation agents (e.g., any of 10b to n). In other embodiments, the recommendation agent (e.g., 10a) may renegotiate with the coordination agent 20.

[0031] In other embodiments, the content that performs this recommendation may include negotiation terms between the recommendation agent and the proposed recipient (e.g., automated negotiation agent technology).

[0032] The recommendation agent 10 then sends the recommended content (list) to the coordination agent 20 (S3). The user's response, the utility (score) obtained by the user, and the date and time may also be sent.

[0033] Furthermore, both the computer installed in the vehicle and the central server of the vehicle control system may have the functions of a recommendation agent or a coordination agent.

[0034] It should be noted that the present invention is not limited to the embodiments described above, and can be modified as appropriate without departing from the spirit of the invention. [Explanation of symbols]

[0035] 1. Destination Recommendation System 10 Recommended Agents 11. Recommended Content Candidate Generation Unit 12. Recommendation Executive Committee 20 Coordination Agent 21 Recommendation History Management Department 22 Recommendation Strategy Decision-Making Department 31 Destination Search History Database 32 Destination Visit History Database 33 Destination Setting History DB 41 CAN (Vehicle Behavior / Driving Operation) 42 External Vehicle Camera Video Database 43 In-car camera video database, 44 Biological Data Database 45 Destination / Route Setting History Database 46 Weather Database 51 Destination Congestion Status Database 52 Parking lot congestion status near destination database 53. Congestion status of routes to destinations (database) 70 Recommended Content List 111 User Preference Estimation Unit 112 User State / Status Estimation Unit 115 Recommended Content Candidate Generation Unit

Claims

1. A destination recommendation system comprising multiple recommendation agents that recommend destinations to each user, and a coordination agent that coordinates the multiple recommendation agents, The aforementioned multiple recommendation agents estimate user preferences for destinations based on at least one of destination search history, destination visit history, destination setting history, and response history to recommended content, and estimate the user's state and situation based on at least one of vehicle behavior, driving operations, external camera video, internal camera video, and user biometric data. Based on the estimated user preferences, the user's state / condition, and at least one of the destination congestion status, the parking lot congestion status near the destination, and the congestion status of the route to the destination, recommended content candidates, the utility of the recommended content candidates, the date and time, and the user's current location are generated, and the generated recommended content candidates, the utility, the date and time, and the user's current location are provided to the adjustment agent, respectively. The aforementioned coordination agent, based on at least one of the following—the congestion status of the destination when it recommended the destination to each recommendation agent, the congestion status of parking lots near the destination, and the congestion status of the route to the destination—and the current or past effectiveness of each recommendation agent, determines for each recommendation agent whether or not to present recommended content, or to select recommended content from the recommended content candidates, and provides each recommendation agent with information on whether or not to present the selected recommended content, and the recommended content itself. A destination recommendation system in which each of the multiple recommendation agents performs the act of presenting the recommended content to each user based on whether or not the determined recommended content is presented and the recommended content itself.

2. The destination recommendation system according to claim 1, wherein the adjustment agent, when recommending destinations to multiple users multiple times over a predetermined period, calculates the average utility of each user's destinations over the predetermined period, and, based on the calculated average utility of each user, determines for each recommendation agent whether or not to present recommendation content, or to select recommendation content from the list of recommendation content candidates.

3. The destination recommendation system according to claim 1, wherein if the coordinating agent rejects a recommended content candidate from the recommendation agent, the recommendation agent provides the recommendation agent with another recommendation agent with whom negotiation is possible.

4. The destination recommendation system according to claim 1, wherein, after presenting the recommended content to each user, the plurality of recommendation agents each provide the coordination agent with the recommended content, the user's response, the benefits obtained by the user, and the date and time.

5. The destination recommendation system according to claim 1, wherein the coordination agent includes AI (Artificial Intelligence).

Images