Real-time travel segmentation method based on TBOX

Abstract

The invention belongs to the technical field of vehicle control, and particularly relates to a real-time travel segmentation method based on a TBOX, and the method comprises the steps: obtaining a real-time signal of a vehicle setting position for a CAN bus through a vehicle-mounted TBOX; the TBOX continuously detects a vehicle power-on signal as a travel triggering condition, and generates a unique travel ID at the same time; synchronously acquiring GPS position data, CAN bus vehicle signals and IMU inertial measurement data in real time; a vehicle power-off signal is continuously monitored, latest effective position information is reserved, and when the power-off signal is monitored, the geographic position of the position is obtained, and the scene type is output through the scene classifier; when the vehicle is not powered on, the TBOX enters a low-power-consumption sleep mode. According to the technical scheme, scene recognition is achieved by fusing multi-dimensional parameters such as the geofence and the parking duration, the differential judgment threshold value is set, and the problem that the travel division precision is insufficient due to a traditional fixed threshold value is solved.

Description

Technical Field

[0001] This invention belongs to the field of vehicle control technology, specifically relating to a real-time trip segmentation method based on TBOX. Background Technology

[0002] Current vehicle trip segmentation and merging logic primarily relies on a fixed-time threshold method, classifying trips as independent trips when the interval between adjacent power-off and power-on signals exceeds a preset duration. However, this method has significant limitations: in non-highway scenarios, frequent short starts and stops can easily lead to the incorrect merging of multiple independent trips; while in high-speed scenarios, reasonable long stops such as rest stops at service areas or charging stops are forcibly segmented due to exceeding the threshold, resulting in trip fragmentation. Traditional solutions lack awareness of scenario characteristics such as road type and driving behavior, and cannot adapt to the threshold requirements of differentiated scenarios such as non-highway congestion, high-speed driving, and charging stops, leading to a high error rate in trip data statistics. With the digitalization of intelligent road networks and the improvement of vehicle sensor accuracy, dynamic threshold adjustment through multi-dimensional scenario recognition has become possible.

[0003] CN202410829479.3 discloses a vehicle trip segmentation method, apparatus, device, and medium. This technical solution proposes a trip segmentation method based on adjacent vehicle start-stop signals: by analyzing adjacent start-stop events in a time sequence, short-term stops exceeding a preset threshold are merged into waypoints. Its core innovation lies in using continuous start-stop signals to determine the start and end of the trip, and updating waypoints according to a preset time threshold, simplifying the algorithm complexity, but it has significant limitations in application scenarios.

[0004] The main drawback of this method, which merges route points based on preset intervals, is that its fixed threshold lacks scenario adaptability. The fixed preset interval cannot dynamically distinguish between different parking types (such as temporary stops, automatic start-stop of new energy vehicles, long stops in highway service areas, etc.), which may cause short-term stops (< preset interval) in non-highway scenarios to be forcibly merged, potentially masking the actual trip nodes. Meanwhile, reasonable long stops in highway scenarios (such as 30 minutes of charging) are incorrectly segmented because they exceed the threshold. At the same time, this method relies on a single time dimension and does not combine multi-source data such as vehicle status (such as gear position, door status), changes in passenger weight, charging gun status, fuel filler cap status, and environmental features (such as GPS location) for comprehensive judgment, thus reducing the accuracy of trip segmentation. Summary of the Invention

[0005] The purpose of this invention is to provide a real-time trip segmentation method based on TBOX to address the shortcomings of existing technologies in terms of scenario adaptability: they rely on fixed thresholds and cannot distinguish between short non-high-speed stops and long high-speed stops, resulting in insufficient trip segmentation longitude. This is achieved by adding the identification of parking scenarios and setting differentiated threshold processing for different scenarios.

[0006] The system suffers from a single-dimensional judgment metric: trip segmentation relies solely on vehicle power-on / off signals, failing to distinguish between functional power outages (such as turning off the engine upon arrival at the destination) and temporary power outages (such as automatic start-stop triggering). This leads to short wait times being included in the driving period, while brief periods of engine shutdown (such as moving the vehicle for 30 seconds) are mistakenly identified as independent trips. Furthermore, the lack of environmental awareness prevents the system from recognizing special states such as charging and refueling, potentially incorrectly marking forced power-offs during charging or refueling as trip completion. Cross-validation using multi-dimensional signals across various scenarios can effectively improve the accuracy of trip segmentation and its adaptability to different scenarios.

[0007] To achieve the above objectives, this application employs the following technical solution:

[0008] A real-time trip segmentation method based on TBOX includes the following steps:

[0009] S1. The vehicle-mounted TBOX obtains real-time signals of the vehicle's set position via the CAN bus;

[0010] S2 and TBOX continuously detect the vehicle power-on signal as the trip trigger condition. When the power-on signal is detected, the latitude and longitude information of the vehicle power-on location is obtained, the geographical location of the location is obtained, and a unique trip ID containing the date, the last 4 digits of the TBOX MAC and a random code is generated.

[0011] S3. Real-time synchronous acquisition of GPS location data, CAN bus vehicle signals and IMU inertial measurement data;

[0012] S4. Continuously monitor the vehicle power-off signal and retain the latest valid location information. When the power-off signal is detected, obtain the geographical location of the location and output the scene type through the scene classifier. When the vehicle is not powered on, TBOX enters a low-power sleep mode.

[0013] Furthermore, in step S1, the vehicle-mounted T-BOX obtains real-time signals of the vehicle's set position via the CAN bus. Specifically, the vehicle-mounted T-BOX obtains real-time signals of the vehicle as a whole, the cabin, the body, and the vehicle's position via the CAN bus, including vehicle power-on / power-off status, gear position, door status, cabin occupant weight value, charging gun status, fuel filler cap status, and GPS location.

[0014] Furthermore, the geographical location of the current location in steps S2 and S4 is obtained through Gaode reverse geocoding.

[0015] Furthermore, if location information is lost, the system will use the vehicle's latest valid location information, the IMU's accelerometer and gyroscope to perform short-term location estimation, combine road network data to complete the location, and then use Gaode reverse geocoding to obtain the geographical location of the current location.

[0016] Furthermore, in step S4, the scene type is output through the scene classifier. The high-speed scene refinement recognition process based on GPS + map includes the following steps:

[0017] S41 and TBOX obtain vehicle latitude and longitude coordinates in real time;

[0018] S42. Use online API or offline geofencing to perform map matching to determine whether the vehicle is located on a highway, gas station, or charging station. If the GPS continues to move after the vehicle is turned off, mark it as a tow truck. Highway sections that do not match service areas or tow truck scenarios are classified as other highways. Sections not on highways are marked as non-highway sections.

[0019] Furthermore, specific highway scenario types are marked as: highway service area, highway tow truck, other highways, non-highway, non-highway gas station, and non-highway charging station.

[0020] Furthermore, if the scene classifier identifies a high-speed service area, the waiting time is extended. If the waiting time exceeds the first threshold, the trip ends; otherwise, the trip continues.

[0021] When the scene classifier is a high-speed trailer, a shorter waiting time is set. If the second threshold is exceeded, the trip ends; otherwise, the trip continues.

[0022] If the scene classifier is classified as high-speed or other, the waiting time is extended. If the waiting time exceeds the third threshold, the process ends; otherwise, the process continues.

[0023] When the scene classifier is a non-high-speed scene, check if the gear is P. If it is not P, continue the trip. Otherwise, if the parking location is a gas station, set a short-term fourth threshold. If the fourth threshold is exceeded, end the trip. Otherwise, continue the trip.

[0024] When the parking location is a charging station, a longer sixth threshold is set. If the sixth threshold is exceeded, the trip ends; otherwise, the trip continues.

[0025] When the parking location is other, the system combines whether the weight reduction of the driver's cabin occupants exceeds a set value or the driver's door is opened. If the conditions are met, a shorter fifth threshold is set. If the fifth threshold is exceeded, the trip ends; otherwise, the trip continues.

[0026] All vehicles that lose power during the journey are marked as stopping points, and the process continues to step S4.

[0027] Furthermore, this includes encrypting the trip data and compressing it using algorithms to reduce the transmission size, and uploading the encrypted data packets to the cloud via the MQTT protocol.

[0028] The beneficial effects of this invention are:

[0029] This technical solution innovatively proposes a method for setting thresholds based on different scenarios. By integrating multi-dimensional parameters such as geofencing and parking duration, it can identify scenarios such as short-distance stops on non-highways, long-distance stops on highways, and destination parking, and sets differentiated judgment thresholds to solve the problem of insufficient trip segmentation accuracy caused by traditional fixed thresholds. Simultaneously, the patent utilizes a multi-modal signal cross-validation algorithm to integrate multi-dimensional data such as vehicle power status, door status, GPS, inertial sensors, and gravity sensors to distinguish between functional power outages and temporary power outages, improving the reliability and scenario adaptability of trip segmentation. Attached Figure Description

[0030] Figure 1 This is a logic diagram of the real-time trip segmentation based on TBOX according to the present invention. Detailed Implementation

[0031] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings. The following embodiments are merely exemplary and can only be used to explain and illustrate the technical solution of the present invention, and should not be construed as limiting the technical solution of the present invention.

[0032] like Figure 1 As shown, this application provides a real-time trip segmentation method based on TBOX, including the following steps:

[0033] (1) Data collection:

[0034] S1. The vehicle-mounted T-BOX obtains real-time signals of the vehicle, cabin, body, and vehicle position via the CAN bus, including vehicle power-on / off status, gear position, door status, passenger weight, charging gun status, fuel filler cap status, and GPS location.

[0035] (2) TBOX Calculation Module:

[0036] S2 and TBOX continuously detect the vehicle power-on signal as the trip trigger condition. When the power-on signal is detected, the latitude and longitude information of the vehicle power-on location is obtained. The geographical location of the location is obtained through Gaode reverse geocoding. At the same time, a unique trip ID containing the date, the last 4 digits of the TBOX MAC and a random code is generated.

[0037] S3. Real-time synchronous acquisition of GPS location data, CAN bus vehicle signals, and IMU inertial measurement data.

[0038] S4. Continuously monitor the vehicle power-off signal and retain the latest valid location information. When the power-off signal is detected, obtain the geographical location of the location through Gaode reverse geocoding and output the scene type through the scene classifier. When the vehicle is not powered on, TBOX enters low-power sleep mode.

[0039] This application also includes, if location information is lost, performing short-term location estimation based on the vehicle's latest valid location information, using the IMU's accelerometer and gyroscope (resetting the IMU speed integral error when the vehicle is stationary), combining road network data to complete the location, and then using Gaode reverse geocoding to obtain the geographical location of the current location.

[0040] In this application, the scene type is output through a scene classifier: The high-speed scene fine-grained recognition process based on GPS + map is as follows:

[0041] S41 and TBOX obtain vehicle latitude and longitude coordinates in real time;

[0042] S42. Use online API or offline geofencing to perform map matching to determine whether the vehicle is located on a highway, gas station, or charging station. If the GPS continues to move after the vehicle is turned off, mark it as a tow truck. Highway sections that do not match service areas or tow truck scenarios are classified as other highways. Sections not on highways are marked as non-highway sections.

[0043] The specific high-speed scenario types are marked as: high-speed service area, high-speed tow truck, other high-speed, non-high-speed, non-high-speed gas station, and non-high-speed charging station.

[0044] In this application, when the scenario classifier is a highway service area, the waiting time is extended (which may be a scenario of waiting for charging or resting in a service area). If the waiting time exceeds the first threshold, the trip ends; otherwise, the trip continues.

[0045] When the scene classifier is a high-speed trailer (i.e., the vehicle moves via GPS after power is off), a short waiting time is set. If the second threshold is exceeded, the trip ends; otherwise, the trip continues.

[0046] If the scenario classifier classifies it as "high-speed other", the waiting time is extended (which may be due to congestion, emergency lane parking, accidents, etc.). If the waiting time exceeds the third threshold, the trip ends; otherwise, the trip continues.

[0047] When the scene classifier is a non-high-speed scene, check if the gear is P. If it is not P, continue the trip. Otherwise, if the parking location is a gas station, set a short-term fourth threshold. If the fourth threshold is exceeded, end the trip. Otherwise, continue the trip.

[0048] When the parking location is a charging station, a relatively long sixth threshold is set (temporary replenishment of power generally does not exceed one hour). If the sixth threshold is exceeded, the trip ends; otherwise, the trip continues.

[0049] When the parking location is other, the system considers whether the weight reduction of the driver's cabin occupants exceeds 30kg or whether the driver's door is opened. If the conditions are met, a shorter fifth threshold is set. If the fifth threshold is exceeded, the trip ends; otherwise, the trip continues.

[0050] All vehicles that lose power during the journey are marked as stopping points, and the process continues to step S4.

[0051] The last power outage after the trip ends will be the actual trip end time.

[0052] The trip data is encrypted and compressed using algorithms to reduce transmission size, and the encrypted data packets are uploaded to the cloud via the MQTT protocol.

[0053] In this application, the scene classifier identifies high-speed and non-high-speed scenes based on online APIs or offline electronic fences, where the vehicle parking time differs significantly between the two scenarios.

[0054] Multi-source signal cross-verification: By using a multi-modal signal cross-verification algorithm, the system integrates multi-dimensional data such as vehicle power status, door status, GPS, inertial sensors, and gravity sensors to distinguish between functional power outages and temporary power outages, thereby improving the reliability of trip division and the adaptability of scenarios.

[0055] Through deep collaboration between cloud-based enhancement modules and TBOX, traditional trip segmentation technology is expanded into an intelligent system with self-learning capabilities: At the technical level, an innovative federated learning architecture for vehicle-cloud collaboration is adopted to achieve dynamic threshold optimization, and scene recognition accuracy is improved through multimodal semantic fusion (GPS trajectory + visual OCR + acoustic features); at the hardware level, a dedicated NPU instruction set is designed to accelerate edge computing, and a lightweight binary protocol is developed to ensure real-time performance; at the application level, cross-domain applications such as UBI insurance dynamic pricing and accurate carbon footprint accounting are derived, forming a complete closed loop of "data collection - scene analysis - commercial monetization".

[0056] The above are preferred embodiments of the present invention. The basic principles and advantages of the present invention have been shown and described above. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are only illustrative of the principles of the present invention. Various changes and modifications can be made to the present invention without departing from the spirit and scope of the present invention. All such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A real-time route segmentation method based on TBOX, characterized in that, Includes the following steps: S1. The vehicle-mounted TBOX obtains real-time signals of the vehicle's set position via the CAN bus; S2 and TBOX continuously detect the vehicle power-on signal as the trip trigger condition. When the power-on signal is detected, the latitude and longitude information of the vehicle power-on location is obtained, the geographical location of the location is obtained, and a unique trip ID containing the date, the last 4 digits of the TBOX MAC and a random code is generated. S3. Real-time synchronous acquisition of GPS location data, CAN bus vehicle signals and IMU inertial measurement data; S4. Continuously monitor the vehicle power-off signal and retain the latest valid location information. When the power-off signal is detected, obtain the geographical location of the location and output the scene type through the scene classifier. When the vehicle is not powered on, TBOX enters a low-power sleep mode.

2. The real-time trip segmentation method based on TBOX according to claim 1, characterized in that, In step S1, the vehicle-mounted T-BOX obtains real-time signals of the vehicle's set position via the CAN bus. Specifically, the vehicle-mounted T-BOX obtains real-time signals of the vehicle as a whole, the cabin, the body, and the vehicle's position via the CAN bus, including vehicle power-on, power-off, gear position, door status, cabin occupant weight, charging gun status, fuel filler cap status, and GPS location.

3. The real-time trip segmentation method based on TBOX according to claim 1, characterized in that, In steps S2 and S4, the geographic location is obtained by using Gaode reverse geocoding.

4. The real-time trip segmentation method based on TBOX according to claim 1, characterized in that, It also includes, if location information is lost, using the vehicle's latest valid location information, short-term location estimation using the IMU's accelerometer and gyroscope, location completion using road network data, and then obtaining the geographic location using Gaode reverse geocoding.

5. The real-time trip segmentation method based on TBOX according to claim 1, characterized in that, In step S4, the scene type is output through the scene classifier. The high-speed scene refinement recognition process based on GPS + map includes the following steps: S41 and TBOX obtain vehicle latitude and longitude coordinates in real time; S42. Use online API or offline geofencing to perform map matching to determine whether the vehicle is located on a highway, gas station, or charging station. If the GPS continues to move after the vehicle is turned off, mark it as a tow truck. Highway sections that do not match service areas or tow truck scenarios are classified as other highways. Sections not on highways are marked as non-highway sections.

6. The real-time trip segmentation method based on TBOX according to claim 5, characterized in that, The specific high-speed scenario types are marked as: high-speed service area, high-speed tow truck, other high-speed, non-high-speed, non-high-speed gas station, and non-high-speed charging station.

7. The real-time trip segmentation method based on TBOX according to claim 6, characterized in that, If the scene classifier identifies it as a high-speed service area, the waiting time is extended. If the waiting time exceeds the first threshold, the trip ends; otherwise, the trip continues. When the scene classifier is a high-speed trailer, a shorter waiting time is set. If the second threshold is exceeded, the trip ends; otherwise, the trip continues. If the scene classifier is classified as high-speed or other, the waiting time is extended. If the waiting time exceeds the third threshold, the process ends; otherwise, the process continues. When the scene classifier is a non-high-speed scene, check if the gear is P. If it is not P, continue the trip. Otherwise, if the parking location is a gas station, set a short-term fourth threshold. If the fourth threshold is exceeded, end the trip. Otherwise, continue the trip. When the parking location is a charging station, a longer sixth threshold is set. If the sixth threshold is exceeded, the trip ends; otherwise, the trip continues. When the parking location is other, the system combines whether the weight reduction of the driver's cabin occupants exceeds a set value or the driver's door is opened. If the conditions are met, a shorter fifth threshold is set. If the fifth threshold is exceeded, the trip ends; otherwise, the trip continues. All vehicles that lose power during the journey are marked as stopping points, and the process continues to step S4.

8. The real-time trip segmentation method based on TBOX according to claim 1, characterized in that, This includes encrypting trip data and compressing it using algorithms to reduce transmission size, and uploading the encrypted data packets to the cloud via the MQTT protocol.

Images