Intelligent route planning system and method based on intelligent riding helmet

Abstract

The invention discloses an intelligent route planning system and method based on an intelligent riding helmet, and the method comprises the steps: converting IMU original data through the low-dimensional structured coding of a full-connection layer through employing a data time queue continuously collected by an IMU sensor, inputting the data into a specially designed motion state nonlinear time sequence type integral coding network for processing, and carrying out the processing of a non-linear time sequence type integral coding network. Key features capable of reflecting a real motion track are extracted from the current position, a position offset compensation value is calculated through a decoder, and finally the current position of the user is accurately corrected. By means of the mode, reliable position information can be provided under the condition that GPS signals are weak or lack, the accuracy and reliability of path planning are ensured, and then the safety and experience feeling of riding are improved.

Description

Technical Field

[0001] The present application relates to the field of intelligent planning, and more specifically, to a route intelligent planning system and method based on a smart cycling helmet. Background Art

[0002] With the rapid development of intelligent transportation systems, cycling safety and efficiency have become a growing concern. While a variety of cycling assistance devices, such as smartwatches and smartphone apps, already exist on the market, they primarily help cyclists choose optimal routes by providing navigation and route planning. However, these devices often rely on external devices or mobile phones as their primary information processing unit. This not only increases the burden of carrying additional equipment but can also affect user experience and safety due to improper device placement.

[0003] Furthermore, existing technical solutions still face challenges in terms of positioning accuracy. Although conventional cycling navigation systems use GPS positioning technology, they do not fully consider the positioning drift caused by multipath interference under the urban canyon effect. This is especially true when cyclists are in complex scenarios such as elevated bridges, dense building clusters, or tunnels. Relying solely on the GPS module will result in cumulative position deviations, seriously affecting the reliability of subsequent path planning. Although some systems have attempted to introduce inertial measurement units for auxiliary positioning, the conventional linear compensation model cannot effectively analyze the nonlinear posture changes in cycling motion, such as the centripetal acceleration interference during sharp turns or the random vibration noise on bumpy roads, resulting in significant deviations between the motion state estimate and the actual trajectory.

[0004] In addition, traditional route planning systems are mostly based on static data for calculations, that is, they only consider the shortest distance or estimated time between the starting and ending points; at the same time, the existing technology uses traffic flow data updated at a fixed period and cannot capture the instantaneous changes in road conditions caused by sudden accidents, resulting in the assessment of the impact of real-time changing traffic conditions and meteorological factors on road accessibility remaining at the static threshold judgment level.

[0005] In response to the above problems, we look forward to an optimized route intelligent planning method based on smart cycling helmets. Summary of the Invention

[0006] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides a route intelligent planning system and method based on a smart cycling helmet, which utilizes the time queue of data continuously collected by the IMU sensor, converts the IMU raw data through the low-dimensional structured coding of the fully connected layer, and inputs it into a specially designed motion state nonlinear time series integral coding network for processing, extracts the key features that can reflect the real motion trajectory, and then calculates the position offset compensation value through the decoder, and finally realizes the accurate correction of the user's current position. In this way, reliable location information can be provided even when the GPS signal is weak or missing, ensuring the accuracy and reliability of path planning, thereby improving the safety and experience of riding.

[0007] According to one aspect of the present application, a method for intelligent route planning based on a smart cycling helmet is provided, which includes:

[0008] The GPS module and IMU sensor deployed in the smart cycling helmet collect the time queue of the user's real-time location and the time queue of IMU data;

[0009] Utilizing an external data receiving module deployed in the smart cycling helmet to receive a time queue of traffic data and a time queue of weather data;

[0010] Extracting the user's current location from the time queue of the user's real-time location;

[0011] Based on the time queue of the IMU data, performing position deviation correction on the user's current position to obtain an optimized user's current position;

[0012] Inputting the optimized user's current location, the time queue of the traffic data, and the time queue of the weather data into a road environment comprehensive state evaluation module to obtain a road environment comprehensive state evaluation result;

[0013] Based on the comprehensive road environment status assessment results and set riding preferences, a dynamic route optimization algorithm is used to adjust the riding route planning in real time.

[0014] According to another aspect of the present application, a route intelligent planning system based on a smart cycling helmet is provided, which includes:

[0015] The data acquisition module is used to collect the time queue of the user's real-time location and the time queue of IMU data through the GPS module and IMU sensor deployed in the smart cycling helmet;

[0016] an external data receiving module, configured to receive a time queue of traffic data and a time queue of weather data using the external data receiving module deployed in the smart cycling helmet;

[0017] A user current location extraction module, configured to extract the user current location from the time queue of the user's real-time location;

[0018] A position deviation correction module, configured to perform position deviation correction on the user's current position based on the time queue of the IMU data to obtain an optimized user's current position;

[0019] a road environment comprehensive state evaluation module, configured to input the optimized user's current location, the time queue of the traffic data, and the time queue of the weather data into the road environment comprehensive state evaluation module to obtain a road environment comprehensive state evaluation result;

[0020] The real-time route adjustment module is used to adjust the cycling route planning in real time based on the comprehensive status evaluation results of the road environment and the set cycling preferences, and use a dynamic route optimization algorithm.

[0021] Compared with the existing technology, the present application provides a route intelligent planning system and method based on a smart cycling helmet. It uses the time queue of data continuously collected by the IMU sensor, converts the IMU raw data through the low-dimensional structured coding of the fully connected layer, and inputs it into a specially designed motion state nonlinear time series integral coding network for processing. It extracts key features that can reflect the actual motion trajectory, and then calculates the position offset compensation value through the decoder, ultimately achieving accurate correction of the user's current position. In this way, reliable location information can be provided even when the GPS signal is weak or missing, ensuring the accuracy and reliability of path planning, thereby improving the safety and experience of cycling. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The above and other purposes, features, and advantages of the present application will become more apparent through a more detailed description of the embodiments of the present application in conjunction with the accompanying drawings. The accompanying drawings are intended to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0023] Figure 1 Flowchart of a method for intelligent route planning based on a smart cycling helmet according to an embodiment of the present application;

[0024] Figure 2 Schematic diagram of data flow of a route intelligent planning method based on a smart cycling helmet according to an embodiment of the present application;

[0025] Figure 3 This is a flowchart of sub-step S4 of the route intelligent planning method based on the smart cycling helmet according to an embodiment of the present application;

[0026] Figure 4Flowchart of sub-step S42 of the route intelligent planning method based on the smart cycling helmet according to an embodiment of the present application;

[0027] Figure 5 4 is a block diagram of an intelligent route planning system based on a smart cycling helmet according to an embodiment of the present application. DETAILED DESCRIPTION

[0028] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described herein.

[0029] As used in this application and the claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" are not intended to refer to the singular but may include the plural. Generally speaking, the terms "comprises" and "include" only indicate the inclusion of the steps and elements specifically identified, and these steps and elements do not constitute an exclusive list. A method or apparatus may also include other steps or elements.

[0030] Although the present application makes various references to certain modules in the system according to embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are illustrative only, and different aspects of the system and method can use different modules.

[0031] Flowcharts are used in this application to illustrate the operations performed by the systems according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, the various steps may be processed in reverse order or simultaneously, as needed. Furthermore, other operations may be added to these processes, or one or more operations may be removed from these processes.

[0032] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described herein.

[0033] To overcome the data silo effect of traditional split-device architectures, this technology leverages a high-precision sensor array integrated into the helmet itself to establish a positioning compensation mechanism and environmental state assessment model with enhanced spatiotemporal continuity in cycling scenarios. The core concept of this technical solution lies in constructing a multimodal data fusion decision-making system integrated into the cycling helmet itself. Through the synergy of an embedded sensor network and intelligent algorithms, this system achieves a technological leap forward in three dimensions: positioning accuracy, environmental perception, and dynamic response. Specifically, considering the shortcomings of existing technologies in positioning accuracy, real-time data processing, and user experience, this technical solution combines a GPS module with an IMU sensor to collect real-time user location and motion status information, effectively overcoming the urban canyon effect caused by relying solely on GPS positioning. Furthermore, by introducing a low-dimensional structured coding method based on fully connected layers and a nonlinear time-series integral coding network, it accurately captures and compensates for the rider's complex motion postures. Furthermore, an external data receiving module acquires time queues of traffic and weather data, which are combined with a comprehensive road environment state assessment module, enabling the system to dynamically adjust to changing road conditions and cycling preferences.

[0034] Specifically, in the technical solution of the present application, a route intelligent planning method based on a smart cycling helmet is proposed. Figure 1 The figure is a flowchart of a method for intelligent route planning based on a smart cycling helmet according to an embodiment of the present application. Figure 2 Schematic diagram of data flow of the intelligent route planning method based on the smart cycling helmet according to the embodiment of the present application. Figure 1 and Figure 2 As shown, according to the embodiment of the present application, the route intelligent planning method based on the smart cycling helmet includes the following steps: S1, collecting the time queue of the user's real-time position and the time queue of IMU data through the GPS module and IMU sensor deployed in the smart cycling helmet; S2, using the external data receiving module deployed in the smart cycling helmet to receive the time queue of traffic data and the time queue of weather data; S3, extracting the user's current position from the time queue of the user's real-time position; S4, based on the time queue of the IMU data, correcting the position deviation of the user's current position to obtain the optimized user's current position; S5, inputting the optimized user's current position, the time queue of traffic data, and the time queue of weather data into the road environment comprehensive state evaluation module to obtain a road environment comprehensive state evaluation result; S6, based on the road environment comprehensive state evaluation result and the set cycling preference, and using the dynamic route optimization algorithm to adjust the cycling route planning in real time.

[0035] Specifically, S1 collects a time queue of the user's real-time location and a time queue of IMU data using the GPS module and IMU sensor deployed in the smart cycling helmet. It should be understood that in urban environments, GPS signals are susceptible to interference, particularly in areas with tall buildings, tunnels, or under bridges. Multipath (i.e., GPS signals reflected from multiple directions reaching the receiver) can cause positioning drift, making positioning results based solely on GPS less accurate. In these situations, GPS may not provide sufficiently accurate location information to support effective route planning. IMU sensors, on the other hand, operate independently of external radio signals, tracking the device's motion by measuring acceleration and angular velocity. Therefore, combining IMU data can complement GPS's shortcomings in complex environments, ensuring a relatively accurate position estimate even in conditions of poor GPS signal quality. Therefore, in the technical solution of this application, position deviation correction is performed on the user's current location based on the time queue of IMU data to obtain an optimized current location. It is worth noting that an IMU (Inertial Measurement Unit) sensor is a device used to measure an object's three-axis attitude angle (or angular rate) and acceleration. In the technical solution of this application, the IMU sensor can provide high-frequency acceleration and angular velocity data. This data is collected and formed into a time sequence, allowing the system to capture very subtle changes in the cyclist's head movement and even the entire riding process. This is very important for understanding the cyclist's real-time motion status, especially in situations where the GPS signal is poor or unavailable (such as when passing through a tunnel or the canyon effect between tall buildings). The continuous data provided by the IMU can serve as a supplement to help maintain the accuracy of position estimation.

[0036] Specifically, S2 utilizes an external data receiving module deployed within the smart cycling helmet to receive a time queue of traffic data and a time queue of weather data. It should be understood that by acquiring a time queue of real-time traffic data, the system can dynamically adjust cycling routes to avoid congested roads or accident scenes. Traditional route planning often relies on static map information and cannot promptly reflect unexpected road conditions, such as traffic accidents and road construction. The introduction of real-time traffic data allows cyclists to choose the optimal route based on the latest road conditions, avoiding delays and potential safety risks. Furthermore, weather conditions have a direct impact on cycling safety. For example, rain can make roads slippery, increasing the risk of falls; strong winds can affect a cyclist's balance, especially on overpasses or in open areas. By receiving a time queue of weather data, the system can provide early warning of impending severe weather and advise cyclists to take appropriate precautions or adjust their itineraries, thereby ensuring safety during cycling. Combining this external data with information collected by the IMU sensor and GPS module allows for a more comprehensive assessment of the overall cycling environment. It's worth mentioning that the external data receiving module connects to the internet via wireless communication technologies (such as Bluetooth, Wi-Fi, or cellular networks), enabling access to the latest traffic conditions and weather forecast information. Specifically, this module regularly downloads updated data streams from the cloud server. These data include traffic flow in the current area, accident reports, road construction information, and detailed weather forecasts, such as rainfall, temperature changes, wind speed, and other factors that may affect riding conditions.

[0037] Specifically, step S3 extracts the user's current location from the time queue of the user's real-time location. This time queue contains the user's latitude and longitude coordinates at different points in time, forming the basis for the user's movement trajectory. With accurate current location information, the system can more effectively execute the path planning algorithm.

[0038] In particular, the S4, based on the time queue of the IMU data, performs position deviation correction on the current position of the user to obtain the optimized current position of the user. Specifically, in a specific example of the present application, as Figure 3 As shown, the S4 includes: S41, performing structured coding on each IMU data in the time queue of the IMU data to obtain a time queue of IMU low-dimensional structured coding vectors; S42, inputting the time queue of the IMU low-dimensional structured coding vectors into the motion state nonlinear time series integral coding network to obtain a motion state time series integral cumulative coding vector; S43, determining a position offset compensation value based on the motion state time series integral cumulative coding vector; S44, performing position deviation correction on the user's current position based on the position offset compensation value to obtain the optimized user current position.

[0039] Specifically, S41 performs structured encoding on each IMU data item in the time queue of IMU data to obtain a time queue of IMU low-dimensional structured encoding vectors. It should be understood that in complex urban cycling scenarios, due to the coupling of multiple sources of noise caused by intermittent occlusion from overpasses, reflective interference from dense buildings, and sudden changes in riding posture (such as sudden braking or swerving), the time queue of raw IMU data often contains a mixture of high-dimensional redundant features and nonlinear motion noise. Traditional linear compensation models, when directly performing differential operations on the raw six-axis inertial data, have difficulty separating the coupling effects of the essential characteristics of the motion state and random sensor drift. To overcome this problem and convert the raw high-dimensional IMU data into more representative and easier-to-process low-dimensional structured encoding vectors, the technical solution of the present application performs low-dimensional structured encoding on each IMU data item in the time queue of IMU data using a fully connected layer to obtain a time queue of IMU low-dimensional structured encoding vectors. By applying low-dimensional structured encoding technology using a fully connected layer, key features that best reflect the rider's true motion state can be extracted from a large number of IMU sensor readings. Doing so not only simplifies the subsequent data processing process, but also improves the effectiveness of data feature extraction, allowing the system to more accurately understand the rider's movement pattern and maintain high positioning accuracy even in the case of poor GPS signals. In particular, the encoding layer adopts a weight sharing mechanism across the sensor axis system to perform feature orthogonal projection of the three-axis time series data of the accelerometer, gyroscope, and magnetometer in the time-space dimension, and screens out the intrinsic motion vectors that are strongly correlated with the riding movement pattern through a nonlinear activation function. Specifically, the six-dimensional raw data (three-axis acceleration + three-axis angular velocity) of the IMU at each timestamp is mapped to a three-dimensional latent space, while retaining the core movement characteristics of the riding posture, filtering out the dimensional noise caused by the inherent zero bias of the sensor and environmental magnetic interference. In this way, not only the accuracy of position positioning is improved, but also more reliable data support is provided for the comprehensive state assessment of the road environment and dynamic route optimization.

[0040] Specifically, S42 involves inputting the time sequence of the IMU low-dimensional structured code vectors into a nonlinear time-series integral coding network for motion states to obtain a motion state time-series integral cumulative code vector. It should be understood that the multimodal inertial characteristics of cycling motion exhibit both long-range dependencies and localized mutations. Traditional linear compensation models struggle to accurately capture complex nonlinear posture changes during cycling, such as centripetal acceleration disturbances from sharp turns, rapid acceleration or deceleration, and random vibration noise from bumpy roads. This makes it difficult for traditional position compensation models based on linear integration or simple recursive networks to accurately analyze the true motion trajectory. To simultaneously capture both global motion trends and instantaneous posture changes during cycling, key features that best reflect the rider's true motion state are extracted from the time sequence of the IMU low-dimensional structured code vectors. In the technical solution of this application, the time sequence of the IMU low-dimensional structured code vectors is input into a nonlinear time-series integral coding network for motion states to obtain a motion state time-series integral cumulative code vector. That is, through the multi-stage encoding method from coarse-grained to fine-grained and global to local, not only the overall motion pattern can be captured, but also the significance of the node's personalized features can be carefully enhanced, thereby realizing multi-level and multi-scale complex feature integration. In this process, the main pattern of cycling motion (such as the overall acceleration trend, the angular velocity change cycle) is first extracted through the information core coarse-grained convergence network to construct a global motion representation during the cycling process; then, the local abnormal motion fragments (such as the sudden drop in acceleration during emergency braking, the sudden turn when avoiding obstacles) are dynamically identified through the core convergence compensation factor, and the compensation weight is explicitly regulated by the gating function, so that the encoding vector can not only represent the smooth displacement in continuous riding, but also highlight the instantaneous impact of sudden actions. Among them, the coarse-grained convergence coding depicts the macroscopic displacement trend of the cyclist along the planned path, while the fine-grained compensation convergence accurately captures the microscopic trajectory deviation caused by environmental interference (such as crosswind disturbance) or active obstacle avoidance. In this way, the real position deviation of the cyclist can be captured, which helps to achieve more accurate position deviation correction, thereby improving the reliability and user experience of the entire route planning system. Specifically, in a specific example of the present application, such as Figure 4As shown, the S42 includes: S421, inputting the time queue of the IMU low-dimensional structured coding vector into the information kernel coarse-grained aggregation network to obtain the IMU coarse-grained aggregation coding vector; S422, based on the IMU coarse-grained aggregation coding vector, determining the IMU kernel aggregation compensation weights of each IMU low-dimensional structured coding vector in the time queue of the IMU low-dimensional structured coding vector to obtain a set of IMU kernel aggregation compensation weight factors; S423, inputting the set of IMU kernel aggregation compensation weight factors, the IMU coarse-grained aggregation coding vector and the time queue of the IMU low-dimensional structured coding vector into the node fine-grained dynamic compensation aggregation network to obtain the IMU fine-grained compensation aggregation coding vector; S424, inputting the IMU coarse-grained aggregation coding vector and the IMU fine-grained compensation aggregation coding vector into the residual unit to obtain the motion state time series integral accumulation coding vector.

[0041] More specifically, the S421 inputs the time queue of the IMU low-dimensional structured coding vector into the information kernel coarse-grained aggregation network to obtain the IMU coarse-grained aggregation coding vector. Since complex environments such as urban canyons and overpasses can cause multipath interference of GPS signals to form cumulative position deviations, and cycling motion includes nonlinear posture changes such as sharp turns and bumpy roads, it is difficult for traditional linear compensation models to accurately capture these dynamic characteristics. At this time, although the accelerometer and gyroscope data collected by the IMU sensor can provide high-frequency motion state information, their original signals have high-frequency noise interference and nonlinear coupling problems, and direct use will lead to amplification of motion state estimation errors. In the technical solution of the present application, the time queue of the IMU low-dimensional structured coding vector is input into the information kernel coarse-grained aggregation network to obtain the IMU coarse-grained aggregation coding vector. Here, the IMU low-dimensional code is globally compressed through the information kernel coarse-grained aggregation network. In essence, the similarity structure between nodes is reconstructed in the implicit high-dimensional space through a nonlinear mapping mechanism to summarize the global features of the time queue of the IMU low-dimensional structured code vector, so as to extract the main mode that best represents the rider's true motion state. In this way, the overall motion trend of the rider in the entire time series can be effectively captured, which is crucial for understanding the rider's macroscopic movement pattern. Specifically, in an embodiment of the present application, the time queue of the IMU low-dimensional structured code vector is input into the information kernel coarse-grained aggregation network using the following information kernel coarse-grained aggregation formula to obtain the IMU coarse-grained aggregation code vector; wherein, the information kernel coarse-grained aggregation formula is:

[0042]

[0043] in, represents the time queue of the IMU low-dimensional structured encoding vector, are the first, second, and third time queues of the IMU low-dimensional structured coding vector respectively. and IMU low-dimensional structured encoding vector, and Indicates taking the maximum and minimum values ​​in a vector. represents the global feature information nuclear factor, for function, is the global feature information kernel weight, is the number of vectors in the time queue of the IMU low-dimensional structured encoding vector, Represents the IMU coarse-grained aggregated encoding vector.

[0044] More specifically, S422 determines, based on the IMU coarse-grained converged code vector, an IMU kernel convergence compensation weight for each IMU low-dimensional structured code vector in the time sequence of the IMU low-dimensional structured code vector to obtain a set of IMU kernel convergence compensation weight factors. In an embodiment of the present application, kernel convergence compensation factors are first calculated for each IMU low-dimensional structured code vector in the time sequence of the IMU low-dimensional structured code vector relative to the IMU coarse-grained converged code vector to obtain a set of IMU kernel convergence compensation factors. It should be understood that while the information kernel coarse-grained convergence network can effectively capture the overall motion trend of the cyclist throughout the entire time series, this compressive aggregation process of global features may dilute or even completely lose the personalized information of certain important nodes. For example, during nonlinear posture changes such as sharp turns or rapid acceleration and deceleration, local detailed features are crucial for accurate position deviation correction. If these details are ignored or lost, significant deviations between the position estimate and the actual trajectory will result. Therefore, the kernel convergence compensation factor of each IMU low-dimensional structured code vector in the temporal sequence of the IMU low-dimensional structured code vector is calculated relative to the IMU coarse-grained convergence code vector. This extracts key features reflecting local details from the temporal sequence of the IMU low-dimensional structured code vector and generates a set of IMU kernel convergence compensation factors. In other words, by calculating the kernel convergence compensation factor, the correction rule for each node (i.e., the IMU data at each time point) relative to the global features is dynamically adjusted, ensuring that the personalized information of important nodes is preserved and providing strong support for subsequent fine-grained compensation modeling.

[0045] In particular, to compensate for the deviation between node feature vectors and coarse-grained aggregated encoding vectors, in a preferred example, a game-based counterfactual framework is used to dynamically quantify the need for correcting the deviation between local features and global representations. Because GPS signals in complex environments such as urban canyons and overpasses are susceptible to multipath interference, resulting in cumulative drift, and nonlinear posture changes during cycling, such as sharp turns and bumpy roads, can cause the IMU raw data to contain high-frequency noise and nonlinear coupling errors, traditional global feature aggregation may over-compress local motion details (such as inertial perturbations caused by arm swing or effective motion information in short-term vibration noise). In this case, the kernel convergence compensation factor is calculated by mapping the vector norm difference between each node feature vector and the coarse-grained encoding vector into the decision point loss of the policy action, and the normalized counterfactual regret value is used to measure the potential benefits of local features not retained in global modeling. This compensation mechanism not only captures the conflict between centripetal acceleration disturbances and global motion trends during sharp turns, but also dynamically adjusts the compensation weights under explicit control of the gating function. This allows the model to maintain consistency in overall motion direction (such as the angular velocity trend during continuous steering) while retaining the local salience of instantaneous obstacle avoidance maneuvers (such as sudden IMU signals during emergency lane changes) in scenarios such as elevated bridges. By treating the compensation factor as a potential reward bias for "not taking action," the model can adaptively balance global and local feature conflicts under the assumption of information kernel compression. Specifically, in an embodiment of the present application, the specific steps for calculating the core convergence compensation factor of each IMU low-dimensional structured coding vector in the time queue of the IMU low-dimensional structured coding vector relative to the IMU coarse-grained convergence coding vector are as follows: modulating the IMU low-dimensional structured coding vector and the IMU coarse-grained convergence coding vector to obtain an IMU low-dimensional structured coding modulation vector and an IMU coarse-grained convergence coding modulation vector; calculating the compensation information between the IMU low-dimensional structured coding modulation vector and the IMU coarse-grained convergence coding modulation vector to obtain an IMU core convergence compensation weight vector; calculating the IMU core convergence deviation compensation factor based on the IMU low-dimensional structured coding vector and the IMU coarse-grained convergence coding vector; calculating the core convergence compensation factor based on the IMU core convergence compensation weight vector and the IMU core convergence deviation compensation factor to obtain the IMU core convergence compensation factor.Among them, the specific process of calculating the IMU core convergence deviation compensation factor based on the IMU low-dimensional structured coding vector and the IMU coarse-grained convergence coding vector is as follows: in response to the two-norm of the IMU low-dimensional structured coding vector being less than the two-norm of the IMU coarse-grained convergence coding vector, the two-norm of the IMU low-dimensional structured coding vector is calculated and divided by the two-norm of the IMU coarse-grained convergence coding vector and then added to the constant one, and the logarithmic function value with base 2 of the numerical value is calculated; in response to the two-norm of the IMU low-dimensional structured coding vector being greater than or equal to the two-norm of the IMU coarse-grained convergence coding vector, the two-norm of the IMU low-dimensional structured coding vector is calculated and divided by the two-norm of the IMU coarse-grained convergence coding vector.

[0046] More specifically, in this embodiment, the following kernel convergence compensation calculation formula is used to calculate the kernel convergence compensation factor of each IMU low-dimensional structured code vector in the time queue of the IMU low-dimensional structured code vector relative to the IMU coarse-grained converged code vector to obtain a set of IMU kernel convergence compensation factors; wherein, the kernel convergence compensation calculation formula is:

[0047]

[0048] in, For point convolution layer processing, is the first weight matrix, is the second weight matrix, for function, is the time queue of the IMU low-dimensional structured encoding vector IMU low-dimensional structured coded modulation vector, is the IMU coarse-grained aggregate coded modulation vector, For positional subtraction, To take the absolute value, Aggregate compensation weight vector for the IMU core, represents the two-norm of the vector, is the logarithmic function value with base 2, represents the IMU kernel convergence bias compensation factor, and denote the compensation weight matrix and compensation bias vector respectively, To compensate the modulation vector, for Corresponding IMU kernel convergence compensation factor.

[0049] Furthermore, the set of IMU core convergence compensation factors is explicitly modeled using a gating function to obtain a set of IMU core convergence compensation weight factors. Because GPS signals in complex environments such as urban canyons and overpasses are susceptible to multipath interference, leading to cumulative drift, and nonlinear posture changes during cycling, such as sharp turns and bumpy roads, can cause the IMU raw data to contain high-frequency noise and nonlinear coupling errors, traditional global feature compression can over-suppress local motion details. By explicitly modeling the core convergence compensation factors using a gating function, the strength of the compensation factors can be parameterized, allowing them to dynamically select local information that is critical for global feature correction under nonlinear constraints. This allows the system to filter out only locally significant features while appropriately suppressing irrelevant or redundant information. This allows for a more precise description of the impact of each node on global features, ensuring accurate position information even in complex dynamic environments. That is, the effectiveness of the compensation factor is flexibly adjusted according to the current motion state, and the core convergence compensation weight factor obtained by explicit modeling of the gated function can dynamically adjust the correction strength of the local feature to the global encoding, so that the model can retain the key local information of the centripetal acceleration during sharp turns in scenes such as elevated bridge areas, and suppress the interference of redundant vibration noise on the overall motion direction. In this way, the model can balance the coherence of the global motion direction and the instantaneous response of the local obstacle avoidance action in scenes such as elevated bridge areas. Specifically, in an embodiment of the present application, the set of IMU core convergence compensation factors is explicitly modeled based on the compensation of the gated function using the following compensation formula to obtain the set of IMU core convergence compensation weight factors; wherein, the compensation formula is:

[0050]

[0051] in, is the preset threshold, To compensate for explicit modeling operations, for The corresponding IMU kernel convergence compensation weight factor.

[0052] More specifically, in S423, the set of IMU core convergence compensation weight factors, the IMU coarse-grained converged code vector, and the time queue of the IMU low-dimensional structured code vector are input into a node fine-grained dynamic compensation convergence network to obtain an IMU fine-grained compensation convergence code vector. That is, by performing node fine-grained dynamic compensation convergence analysis on the set of IMU core convergence compensation weight factors, the IMU coarse-grained converged code vector, and the time queue of the IMU low-dimensional structured code vector, the key features that best reflect the rider's actual motion state are extracted from the set of IMU core convergence compensation weight factors, the IMU coarse-grained converged code vector, and the time queue of the IMU low-dimensional structured code vector. In this way, the model can dynamically optimize the coordinated expression of global and local features based on the motion state (e.g., riding speed, steering amplitude) and environmental complexity (e.g., road density, obstacle distribution) in real-time scenarios, thereby improving the accuracy of position deviation correction. Specifically, in an embodiment of the present application, the set of the IMU core convergence compensation weight factors, the IMU coarse-grained convergence coding vector, and the time queue of the IMU low-dimensional structured coding vector are input into the node fine-grained dynamic compensation convergence network to obtain the IMU fine-grained compensation convergence coding vector according to the following fine-grained dynamic compensation convergence formula; wherein, the fine-grained dynamic compensation convergence formula is:

[0053]

[0054] in, Represents the IMU fine-grained compensation aggregate coding vector.

[0055] More specifically, in S424, the IMU coarse-grained converged coding vector and the IMU fine-grained compensation converged coding vector are input into the residual unit to obtain the motion state time series integral cumulative coding vector. It should be understood that although the information kernel coarse-grained convergence network can effectively capture the overall motion trend of the cyclist and correct the global features by calculating the kernel convergence compensation factor, and the fine-grained dynamic compensation convergence network further refines the local detail features, if these two parts of information are directly merged, some key details may be lost or unnecessary noise may be introduced. Therefore, in the technical solution of the present application, the IMU coarse-grained converged coding vector and the IMU fine-grained compensation converged coding vector are input into the residual unit to extract the key features that best reflect the actual motion state of the cyclist from the IMU coarse-grained converged coding vector and the IMU fine-grained compensation converged coding vector, and generate an optimized motion state time series integral cumulative coding vector. In addition, by inputting the IMU coarse-grained converged coding vector and the IMU fine-grained compensated converged coding vector into the residual unit, the system can avoid the gradient vanishing problem while ensuring feature transfer, thereby making the model easier to optimize and further improving the significance and robustness of the coding results. Through such fine adjustments, the smart cycling helmet can provide cyclists with reliable location services and optimal cycling route suggestions in various complex environments. Specifically, in an embodiment of the present application, the IMU coarse-grained converged coding vector and the IMU fine-grained compensated converged coding vector are input into the residual unit using the following residual calculation formula to obtain the motion state time series integral cumulative coding vector; wherein, the residual calculation formula is:

[0056]

[0057] in, and is the residual weight coefficient, Represents the time series integral cumulative coding vector of the motion state.

[0058] Specifically, S43 determines a position offset compensation value based on the motion state time series integral cumulative code vector. That is, in the technical solution of the present application, the motion state time series integral cumulative code vector is input into a decoder-based position compensation estimation module to obtain a position offset compensation value. During this process, the decoder extracts the key features that best reflect the rider's actual motion state from the motion state time series integral cumulative code vector and generates an accurate position offset compensation value. The position offset compensation value is calculated based on the rider's actual motion trajectory and can effectively compensate for GPS positioning errors and noise in IMU data, ensuring the accuracy and reliability of position information. Furthermore, the user's current position is corrected for position deviation based on the position offset compensation value to obtain the optimized user current position. By correcting the user's current position for position deviation based on the position offset compensation value, the system can significantly improve the accuracy of position estimation, especially in complex urban environments. This means that in various complex urban riding scenarios, such as elevated bridge areas or tunnels, riders can obtain continuous and stable position information, greatly improving riding safety and comfort.

[0059] Specifically, in S44, the position offset of the user's current location is corrected based on the position offset compensation value to obtain the optimized user's current location. That is, the calculated position offset compensation value is combined with the preliminary position estimate provided by the current GPS. In this way, errors introduced by GPS positioning can be effectively corrected, particularly positioning drift caused by multipath effects, which are common in complex environments such as urban canyons, between high-rise buildings, or in tunnels. Ultimately, after such position offset correction, the system can obtain a more accurate user's current location, namely, the optimized user's current location.

[0060] Specifically, S5 inputs the optimizing user's current location, the time series of traffic data, and the time series of weather data into the road environment comprehensive status assessment module to generate a comprehensive road environment status assessment result. The road environment comprehensive status assessment module is more than just a simple data processing unit; it is an intelligent analysis platform that integrates multiple algorithms and technologies. By integrating multi-source data, it provides a comprehensive, dynamic, and personalized road environment assessment result, recommending optimal routes for cyclists and ensuring their safety and comfort. The module utilizes advanced data analysis techniques, such as machine learning algorithms, time series analysis, and pattern recognition, to deeply analyze this data. Specifically, the road environment comprehensive status assessment module conducts a detailed analysis of traffic conditions near the optimizing user's current location, identifying, for example, the presence of congested sections or accident scenes, and based on this information, calculates the efficiency of each possible route. Furthermore, the module also considers the impact of weather factors on cycling safety. For example, if heavy rain is imminent on the road ahead, it may recommend a safer route. Throughout this process, the road environment comprehensive status assessment module goes beyond simply aggregating and displaying data; instead, it comprehensively considers all relevant factors to generate a comprehensive road environment status assessment result. This includes but is not limited to recommended optimal cycling routes, estimated arrival times, and potential risk warnings.

[0061] Specifically, S6 utilizes a dynamic route optimization algorithm to adjust the cycling route plan in real time based on the comprehensive road environment assessment results and set cycling preferences. That is, after obtaining the comprehensive road environment assessment results, the system combines this information with the user's cycling preferences. Cycling preferences may include, but are not limited to, a preference for scenic routes, a desire to take the shortest path, or avoiding busy roads. These preferences are typically set by the user in the system or automatically generated based on historical cycling data. By understanding the user's personal preferences, the system can better meet their specific needs and provide more personalized services. Therefore, in the technical solution of the present application, a dynamic route optimization algorithm is used to adjust the cycling route plan in real time based on the comprehensive road environment assessment results and set cycling preferences. The dynamic route optimization algorithm utilizes a series of advanced computational methods to process this complex data, including but not limited to heuristic search algorithms, genetic algorithms, and simulated annealing algorithms. These algorithms are able to find the optimal solution among a large number of feasible paths, balancing multiple dimensions such as time cost, distance, and safety. More importantly, they can be dynamically adjusted based on real-time data updates, meaning that as road conditions change during cycling, the system can instantly recalculate the optimal route. The system then provides the user with the optimized cycling route. This is usually presented as navigation instructions, guiding the rider to follow the recommended path. This not only improves cycling safety and efficiency, but also provides users with a more intelligent and comfortable cycling experience.

[0062] In summary, the route intelligent planning method based on the smart cycling helmet according to the embodiment of the present application is explained. It uses the time queue of data continuously collected by the IMU sensor, converts the IMU raw data through the low-dimensional structured coding of the fully connected layer, and inputs it into a specially designed motion state nonlinear time series integral coding network for processing. The key features that can reflect the actual motion trajectory are extracted from it, and the position offset compensation value is calculated by the decoder to finally achieve accurate correction of the user's current position. In this way, reliable location information can be provided even when the GPS signal is weak or missing, ensuring the accuracy and reliability of path planning, thereby improving the safety and experience of cycling.

[0063] Furthermore, a route intelligent planning system based on a smart cycling helmet is also provided.

[0064] Figure 5 FIG is a block diagram of an intelligent route planning system based on an intelligent cycling helmet according to an embodiment of the present application. Figure 5As shown, according to an embodiment of the present application, the route intelligent planning system 300 based on the smart cycling helmet includes: a data acquisition module 310, which is used to collect the time queue of the user's real-time position and the time queue of IMU data through the GPS module and IMU sensor deployed in the smart cycling helmet; an external data receiving module 320, which is used to use the external data receiving module deployed in the smart cycling helmet to receive the time queue of traffic data and the time queue of weather data; a user current position extraction module 330, which is used to extract the user's current position from the time queue of the user's real-time position; a position deviation correction module 340, which is used to correct the position deviation of the user's current position based on the time queue of the IMU data to obtain an optimized user current position; a road environment comprehensive state evaluation module 350, which is used to input the optimized user current position, the time queue of traffic data, and the time queue of weather data into the road environment comprehensive state evaluation module to obtain a road environment comprehensive state evaluation result; a route real-time adjustment module 360, which is used to adjust the cycling route planning in real time based on the road environment comprehensive state evaluation result and the set cycling preferences, and use a dynamic route optimization algorithm.

[0065] As described above, the route intelligent planning system 300 based on the smart cycling helmet according to the embodiment of the present application can be implemented in various wireless terminals, such as a server with an intelligent route planning algorithm based on the smart cycling helmet. In one possible implementation, the route intelligent planning system 300 based on the smart cycling helmet according to the embodiment of the present application can be integrated into the wireless terminal as a software module and / or hardware module. For example, the route intelligent planning system 300 based on the smart cycling helmet can be a software module in the operating system of the wireless terminal, or it can be an application developed for the wireless terminal; of course, the route intelligent planning system 300 based on the smart cycling helmet can also be one of the many hardware modules of the wireless terminal.

[0066] Alternatively, in another example, the smart cycling helmet-based route intelligent planning system 300 and the wireless terminal may also be separate devices, and the smart cycling helmet-based route intelligent planning system 300 may be connected to the wireless terminal via a wired and / or wireless network and transmit interactive information in accordance with an agreed data format.

[0067] While various embodiments of the present disclosure have been described above, the above descriptions are illustrative, non-exhaustive, and not intended to be limiting of the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is selected to best explain the principles of the embodiments, their practical applications, or improvements to existing technologies, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A route intelligent planning method based on a smart cycling helmet, characterized in that: include: The GPS module and IMU sensor deployed in the smart cycling helmet collect the time queue of the user's real-time location and the time queue of IMU data; Using an external data receiving module deployed in the smart cycling helmet to receive a time queue of traffic data and a time queue of weather data; Extracting the user's current location from the time queue of the user's real-time location; Based on the time queue of the IMU data, performing position deviation correction on the current position of the user to obtain an optimized current position of the user; Inputting the optimized user's current location, the time queue of the traffic data and the time queue of the weather data into a road environment comprehensive state evaluation module to obtain a road environment comprehensive state evaluation result; Based on the comprehensive status evaluation results of the road environment and the set riding preferences, a dynamic route optimization algorithm is used to adjust the riding route planning in real time.

2. The route intelligent planning method based on the smart cycling helmet according to claim 1 is characterized in that: Based on the time queue of the IMU data, the position deviation of the user's current position is corrected to obtain an optimized user's current position, including: Performing structured coding on each IMU data in the time queue of the IMU data to obtain a time queue of IMU low-dimensional structured coding vectors; Inputting the time queue of the IMU low-dimensional structured coding vector into the motion state nonlinear time series integral coding network to obtain the motion state time series integral cumulative coding vector; Determine a position offset compensation value based on the motion state time series integral cumulative coding vector; The user's current position is corrected for position deviation based on the position offset compensation value to obtain the optimized user's current position.

3. The route intelligent planning method based on the smart cycling helmet according to claim 2 is characterized in that: Performing structured coding on each IMU data in the time queue of the IMU data to obtain a time queue of an IMU low-dimensional structured coding vector, including: Each IMU data in the time queue of the IMU data is subjected to low-dimensional structured coding based on a fully connected layer to obtain a time queue of IMU low-dimensional structured coding vectors.

4. The route intelligent planning method based on the smart cycling helmet according to claim 3 is characterized in that: Inputting the time queue of the IMU low-dimensional structured coding vector into the motion state nonlinear time series integral coding network to obtain the motion state time series integral cumulative coding vector, including: Inputting the time queue of the IMU low-dimensional structured coding vector into the information core coarse-grained aggregation network to obtain the IMU coarse-grained aggregation coding vector; Based on the IMU coarse-grained converged coding vector, determining the IMU core convergence compensation weight of each IMU low-dimensional structured coding vector in the time queue of the IMU low-dimensional structured coding vector to obtain a set of IMU core convergence compensation weight factors; Input the set of the IMU core convergence compensation weight factors, the IMU coarse-grained converged coding vector and the time queue of the IMU low-dimensional structured coding vector into a node fine-grained dynamic compensation convergence network to obtain an IMU fine-grained compensation converged coding vector; The IMU coarse-grained aggregated coding vector and the IMU fine-grained compensation aggregated coding vector are input into a residual unit to obtain the motion state time series integral cumulative coding vector.

5. The route intelligent planning method based on the smart cycling helmet according to claim 4 is characterized in that: Based on the IMU coarse-grained converged coding vector, determining the IMU core convergence compensation weight of each IMU low-dimensional structured coding vector in the time queue of the IMU low-dimensional structured coding vector to obtain a set of IMU core convergence compensation weight factors, including: Calculating the core convergence compensation factor of each IMU low-dimensional structured code vector in the time queue of the IMU low-dimensional structured code vector relative to the IMU coarse-grained convergence code vector to obtain a set of IMU core convergence compensation factors; The set of IMU core convergence compensation factors is subjected to explicit compensation modeling based on a gating function to obtain a set of IMU core convergence compensation weight factors.

6. The route intelligent planning method based on the intelligent cycling helmet according to claim 5 is characterized in that: Calculating the core convergence compensation factor of each IMU low-dimensional structured coding vector in the time queue of the IMU low-dimensional structured coding vector relative to the IMU coarse-grained convergence coding vector to obtain a set of IMU core convergence compensation factors, including: Modulating the IMU low-dimensional structured coding vector and the IMU coarse-grained converged coding vector to obtain an IMU low-dimensional structured coding modulation vector and an IMU coarse-grained converged coding modulation vector; Calculating compensation information between the IMU low-dimensional structured coded modulation vector and the IMU coarse-grained converged coded modulation vector to obtain an IMU core converged compensation weight vector; Calculating an IMU core convergence bias compensation factor based on the IMU low-dimensional structured coding vector and the IMU coarse-grained convergence coding vector; A core convergence compensation factor is calculated based on the IMU core convergence compensation weight vector and the IMU core convergence deviation compensation factor to obtain the IMU core convergence compensation factor.

7. The route intelligent planning method based on the intelligent cycling helmet according to claim 6 is characterized in that: Based on the IMU low-dimensional structured coding vector and the IMU coarse-grained converged coding vector, an IMU core convergence deviation compensation factor is calculated, including: In response to the fact that the second norm of the IMU low-dimensional structured coding vector is less than the second norm of the IMU coarse-grained converged coding vector, the second norm of the IMU low-dimensional structured coding vector is calculated and divided by the second norm of the IMU coarse-grained converged coding vector and then added to a constant 1, and a logarithmic function value with base 2 of the numerical value is calculated; In response to the binary norm of the IMU low-dimensional structured coding vector being greater than or equal to the binary norm of the IMU coarse-grained aggregated coding vector, the binary norm of the IMU low-dimensional structured coding vector is calculated by dividing the binary norm of the IMU coarse-grained aggregated coding vector by the binary norm of the IMU coarse-grained aggregated coding vector.

8. The route intelligent planning method based on the intelligent cycling helmet according to claim 7 is characterized in that: Determining a position offset compensation value based on the motion state time series integral cumulative coding vector includes: The motion state time series integral accumulation coding vector is input into a decoder-based position compensation estimation module to obtain a position offset compensation value.

9. A route intelligent planning system based on a smart cycling helmet, characterized in that: include: A data collection module is used to collect the time queue of the user's real-time position and the time queue of IMU data through the GPS module and IMU sensor deployed in the smart cycling helmet; An external data receiving module, used for receiving a time queue of traffic data and a time queue of weather data using the external data receiving module deployed in the smart cycling helmet; A user current location extraction module, used to extract the user current location from the time queue of the user's real-time location; A position deviation correction module, used for performing position deviation correction on the user's current position based on the time queue of the IMU data to obtain an optimized user's current position; A road environment comprehensive state evaluation module, used for inputting the optimized user's current position, the time queue of the traffic data and the time queue of the weather data into the road environment comprehensive state evaluation module to obtain a road environment comprehensive state evaluation result; The real-time route adjustment module is used to adjust the cycling route planning in real time based on the comprehensive status evaluation results of the road environment and the set cycling preferences, and use a dynamic route optimization algorithm.

Images