A cloud-based intelligent driving assistance system and method for off-road environments

Abstract

This invention provides a cloud-based intelligent driving assistance system and method for off-road environments. The method includes: Step 1, when a vehicle first enters the current area, real-time environmental information around the vehicle is collected; Step 2, a map is constructed based on the environmental information for use by the intelligent driving vehicle in assisted driving; Step 3, the driver's operational behavior data during manual driving is recorded; Step 4, the communication module uploads the collected environmental information and recorded operational behavior data to a cloud server for reception and storage; Step 5, when subsequent vehicles enter the current area, the cloud server distributes the stored environmental information and driving behavior data of the current area to the subsequent vehicles; Step 6, the subsequent vehicles, based on the environmental information and driving behavior data distributed by the cloud server, implement intelligent driving assistance functions through path planning algorithms, obstacle avoidance algorithms, and speed control algorithms.

Description

Technical Field

[0001] This application relates to the field of intelligent driving technology, and in particular to an intelligent driving assistance system and method for off-road environments based on cloud sharing. Background Technology

[0002] In off-road environments, the terrain is complex and varied, and detailed electronic maps are often lacking. For novice drivers, relying solely on manual driving poses a high safety risk. Current intelligent driving technologies largely depend on high-precision / lightweight maps and advanced sensor systems, but these conditions are often difficult to meet in off-road environments. Therefore, there is an urgent need for a solution that can provide intelligent driving assistance in off-road environments. Summary of the Invention

[0003] In view of this, the present invention provides an intelligent driving assistance system and method for off-road environments based on cloud sharing, in order to solve the technical problems that the existing technology cannot adapt to the driving needs of off-road environments and lacks sharing of driver operation behavior data.

[0004] This invention provides a cloud-based intelligent driving assistance system for off-road environments. The system includes: a sensor input module, mounted on the intelligent driving vehicle, for collecting environmental information surrounding the vehicle; a map building module, mounted on the intelligent driving vehicle and connected to the sensor input module, for building a map based on the environmental information for assisted driving; a data acquisition module, mounted on the intelligent driving vehicle, for recording the driver's operational behavior data during manual driving; a communication module, mounted on the intelligent driving vehicle and connected to the sensor input module, map building module, and data acquisition module, and connected to a cloud server via a network, for uploading collected environmental information and recorded operational behavior data to the cloud server, and receiving environmental information and operational behavior data uploaded by other vehicles from the cloud server; a cloud server, interacting with the communication module via wireless communication technology, for storing and managing the environmental information and driving behavior data uploaded by vehicles, and sending relevant data to subsequent vehicles entering the same area; and an intelligent driving control module, mounted on the intelligent driving vehicle and connected to the communication module, for implementing intelligent driving assistance functions based on the environmental information and driving behavior data sent from the cloud server.

[0005] Furthermore, the environmental information includes terrain, obstacles, and road slope.

[0006] Furthermore, the operational behavior data includes steering wheel angle, accelerator pedal position, and brake pedal pressure.

[0007] Furthermore, the sensor input module includes a lidar, a camera, and a millimeter-wave radar.

[0008] This invention also provides a method for a cloud-based intelligent driving assistance system for off-road environments, comprising: Step 1, when a vehicle first enters the current area, real-time collection of environmental information surrounding the vehicle; Step 2, constructing a map based on the environmental information for use by the intelligent driving vehicle in assisted driving; Step 3, recording the driver's operational behavior data during manual driving; Step 4, the communication module uploading the collected environmental information and recorded operational behavior data to a cloud server for reception and storage by the cloud server; Step 5, when subsequent vehicles enter the current area, the cloud server distributing the stored environmental information and driving behavior data of the current area to the subsequent vehicles; Step 6, the subsequent vehicles, based on the environmental information and driving behavior data distributing from the cloud server, implement intelligent driving assistance functions through path planning algorithms, obstacle avoidance algorithms, and speed control algorithms.

[0009] Furthermore, the intelligent driving assistance functions include automatic obstacle avoidance, path planning, and speed control.

[0010] Furthermore, step 2 includes: step 21, performing data calibration and synchronization, noise filtering, and data augmentation on the received environmental information; step 22, constructing a map using LiDAR positioning, camera visual perception, and multi-sensor fusion positioning and mapping methods; step 23, storing the constructed map using point cloud maps, grid maps, and topology maps; and step 24, updating and optimizing the map using incremental updates, closed-loop detection and optimization, and data compression and simplification methods.

[0011] Furthermore, step 3 includes: step 31, acquiring driver operation behavior data through the vehicle's electronic control unit; step 32, transmitting the operation behavior data to the data acquisition module through the vehicle's communication bus and recording it in real time.

[0012] Furthermore, step 5 includes: step 51, after a subsequent vehicle enters the current area, it sends location information to the cloud server through the communication module; step 52, the cloud server retrieves and finds matching environmental information and driving behavior data based on the location information using a location matching algorithm; step 53, the cloud server sends the retrieved data to the corresponding vehicle through the communication module.

[0013] Furthermore, the cloud server employs a distributed storage and computing architecture to efficiently store and manage large amounts of data.

[0014] This invention provides an intelligent driving assistance system and method for off-road environments based on cloud sharing. This technical solution collects environmental information and manual driving behavior data when a vehicle first passes through an unfamiliar environment and shares it to the cloud to enable intelligent driving assistance functions for subsequent vehicles in the same area, thereby ensuring the driving safety of drivers, especially novice drivers, in off-road environments. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the framework of an intelligent driving assistance system for off-road environments based on cloud sharing, provided by the present invention. Figure 2 This is a schematic diagram of a cloud-based intelligent driving assistance method for off-road environments provided by the present invention; Figure 3 This is a schematic diagram of the method for constructing a map based on environmental information provided by the present invention; Figure 4 This is a schematic diagram of the method provided by the present invention for a cloud server to send the stored environmental information and driving behavior data of the current area to subsequent vehicles. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Device Example: This invention provides a cloud-based intelligent driving assistance system for off-road environments, such as... Figure 1 As shown, the system includes a sensor input module, a map building module, a data acquisition module, a communication module, a cloud server, and an intelligent driving control module. The sensor input module, map building module, data acquisition module, communication module, and intelligent driving control module are located on the vehicle side, while the cloud server is located in the cloud.

[0018] The sensor input module, installed on the intelligent driving vehicle, is used to collect environmental information around the vehicle; The map building module, installed on the intelligent driving vehicle and connected to the sensor input module, is used to select geologically stable natural objects as feature points based on environmental information. It integrates the feature points with geometric shape, material reflectivity and terrain slope to generate rotation-invariant feature codes, which are then matched with a pre-established seasonal feature library to extract off-road features. The off-road features are then used to build a map for the intelligent driving vehicle to use for assisted driving. The data acquisition module, installed on the intelligent driving vehicle, is used to record the driver's operational intensity characteristics, dynamic response characteristics, and environmental adaptation characteristics during manual driving in the off-road area. The communication module, installed on the intelligent driving vehicle, is connected to the sensor input module, map building module, and data acquisition module respectively. It is connected to the cloud server via the network and is used to upload newly built maps and recorded operation behavior data to the cloud server, as well as to receive adjusted map information and simulated driving style data sent by the cloud server. The cloud server interacts with the communication module via wireless communication technology. It generates simulated driving style data based on a multi-scenario classifier and recorded operation behavior data. Then, it uses a hidden Markov model to model the driving style data so that it changes with time and scenario. A deep learning-based difference detection network is used to downweight the original map information in the cloud server and adaptively adjust it according to the newly constructed map through a map confidence decay model to achieve map update and optimization. The adjusted map information and simulated driving style data are then sent to subsequent vehicles. The intelligent driving control module, installed on the intelligent driving vehicle and connected to the communication module, is used to realize the vehicle's intelligent driving assistance functions based on map information and simulated driving style data sent by the cloud server, through path planning algorithms, obstacle avoidance algorithms, and speed control algorithms.

[0019] This invention provides an intelligent driving assistance system for off-road environments based on cloud sharing. The system includes a sensor input module, a map building module, a data acquisition module, a communication module, a cloud server, and an intelligent driving control module. It collects map environment information and manual driving behavior data when the vehicle first passes through an unfamiliar environment and shares it to the cloud to enable intelligent driving assistance functions for subsequent vehicles in the same area. This solves the technical problems of existing technologies, such as lack of data sharing mechanism, inability to meet the needs of off-road driving, and insufficient real-time performance.

[0020] Method Example: This invention provides a cloud-based intelligent driving assistance system for off-road environments, such as... Figure 2 As shown, the method includes the following steps.

[0021] Step 1: When the vehicle first enters the current off-road area, collect environmental information around the vehicle in real time and select natural objects with stable geological structures as feature points. Off-road environment map construction differs fundamentally from traditional urban road maps, primarily in the following aspects: 1. Unstructured terrain processing: There are no clear road boundaries or lane lines; terrain features are complex and varied (slope, surface material, obstacle distribution), requiring the identification of natural features as navigation benchmarks (landmarks such as rocks and trees); 2. Dynamic environmental adaptability: Surface conditions change in real time with weather (mudliness, humidity, etc.); temporary obstacles frequently appear (fallen trees, animal activity, etc.); and vegetation growth causes seasonal changes; 3. Multi-dimensional information fusion: Geometric information, surface attributes, and dynamic characteristics must be recorded simultaneously; different vehicle types exhibit significant differences in their passability on the same terrain. Therefore, before mapping, environmental information needs to be collected, dynamic interference such as vegetation eliminated, and naturally stable feature points selected. A seasonal feature database (dry season / rainy season feature correspondence) also needs to be established to extract off-road features. The relationship between the selected natural feature points and geometric parameters is shown in Table 1.

[0022]

[0023] Table 1 Geometric parameters of natural feature points Step 2: The feature points are fused with geometry, material reflectivity and terrain slope to generate rotation-invariant feature codes, which are then matched with a pre-established seasonal feature library to extract off-road features. Maps are then constructed using these off-road features for use by intelligent driving vehicles in assisted driving. In addition to the geometric parameters mentioned above, when generating feature codes, feature points also incorporate material reflectivity and terrain slope, as shown in Tables 2 and 3.

[0024]

[0025] Table 2. LiDAR reflectivity of typical off-road materials

[0026] Table 3 Terrain Slope and Category The map includes: categorization of passable areas, terrain difficulty rating, recommended routes, and special risk markers. The intelligent driving assistance functions include automatic obstacle avoidance, route planning, and speed control. Figure 3 As shown, the method for constructing the map includes: Step 21: Perform data calibration and synchronization, noise filtering, and data augmentation on the received off-road feature information; Step 22: Map construction is achieved by using LiDAR positioning, camera visual perception, and multi-sensor fusion positioning and mapping methods. Step 23: Store the constructed map using point cloud map, grid map, and topology map; Step 24: Map updates and optimizations are achieved using incremental updates, closed-loop detection and optimization, and data compression and simplification.

[0027] Step 3: Record the driver's operational intensity characteristics, dynamic response characteristics, and environmental adaptation characteristics during manual driving in the off-road area. The communication module then uploads the newly constructed map and the recorded operational behavior data to the cloud server. This step records the driving style exhibited by the driver during manual driving, including: 1. Operation intensity characteristics: accelerator / brake pedal change rate (Δ% / s) (aggressive driving is usually accompanied by rapid and deep pressing of the accelerator or sudden braking), steering wheel angle velocity (° / s) (aggressive driving is more abrupt in steering, while conservative driving is smoother); 2. Dynamic response characteristics: peak longitudinal acceleration (m / s²) (reflecting the intensity of acceleration / braking), standard deviation of lateral acceleration (reflecting cornering stability, with conservative driving having a lower standard deviation); 3. Environmental adaptation characteristics: terrain type and operation matching degree (e.g., aggressive driving is more prone to tire slippage in sandy scenarios, and the slippage rate is calculated through the difference in wheel speed sensors), and time to pass through special terrains (aggressive driving takes less time to pass through rocky areas but experiences higher bumps).

[0028] This application collects map environment information and driving style data when a vehicle first passes through an unfamiliar environment and shares it to the cloud. The vehicle can then utilize this data to implement intelligent driving assistance functions. Compared to existing technologies, this application places greater emphasis on the sharing and application of driving behavior data. The cloud server employs a distributed storage and computing architecture for efficient storage and management of large amounts of data.

[0029] Step 4: The cloud server generates simulated driving style data based on the recorded operation behavior data using a multi-scenario classifier. Then, a hidden Markov model is used to make the driving style data change with time and scenario. The multi-scene classifier integrates terrain labels (sand, mud, rock, etc.), operational features, and dynamic data, enabling three-class labeling (aggressive, balanced, and conservative) and supporting soft classification (e.g., 70% aggressive + 30% balanced). It addresses the imbalanced sample problem by simulating extreme driving style data using a Generative Adversarial Network (GAN). A Hidden Markov Model (HMM) is employed to model style changes over time and scenario. For example, the same driver might switch to an aggressive mode (rapidly climbing slopes) on sand and a conservative mode (low-speed anti-sinking) on ​​muddy roads. The cloud server maintains a style state transition matrix for each driver, recording the probability of style preference under different terrains.

[0030] Step 5: Using a deep learning-based difference detection network, the original map information in the cloud server is downweighted based on the newly constructed map and a map confidence decay model. Through adaptive adjustment, the map is updated and optimized. The adaptive adjustment is a map update strategy that adjusts based on the degree of change: small changes result in local updates; large changes result in global re-optimization. As mentioned earlier, step 5 for off-road and urban road environments also includes: quality verification and supplementary standards for the newly constructed map to handle special scenarios. This special scenario handling includes: marking vehicle roll angle safety thresholds on axle crossings, marking water depth gradients and riverbed hardness in wading areas, and marking tire pressure adjustment suggestions and passing speeds on soft ground.

[0031] Step 6: When subsequent vehicles enter the current area, the cloud server will send the adjusted map information and simulated driving style data to the subsequent vehicles. As mentioned earlier, the cloud server generates driving styles based on the driver's operational behavior data and sends them to subsequent vehicles according to the distribution strategy. The distribution strategy includes: 1. Style-based recommendation: When a subsequent vehicle requests data, the cloud automatically pushes the corresponding style data based on the current terrain (e.g., pushing one set of data each for aggressive, balanced, and conservative driving styles in a sandy environment); 2. User-selected distribution: The driver can manually select a style through the in-vehicle HMI (e.g., "imitate aggressive driving data"), and the cloud distributes corresponding style map annotations (e.g., aggressive route markers) and control parameters; 3. Hybrid mode: The system defaults to a "balanced + current driver style compensation" strategy. For example, a conservative driver can select "balanced data but with 10% more throttle response." Figure 4 As shown, step 6 includes: Step 61: After subsequent vehicles enter the current area, they send location information to the cloud server via the communication module; Step 62: Based on the location information, the cloud server uses a location matching algorithm to retrieve and find matching map and driving style data. The cloud server distributes map elements on demand: for global path planning needs, local decision-making needs, and real-time control needs.

[0032] Step 63: The cloud server sends the retrieved data to the corresponding vehicle via the communication module. Step 64: When environmental changes or new obstacles occur, an update request is triggered to correct and update the map.

[0033] Once subsequent vehicles enter the same area, they send their location information to the cloud server. Since the cloud server has already received the environmental information and driving behavior data uploaded by the vehicles that first entered each area during the preceding steps, it can now retrieve and locate the corresponding environmental information and driving behavior data for that area based on the location information of subsequent vehicles, and send it to the respective vehicles. Subsequent vehicles then use the received data to implement intelligent driving assistance functions, including automatic obstacle avoidance, path planning, and speed control.

[0034] Step 7: Subsequently, based on the map information and simulated driving style data sent by the cloud server, the vehicle uses path planning algorithms, obstacle avoidance algorithms, and speed control algorithms to realize the vehicle's intelligent driving assistance functions.

[0035] This invention provides an intelligent driving assistance method for off-road environments based on cloud sharing. This method collects map environment information and manual driving behavior data when a vehicle first passes through an unfamiliar environment, and shares it to the cloud to enable intelligent driving assistance functions for subsequent vehicles in the same area. This solves the technical problems of existing technologies, such as the lack of a data sharing mechanism, narrow data range, and inability to meet the needs of off-road driving.

[0036] In summary, this invention provides a cloud-based intelligent driving assistance system and method for off-road environments. This technical solution shares map environment information and driving behavior data collected during the initial passage through an unfamiliar environment, enabling subsequent vehicles to achieve intelligent driving assistance in the same area. This reduces safety risks for novice drivers in off-road environments, thereby improving off-road driving safety. The intelligent driving assistance function helps drivers better cope with complex off-road environments, reduces driving burden, improves driving comfort, and enhances the driving experience. As more vehicles participate in data collection and sharing, the cloud server will accumulate a large amount of off-road environment data, achieving data sharing and accumulation, further improving the performance and reliability of the intelligent driving assistance system.

[0037] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for an intelligent driving assistance system for off-road environments based on cloud sharing, characterized in that, The method includes: Step 1: When the vehicle first enters the current off-road area, collect environmental information around the vehicle in real time and select natural objects with stable geological structures as feature points. Step 2: The feature points are fused with geometry, material reflectivity and terrain slope to generate rotation-invariant feature codes, which are then matched with a pre-established seasonal feature library to extract off-road features. Maps are then constructed using these off-road features for use by intelligent driving vehicles in assisted driving. Step 3: Record the driver's operational intensity characteristics, dynamic response characteristics, and environmental adaptation characteristics during manual driving in the off-road area. The communication module then uploads the newly constructed map and the recorded operational behavior data to the cloud server. Step 4: The cloud server generates simulated driving style data based on the recorded operation behavior data using a multi-scenario classifier. Then, a hidden Markov model is used to make the driving style data change with time and scenario. Step 5: Using a deep learning-based difference detection network, the original map information in the cloud server is downweighted based on the newly constructed map and a map confidence decay model. Through adaptive adjustment, the map is updated and optimized. Step 6: When subsequent vehicles enter the current area, the cloud server will send the adjusted map information and simulated driving style data to the subsequent vehicles. Step 7: Subsequently, based on the map information and simulated driving style data sent by the cloud server, the vehicle uses path planning algorithms, obstacle avoidance algorithms, and speed control algorithms to realize the vehicle's intelligent driving assistance functions.

2. The intelligent driving assistance method for off-road environments based on cloud sharing according to claim 1, characterized in that, The intelligent driving assistance functions include automatic obstacle avoidance, path planning, and speed control.

3. The intelligent driving assistance method for off-road environments based on cloud sharing according to claim 1, characterized in that, The method for constructing the map includes: Step 21: Perform data calibration and synchronization, noise filtering, and data augmentation on the received off-road feature information; Step 22: Map construction is achieved by using LiDAR positioning, camera visual perception, and multi-sensor fusion positioning and mapping methods. Step 23: Store the constructed map using point cloud map, grid map, and topology map; Step 24: Map updates and optimizations are achieved using incremental updates, closed-loop detection and optimization, and data compression and simplification.

4. The intelligent driving assistance method for off-road environments based on cloud sharing according to claim 1, characterized in that, The map includes: traversable area division, terrain difficulty rating, recommended routes, and special risk markings.

5. The intelligent driving assistance method for off-road environments based on cloud sharing according to claim 1, characterized in that, Step 5 further includes: performing quality verification and supplementing standards on the newly constructed map for special scene processing.

6. The intelligent driving assistance method for off-road environments based on cloud sharing according to claim 5, characterized in that, The special scenario processing includes: marking the vehicle roll angle safety threshold on cross-axle road sections, marking the water depth change gradient and riverbed hardness in wading areas, and marking tire pressure adjustment suggestions and passing speed on soft ground.

7. The intelligent driving assistance method for off-road environments based on cloud sharing according to claim 1, characterized in that, Step 6 includes: Step 61: After subsequent vehicles enter the current area, they send location information to the cloud server via the communication module; Step 62: Based on the location information, the cloud server uses a location matching algorithm to retrieve and find matching map and driving style data. Step 63: The cloud server sends the retrieved data to the corresponding vehicle through the communication module.

8. The intelligent driving assistance method for off-road environments based on cloud sharing according to claim 7, characterized in that, Step 6 further includes: Step 64, when environmental changes or new obstacles occur, an update request is triggered to correct and update the map.

9. A cloud-based intelligent driving assistance system for off-road environments, implementing any one of claims 1-8, characterized in that, The system includes: The sensor input module, installed on the intelligent driving vehicle, is used to collect environmental information around the vehicle; The map building module, installed on the intelligent driving vehicle and connected to the sensor input module, is used to select geologically stable natural objects as feature points based on environmental information. It integrates the feature points with geometric shape, material reflectivity and terrain slope to generate rotation-invariant feature codes, which are then matched with a pre-established seasonal feature library to extract off-road features. The off-road features are then used to build a map for the intelligent driving vehicle to use for assisted driving. The data acquisition module, installed on the intelligent driving vehicle, is used to record the driver's operational intensity characteristics, dynamic response characteristics, and environmental adaptation characteristics during manual driving in the off-road area. The communication module, installed on the intelligent driving vehicle, is connected to the sensor input module, map building module, and data acquisition module respectively. It is connected to the cloud server via the network and is used to upload newly built maps and recorded operation behavior data to the cloud server, as well as to receive adjusted map information and simulated driving style data sent by the cloud server. The cloud server interacts with the communication module via wireless communication technology. It generates simulated driving style data based on a multi-scenario classifier and recorded operation behavior data. Then, it uses a hidden Markov model to model the driving style data so that it changes with time and scenario. A deep learning-based difference detection network is used to downweight the original map information in the cloud server and adaptively adjust it according to the newly constructed map through a map confidence decay model to achieve map update and optimization. The adjusted map information and simulated driving style data are then sent to subsequent vehicles. The intelligent driving control module, installed on the intelligent driving vehicle and connected to the communication module, is used to realize the vehicle's intelligent driving assistance functions based on map information and simulated driving style data sent by the cloud server, through path planning algorithms, obstacle avoidance algorithms, and speed control algorithms.

10. The cloud-based intelligent driving assistance system for off-road environments according to claim 9, characterized in that, The sensor input module includes lidar, camera, and millimeter-wave radar.

Images