Heavy truck optimal driving control method and system based on road condition and vehicle condition prediction

Abstract

The application provides a heavy truck optimal driving control method and system based on road condition and vehicle condition prediction, relates to the technical field of driving control, and comprises the following steps: matching the road section speed limit according to the real-time positioning of the heavy truck, and dynamically setting a road condition perception distance window; road condition scanning is performed to obtain real-time point cloud data; after loading a twin road condition map, road condition confidence fusion verification is performed, and a local traffic cost atlas is output; limited time domain multi-objective optimization is performed, and a benchmark longitudinal and lateral control spectrum is output; nonlinear disturbance compensation is performed, and an auxiliary driving control spectrum is output; the timing manipulation instructions of the driver are cooperatively supervised and dynamically weighted, until the target heavy truck reaches the boundary of the twin road condition map, and iteration driving control is triggered. The application solves the technical problem that the heavy truck driving control in the prior art mainly focuses on road condition perception and path planning, ignores the influence of vehicle conditions on driving control, and leads to inaccurate driving control and increased traffic accident risk.

Description

Technical Field

[0001] This invention relates to the field of driving control technology, specifically to an optimal driving control method and system for heavy-duty trucks based on road and vehicle condition prediction. Background Technology

[0002] Heavy trucks often face complex road conditions and changing vehicle conditions during long-distance transportation. Therefore, achieving safe, economical, and efficient autonomous driving control is of great significance for reducing traffic accidents, improving transportation efficiency, and reducing operating costs.

[0003] Most existing autonomous driving technologies focus primarily on road condition perception and path planning, with less consideration for the vehicle's own dynamic changes. However, changes in vehicle conditions affect traction, steering precision, braking response, and fuel efficiency, all of which significantly impact the safety and economy of heavy-duty trucks. Because autonomous driving systems cannot adapt to changes in vehicle conditions, vehicles may experience inaccurate handling and reduced braking performance under heavy loads, poor vehicle conditions, or severe tire wear, thereby increasing the risk of traffic accidents and energy consumption. Summary of the Invention

[0004] This application provides a method and system for optimal driving control of heavy-duty trucks based on road condition and vehicle condition prediction. It aims to solve the technical problem that existing heavy-duty truck driving control technologies mainly focus on road condition perception and path planning, ignoring the impact of vehicle condition on driving control, resulting in inaccurate driving control and increased risk of traffic accidents.

[0005] The first aspect disclosed in this application provides an optimal driving control method for heavy-duty trucks based on road condition and vehicle condition prediction. The method includes: matching road segment speed limits based on the real-time location of the heavy-duty truck and dynamically setting a road condition perception distance window in conjunction with driving response delay; driving a lidar system to scan a forward-looking driving corridor based on the road condition perception distance window to acquire real-time point cloud data; loading a twin road condition map based on the real-time location of the heavy-duty truck and performing road condition confidence fusion verification using the real-time point cloud data to output a local traffic cost map; performing finite-time-domain multi-objective optimization on the local traffic cost map using the wheel-rail contact geometric envelope of the target heavy-duty truck as a constraint boundary to output a baseline longitudinal and lateral control spectrum; interactively obtaining the real-time vehicle condition characteristics of the target heavy-duty truck and using the real-time vehicle condition characteristics to perform nonlinear disturbance compensation on the baseline longitudinal and lateral control spectrum to output an assisted driving control spectrum; and performing collaborative supervision and dynamic weight allocation of the driver's timing manipulation commands based on the assisted driving control spectrum until the target heavy-duty truck reaches the boundary of the twin road condition map, triggering iterative driving control.

[0006] The second aspect of this application discloses an optimal driving control system for heavy-duty trucks based on road condition and vehicle condition prediction. The system is used in the aforementioned optimal driving control method for heavy-duty trucks based on road condition and vehicle condition prediction. The system includes: a window setting module, used to dynamically set a road condition perception distance window based on the real-time positioning of the heavy-duty truck, matching the road segment speed limit, and combining the driving response delay; a road condition scanning module, used to drive a lidar system to scan the forward-looking driving corridor based on the road condition perception distance window, acquiring real-time point cloud data; and a fusion verification module, used to load a twin road condition map based on the real-time positioning of the heavy-duty truck, and then perform road condition confidence fusion verification using the real-time point cloud data. The system performs a multi-objective optimization (MAO) on the local traffic cost map, using the wheel-rail contact geometric envelope of the target heavy truck as the constraint boundary, and outputs a baseline longitudinal and lateral control spectrum. A disturbance compensation module is used to interactively obtain the real-time vehicle condition characteristics of the target heavy truck, and then use these characteristics to perform nonlinear disturbance compensation on the baseline longitudinal and lateral control spectrum, outputting an assisted driving control spectrum. A driving control module is used to perform collaborative supervision and dynamic weight allocation of the driver's timing manipulation commands based on the assisted driving control spectrum, until the target heavy truck reaches the boundary of the twin road condition map, triggering iterative driving control.

[0007] One or more technical solutions provided in this application have at least the following beneficial effects:

[0008] By acquiring the real-time location of the heavy truck and dynamically setting the road condition perception distance window in conjunction with the driving response delay, the system's reaction time and predictive capabilities are optimized. This allows the system to anticipate and adapt to changes in road conditions, effectively avoiding the impact of sudden obstacles or changes in traffic conditions on driving behavior, and improving driving safety and stability. Based on the set road condition perception distance window, the LiDAR is driven to perform high-precision scanning of the road ahead, acquiring real-time point cloud data, enhancing the system's real-time adaptability to complex road conditions. By combining real-time point cloud data and twin road condition maps for road condition confidence fusion verification, the accuracy of identifying local road conditions is improved. This fusion can effectively eliminate sensor errors or environmental interference, ensuring that the prediction and assessment of actual road conditions are always accurate. By using the wheel-rail contact geometric envelope of the target heavy truck as the constraint boundary... It can precisely limit the vehicle's steering range, avoiding instability or collision risks caused by oversteering or improper control in emergency situations. It performs finite-time multi-objective optimization on the local traffic cost map to generate a baseline longitudinal and lateral control spectrum, ensuring the globally optimal control scheme. After obtaining real-time vehicle condition characteristics, it performs nonlinear disturbance compensation on the baseline longitudinal and lateral control spectrum to correct control deviations caused by changes in vehicle conditions, ensuring that heavy trucks can achieve optimal driving performance under various operating conditions. It synchronizes the driver's input control commands with the assisted driving control spectrum in real time, performs conflict detection and priority determination on the driver's input. This real-time collaborative control can ensure harmonious cooperation between driver operation and automatic system commands, avoid operational conflicts, improve driving stability, and achieve adaptive and continuous driving optimization when combined with iterative driving control.

[0009] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0010] Figure 1 This is a schematic flowchart of the optimal driving control method for heavy trucks based on road condition and vehicle condition prediction provided in an embodiment of this application.

[0011] Figure 2 This is a schematic diagram of the structure of the optimal driving control system for heavy trucks based on road condition and vehicle condition prediction, provided in an embodiment of this application.

[0012] Explanation of reference numerals in the attached diagram: Window setting module 10, Road condition scanning module 20, Fusion verification module 30, Multi-target optimization module 40, Disturbance compensation module 50, Driving control module 60. Detailed Implementation

[0013] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0014] Example 1, as Figure 1 As shown in the embodiments of this application, an optimal driving control method for heavy trucks based on road condition and vehicle condition prediction is provided. The method includes:

[0015] A100: Matches road speed limits based on real-time truck location and dynamically sets road condition perception distance windows based on driving response delay.

[0016] The positioning system accurately determines the real-time location of the heavy-duty truck, including its current position, direction of travel, and speed. Based on the truck's real-time location, the system matches the specific road segment on a high-precision map. Each road segment has a set speed limit. The speed limit information for that road segment is read from the high-precision map based on the truck's current position. This speed limit reading is accomplished by combining the map database with the truck's positioning information. Driving response latency refers to the delay between the driver receiving information and actually executing the action. This latency is affected by factors such as the driver's reaction speed and the control latency of the vehicle system. Based on the speed limit and latency, the road condition perception distance window is dynamically calculated. This window determines the maximum distance at which the heavy-duty truck can effectively perceive changes in road conditions at its current speed. For example, if the speed limit is low and the driver's response latency is short, the perception window is short; conversely, it is longer. The purpose of this window setting is to ensure that the heavy-duty truck can perceive road condition information ahead within the optimal distance range.

[0017] A200: Based on the road condition perception distance window, drive the lidar system to scan the road condition of the forward driving corridor and acquire real-time point cloud data.

[0018] A lidar system is a data acquisition device that scans the surrounding environment using a laser beam and collects the returned signals. It can generate high-precision 3D point cloud data of the environment and is widely used in autonomous driving and environmental perception. The lidar system scans a defined forward driving corridor, which is a specific area in front of the vehicle. By scanning this area, the lidar acquires corresponding real-time point cloud data, representing the 3D structure and shape of the vehicle's surrounding environment, used to perceive road obstacles, road conditions, and other dynamic factors.

[0019] A300: After loading the twin road condition map based on the real-time positioning of the heavy truck, the real-time point cloud data is used to perform road condition confidence fusion verification and output a local traffic cost map.

[0020] Twin road condition maps are a dynamic representation method based on high-precision maps. They include not only static geographic information, such as road topology and traffic signs, but also dynamic, real-time updated road condition information, such as traffic flow and traffic incidents. By using real-time positioning of heavy trucks, twin road condition maps of the vicinity are loaded from a map database. Real-time point cloud data is used to fuse and verify the road condition confidence of the twin road condition maps, including confidence assessment and road condition change detection, generating a local toll cost map. This map divides the entire driving area ahead into several smaller regions, and the toll cost for each region is calculated.

[0021] A400: Using the wheel-rail contact geometry envelope of the target heavy truck as the constraint boundary, perform finite-time multi-objective optimization on the local traffic cost map and output the benchmark longitudinal and transverse control spectrum.

[0022] Determine the dynamic parameters of the heavy truck, such as wheelbase, track width, and maximum steering angle. These parameters together determine the vehicle's range of motion and driving stability. The wheel-rail contact geometric envelope is a minimum driving space obtained by modeling based on these parameters. It is a region defined by geometric constraints, representing the spatial range of contact between the heavy truck's wheels and the ground during driving.

[0023] Based on the local traffic cost map and the wheel-rail contact geometry envelope, a finite-time domain multi-objective optimization is performed. Here, "finite-time domain" means the optimization takes place within a specific timeframe, determined by the truck's current speed, road conditions, and expected driving time. Multi-objective optimization considers multiple control objectives simultaneously during the optimization process. These objectives may include obstacle avoidance costs, projected energy consumption, driving efficiency, and control feasibility, each with different weights under different conditions. The optimization algorithm selects the optimal longitudinal and lateral control strategies based on the local traffic cost map. The final output baseline longitudinal and lateral control spectra encompass all optimization objectives, providing the vehicle with initial control commands.

[0024] A500: After interactively obtaining the real-time vehicle condition characteristics of the target heavy truck, the real-time vehicle condition characteristics are used to perform nonlinear disturbance compensation on the reference longitudinal and lateral control spectrum, and output the assisted driving control spectrum.

[0025] By using onboard sensors, such as IMU and wheel speed sensors, the real-time vehicle condition characteristics of heavy trucks are obtained, including vehicle acceleration, steering angle, tire grip, throttle and brake control signals, engine status, etc., to reflect the current working status of heavy trucks and possible performance changes.

[0026] In actual driving, heavy trucks encounter nonlinear disturbances such as slippery road surfaces, sharp turns, and speed changes, all of which can cause deviations in the baseline longitudinal and lateral control spectra. Nonlinear disturbance compensation refers to dynamically adjusting the baseline longitudinal and lateral control spectra based on real-time vehicle condition characteristics to compensate for these disturbances. For example, if wheels slip, the control signal is corrected based on the current change in tire grip to ensure the vehicle can still drive safely and efficiently. Compensation is achieved through nonlinear mapping. Based on real-time vehicle condition data, a disturbance observer estimates the disturbance amount and generates compensation signals. These compensation signals are applied to the baseline longitudinal and lateral control spectra, producing a more accurate assisted driving control spectrum, further improving driving accuracy and stability.

[0027] A600: Based on the assisted driving control spectrum, perform collaborative supervision and dynamic weight allocation of the driver's timing operation commands until the target heavy truck reaches the boundary of the twin road condition map, triggering iterative driving control.

[0028] Timing-based control commands are the input signals from the driver to operate the vehicle in real time, including steering wheel operation, accelerator, and brake. The driver assistance control spectrum is used for coordinated supervision of these commands. This supervision involves monitoring the driver's inputs and the outputs of the assistance systems to ensure that there are no inconsistencies or conflicts in their operation. During this supervision process, the weights of the driver assistance control spectrum are dynamically adjusted based on driver inputs and vehicle conditions. This means that the weight distribution between the autonomous driving control system and the driver is dynamically adjusted according to the current environment and the driver's control. For example, at high speeds, more reliance is placed on automatic control, while at low speeds or in complex road conditions, driver control has a greater weight.

[0029] When the target heavy truck approaches the boundary of the twin road condition map, iterative driving control is triggered. This means that as the vehicle moves, the control system continuously evaluates and adjusts the real-time road and vehicle conditions to generate new control strategies and repeatedly optimize the control process to ensure that the vehicle always maintains the best handling state in the dynamic environment.

[0030] Furthermore, after loading the twin road condition map based on the real-time positioning of the heavy truck, the real-time point cloud data is used to perform road condition confidence fusion verification, and a local traffic cost map is output. The method includes:

[0031] A310: Perform dense 3D reconstruction based on the real-time point cloud data to dynamically reconstruct the real-time road surface elevation profile; A320: After spatially aligning and registering the real-time road surface elevation profile and the twin road condition map, perform multimodal road condition correlation difference quantification to extract multiple road condition change information; A330: Based on the multiple road condition change information, conduct a local traffic risk assessment of the twin road condition map and output the local traffic cost map.

[0032] Dense 3D reconstruction transforms discrete 3D data points into a continuous, dense 3D model. This point cloud data represents the environmental structure surrounding the vehicle, including road surfaces, obstacles, and buildings. A point cloud-based surface reconstruction algorithm fills in the gaps in the sparse point cloud data, generating a complete 3D environment. By analyzing the dense 3D reconstruction model, a real-time road surface elevation profile for the current road segment is calculated and reconstructed. This profile contains information on elevation changes in the vehicle's current driving area, such as slope undulations and road surface unevenness. Dynamic reconstruction means that this profile is continuously updated as the vehicle moves, constantly adjusting the road surface elevation information based on real-time point cloud data, allowing the control system to always grasp the current road surface condition.

[0033] Real-time road elevation profiles and twin road condition maps may differ in their representation and accuracy, necessitating spatial alignment and registration to ensure consistency in coordinate systems and scales. This allows for accurate location and assessment of road surface changes on the map. Multimodal road condition correlation differential quantification involves comparing differences between different types of data and extracting key road condition change information, including: changes in road elevation (such as slope changes, potholes, or obstacles); road surface change trends (e.g., from flat surfaces to steep slopes or severe road conditions); and environmental changes (such as the appearance of obstacles, road damage, or water accumulation).

[0034] Based on road condition change information, a local traffic risk assessment is conducted. This risk assessment refers to a quantitative analysis of the safety and traffic difficulty of the current road conditions, including calculating the traffic cost of different road segments, i.e., the traffic difficulty or danger level of that road segment. Based on the local traffic risk assessment, a local traffic cost map is generated, and the cost reflects the traffic difficulty or safety risk of that road segment.

[0035] Furthermore, after spatially aligning and registering the real-time road surface elevation profile and the twin road condition map, multimodal road condition correlation difference quantification is performed to extract multiple road condition change information. The method includes:

[0036] A321: After aligning and registering the real-time road surface elevation profile with the twin road condition map, perform multi-scale sliding window partitioning along the driving corridor to obtain multiple road cross-section analysis units; A322: Traverse the multiple road cross-section analysis units to calculate the slope change and output the slope variation vector field; A323: Traverse the multiple road cross-section analysis units to calculate the roughness change and output the roughness variation vector field; A324: Traverse the multiple road cross-section analysis units to perform new obstacle identification and output the obstacle semantic segmentation map; A325: Fuse the slope variation vector field, roughness variation vector field, and obstacle semantic segmentation map, and perform three-dimensional spatial topological identification of the road condition change area to obtain the multiple road condition change information.

[0037] Aligning and registering real-time road elevation profiles with twin road condition maps ensures consistency in coordinate systems and scales, providing accurate geolocation and road condition data for subsequent analysis. Sliding window partitioning divides the driving corridor (i.e., the area currently being traveled by the heavy truck) along the direction of travel. The window can slide at different scales to detect road condition changes at different scales. Multi-scale means the size of the sliding window can be adjusted according to specific needs, such as vehicle speed, road condition complexity, or other environmental factors. Each sliding window generates a road cross-section analysis unit, which represents a segment of the road in that area, containing information such as road geometry and pavement condition. By dividing the driving corridor into multiple analysis units, the analysis of different parts of the road can be refined, enabling more precise control and decision-making.

[0038] For each road cross-section analysis unit, the slope change is calculated. Slope describes the angle change of a road's ascent or descent. The slope change is derived by processing real-time acquired road surface elevation profile data. The calculation primarily focuses on the vertical change along the road, such as whether there are steep slopes, gentle slopes, or sudden slope changes, as these affect vehicle acceleration, braking, and stability. The calculation results are output as a slope variation vector field, representing the slope changes between different road cross-sections. Each vector corresponds to the slope change of a small segment of the road.

[0039] For each road cross-section analysis unit, the road surface roughness variation is calculated. Roughness refers to the degree of unevenness of the road surface, which directly affects vehicle driving comfort, grip, and safety. Point cloud data is used to detect uneven areas on the road surface, as well as variations in these uneven parts, such as potholes, cracks, and damage. The roughness variation calculation focuses on the differences in these unevennesses between different road sections, especially whether there are drastic changes. The calculation results are output in the form of a roughness variation vector field, which represents the roughness variation between different road sections.

[0040] Real-time point cloud data is used to identify and locate newly added obstacles, such as other vehicles, pedestrians, construction signs, and fallen objects. By performing semantic segmentation on the point cloud data of the road cross-section analysis unit, different types of objects on the road surface can be distinguished, and it can be identified which parts are obstacles. The identified obstacle information is output in the form of a semantic segmentation map, which marks obstacles in different areas of the road.

[0041] The obtained slope variation vector field, roughness variation vector field, and obstacle semantic segmentation map are fused together to form a comprehensive road condition description. Road condition change areas are marked using 3D spatial topology identification. Topology identification spatially divides the areas of road condition change in 3D space, that is, it identifies the changed areas of road segments on a map or in a 3D environment. Finally, multiple road condition change information is output, including dynamic changes in slope, roughness, and obstacles, and the location and extent of these changes are clearly defined through 3D spatial topology identification.

[0042] Furthermore, based on the multiple road condition change information, a local traffic risk assessment is performed on the twin road condition map, and the local traffic cost map is output. The method includes:

[0043] A331: Separate multiple static terrain-related variables from the multiple road condition change information, wherein the static terrain-related variables include elevation change amplitude, slope change intensity, and road surface roughness vector; A332: Quantify and output rollover risk index and skid risk index based on the real-time vehicle condition characteristics; A333: After matching the real-time driving scenario based on the real-time positioning of the heavy truck, call the real-time weight allocation according to the real-time driving scenario; A334: Using the real-time weight allocation as the weighted quantization benchmark, combine the rollover risk index and skid risk index to perform weighted fusion calculation of the multiple static terrain-related variables, and output multiple discrete risk levels; A335: After performing nonlinear mapping of the multiple discrete risk levels to the traffic cost value, project them onto the two-dimensional raster coordinate system of the twin road condition map to generate the local traffic cost map.

[0044] Multiple static terrain-related variables are extracted from various road condition change data. These variables are static indicators describing the basic characteristics of the road and typically do not change over time. Elevation change amplitude refers to the magnitude of elevation change at different locations on the road surface. For example, sudden slopes or undulations can affect vehicle control. Slope variation intensity describes the intensity of changes in road slope, especially in transitional sections between steep and gentle slopes. Areas with significant slope changes require more driver control to avoid unnecessary risks. The road surface roughness vector represents the distribution of road surface unevenness. Areas with higher roughness increase vehicle vibration and may affect grip.

[0045] The vehicle's current stability and control are assessed based on real-time vehicle condition characteristics. The rollover risk index measures the likelihood of the vehicle rolling over under the current driving conditions. It depends on factors such as vehicle speed, road slope, vehicle steering angle, and dynamic center of gravity. The skid risk index reflects the risk of the vehicle skidding under the current road conditions. It is affected by factors such as road surface roughness, wetness, and tire grip.

[0046] The system utilizes real-time positioning of heavy trucks to match the current real-time driving scenario, including the specific road segment where the vehicle is located, the current road conditions, and the surrounding environment. By matching the real-time driving scenario, the system identifies the characteristics of the current road segment, such as whether there are sharp turns, steep slopes, or obstacles. Based on these characteristics, the system calls for real-time weight allocation, that is, it allocates weights according to the real-time driving scenario and road conditions.

[0047] Using real-time weight allocation as a benchmark, and combining rollover risk and skidding risk indices, a weighted fusion calculation is performed on static terrain-related variables. This means adjusting the weight of each static terrain variable based on the risk level of the current road and vehicle conditions. For example, if the rollover risk is high, a higher weight is given to the slope variation intensity, thus prioritizing the impact of slopes. After weighted fusion, the system outputs multiple discrete risk levels, each representing the safety or ease of passage of the current road segment after considering factors such as vehicle conditions, road conditions, and risk indices.

[0048] A nonlinear mapping is applied to multiple discrete risk levels, transforming these levels into continuous passage cost values ​​using nonlinear functions such as exponential or logarithmic functions. The purpose of this nonlinear mapping is to ensure that the impact of different risk levels on the passage cost map is presented non-linearly, resulting in higher-risk areas having higher cost values. These passage cost values ​​are projected onto the two-dimensional raster coordinate system of the twin road condition map to generate an intuitive local passage cost map. This map displays the passage difficulty of each small area on the road, helping the system and drivers identify optimal routes or potential hazards.

[0049] Furthermore, using the wheel-rail contact geometric envelope of the target heavy truck as the constraint boundary, finite-time multi-objective optimization is performed on the local traffic cost map to output a baseline longitudinal and lateral control spectrum. The method includes:

[0050] A410: Calculate the envelope boundary based on the target heavy truck's wheelbase, track width, and maximum steering angle dynamic parameters to obtain the wheel-rail contact geometric envelope; A420: Using the driving direction vector and driving time domain as two-dimensional constraints, perform continuous trajectory fitting on the twin road condition map using the wheel-rail contact geometric envelope to obtain multiple candidate trajectories; A430: Project the multiple candidate trajectories onto the local traffic cost map and perform control sequence fitting to obtain multiple candidate longitudinal and lateral control spectra; A440: Traverse the multiple candidate longitudinal and lateral control spectra to extract multiple sets of driving control associated targets; A450: Weight quantize the multiple sets of driving control associated targets to obtain multiple scalarized performance values; A460: Based on the ascending order of the multiple scalarized performance values, locate the baseline longitudinal and lateral control spectrum among the multiple candidate longitudinal and lateral control spectra.

[0051] Wheelbase refers to the distance between the front and rear axles of a vehicle, track width refers to the distance between the left and right tires, and maximum steering angle represents the maximum angle the steering wheel can turn, directly affecting the vehicle's turning radius and trajectory. The wheel-rail contact geometric envelope describes the area of ​​contact between the front and rear tires and the ground. This envelope considers the range of wheel motion, steering angle, relative position between wheels during turning, and the tire contact state during vehicle operation. By calculating the envelope boundaries, the geometric constraints of the vehicle during travel can be clearly defined.

[0052] The driving direction vector refers to the vehicle's current driving direction, which can be determined using information such as the vehicle's real-time positioning and heading angle. The driving time domain refers to the range of the vehicle's trajectory within a specific time period, reflecting the vehicle's motion state over a future period. By using the driving direction vector and the driving time domain as dual constraints, the vehicle's trajectory is limited. These two parameters ensure that the fitted trajectory not only conforms to the current driving direction but also meets the vehicle's motion characteristics and dynamic constraints over a future period.

[0053] Under the constraints of the wheel-rail contact geometric envelope, continuous trajectory fitting is performed. This means that multiple driving trajectories are calculated based on existing road condition data and dynamic constraints. These trajectories conform to the vehicle's steering, speed, and dynamic characteristics and do not exceed the wheel-rail contact geometric envelope. Continuous trajectory fitting implies that the resulting trajectory is smooth, without abrupt changes or discontinuities, ensuring vehicle stability and safety. After fitting, multiple candidate trajectories are generated.

[0054] Projecting multiple alternative routes onto a local traffic cost map means evaluating the traffic cost of each alternative route based on road condition information. Traffic cost is a comprehensive indicator that measures the difficulty, risk, and time consumption of a route under specific road conditions. By projecting onto a local traffic cost map, the cost of each route can be clearly identified, helping to determine the optimal route.

[0055] Control sequence fitting involves generating corresponding control command sequences based on trajectory changes and costs. These sequences indicate how to guide the vehicle along the fitted trajectory through longitudinal control (e.g., acceleration, deceleration) and lateral control (e.g., steering). Control sequence fitting yields multiple alternative longitudinal and lateral control spectra, providing different control schemes for each alternative trajectory.

[0056] For each alternative longitudinal and lateral control spectrum, control objectives related to the driving task are extracted, namely driving control associated objectives. These objectives are designed based on the vehicle's driving needs and environmental factors, and are used to measure the merits of each trajectory control scheme. Specific driving control associated objectives include obstacle avoidance cost, expected energy consumption, driving efficiency, and control feasibility.

[0057] Each control objective has a different level of importance. Different weights are assigned to different control objectives based on real-time driving needs and road conditions. Through weighting, each control objective can be quantified according to its priority, generating a comprehensive scalar performance value that represents the performance of that control objective under the current circumstances. For example, control objectives with lower obstacle avoidance costs receive higher performance values, while those with lower control feasibility receive lower performance values.

[0058] The candidate longitudinal and transverse control spectra are sorted in ascending order based on multiple scalar performance values. That is, control spectra with lower performance values ​​are ranked higher, indicating that these spectra perform better overall across multiple control objectives and meet higher optimization criteria. After ascending sorting, the control spectrum with the highest ranking is selected as the baseline longitudinal and transverse control spectrum, i.e., the optimal control scheme.

[0059] Furthermore, each set of driving control-related objectives includes obstacle avoidance cost, expected energy consumption, driving efficiency, and control feasibility.

[0060] The core concern of obstacle avoidance cost is ensuring that the vehicle avoids collisions with obstacles during operation, especially in complex road environments. If the control scheme causes the vehicle to approach obstacles, the obstacle avoidance cost increases, thus affecting the overall evaluation of the control spectrum. Vehicle energy consumption is mainly affected by the choice of driving path, vehicle acceleration and braking operations, and road conditions. When selecting a control spectrum, priority should be given to trajectories that optimize expected energy efficiency, avoiding frequent rapid acceleration and braking to reduce fuel or battery consumption. Driving efficiency assesses the vehicle's speed and time efficiency. Selecting a control spectrum with high driving efficiency means that the vehicle can reach its destination with minimal time cost, reducing unnecessary stops and decelerations. Driving efficiency is also closely related to road conditions, traffic flow, and other factors. The core of control feasibility is ensuring that the selected trajectory and control actions are physically feasible and do not exceed vehicle dynamics limitations, such as maximum steering angle, maximum acceleration, and maximum braking capacity.

[0061] Furthermore, after interactively obtaining the real-time vehicle condition characteristics of the target heavy truck, the real-time vehicle condition characteristics are used to perform nonlinear disturbance compensation on the reference longitudinal and lateral control spectra, and the assisted driving control spectrum is output. The method includes:

[0062] A510: Identify disturbance sources for the real-time vehicle condition features to obtain multiple real-time disturbance sources; A520: Input the multiple real-time disturbance sources into a nonlinear disturbance observer to perform nonlinear mapping to obtain multiple disturbance quantity estimation vectors; A530: Calculate multiple nonlinear feedforward compensation quantities based on the multiple disturbance quantity estimation vectors; A540: After superimposing the multiple nonlinear feedforward compensation quantities onto the reference longitudinal and lateral control spectra, perform spectrum adaptive smoothing and shaping to output the assisted driving control spectrum.

[0063] During vehicle operation, various disturbances may occur, such as changes in road conditions, driver operation, environmental influences, and changes in the vehicle's own state. These disturbances can affect the vehicle's dynamic behavior and control response. They need to be identified and categorized into multiple real-time disturbances for subsequent compensation and optimization.

[0064] A nonlinear disturbance observer is a mathematical tool used to estimate and observe disturbances in a system that are not directly measured. Unlike traditional linear observers, nonlinear disturbance observers can handle more complex and nonlinear disturbance sources closely related to the nonlinear dynamics of a vehicle, such as tire friction and suspension characteristics. Based on the vehicle's input and output signals, such as acceleration, steering angle, and braking force, as well as vehicle condition characteristics, the nonlinear disturbance observer estimates the disturbance source. It can accurately identify different types of disturbances and convert them into numerical disturbance quantities. The resulting multiple disturbance quantity estimates can be seen as the representation of the disturbance source in the vehicle control model; for example, deviations in speed and direction caused by sudden changes in road conditions, wind, or gradient.

[0065] Feedforward compensation refers to making corresponding control adjustments in advance based on known disturbances in a control system, rather than making feedback adjustments afterward. Nonlinear feedforward compensation is a control method used to handle disturbances in nonlinear systems. It optimizes the system response by predicting the impact of disturbances in advance and taking corresponding compensation measures. In this step, based on multiple disturbance estimate vectors, the compensation amount for each disturbance is calculated. The compensation amount refers to the control quantity that needs to be adjusted to counteract the influence of the disturbance source, such as adjusting acceleration, braking force, or steering angle.

[0066] Multiple calculated nonlinear feedforward compensation values ​​are superimposed onto the baseline longitudinal and lateral control spectra. This superposition process is equivalent to applying adjustment signals to the original control scheme in response to disturbances, thereby optimizing vehicle driving control. Adaptive spectrum smoothing and shaping is a signal processing method used to smooth control signals, avoiding abrupt changes or drastic fluctuations in control commands. This ensures that the compensated control signal is not too drastic, preventing over-adjustment of the vehicle and resulting in an uncomfortable or unsafe driving experience. The compensated and adjusted assisted driving control spectrum is used in actual driving for the assisted driving system to control the vehicle in real time.

[0067] Furthermore, based on the assisted driving control spectrum, the method includes collaborative supervision and dynamic weight allocation of the driver's timing control commands until the target heavy truck reaches the boundary of the twin road condition map, triggering iterative driving control.

[0068] A610: After synchronizing the timing control commands and assisted driving control spectrum output by the driver, conflict detection is performed to obtain multiple real-time control conflicts; A620: Based on the multiple control priorities of the multiple real-time control conflicts, real-time coupling arbitration of driving control is performed to generate an arbitration preference value; A630: After dynamically adjusting the bias of the assisted driving control spectrum according to the arbitration preference value, the assisted driving control spectrum is executed in a closed loop until the target heavy truck reaches the boundary of the twin road condition map, triggering iterative driving control.

[0069] Synchronizing the driver's timing commands with the driver assistance control spectrum means mapping them to the same timeline, facilitating comparison and analysis of the differences between driver actions and system commands at each point in time. Conflict detection is then performed on the synchronized signals to identify inconsistencies between driver actions and the driver assistance control spectrum. For example, if the driver turns the steering wheel left while the driver assistance spectrum suggests turning right, the detection results in multiple real-time control conflicts, each describing a contradiction between the driver's intention and the system control.

[0070] Control priorities are assigned based on different types of real-time control conflicts. These priorities are dynamically adjusted according to driving scenarios, vehicle speed, road conditions, and real-time risk assessments. Driver intentions are coupled with control commands from the assisted driving spectrum in real time, and arbitration decisions are made according to priority. The arbitration process weights each conflict: if the conflict involves safety, assisted driving control is prioritized; if the conflict has a minor impact on safety, driver operation is retained. The arbitration result is quantified as an arbitration preference value, reflecting the priority of each control command in the current scenario.

[0071] Based on the arbitration preference value, the assisted driving control spectrum is dynamically adjusted: when the driver's operation is safe and feasible, the original control spectrum is appropriately deviated; when the driver's operation may pose a risk, the system control is maintained or enhanced. Dynamic biasing allows the assisted driving control spectrum to adapt to real-time driving needs, balancing the flexibility and safety of human-machine operation. The adjusted assisted driving control spectrum is sent to the vehicle controller in real time, forming a closed-loop control. When the vehicle reaches the boundary of the twin road condition map, the next iteration is triggered, starting a new round of human-machine cooperative driving optimization.

[0072] Furthermore, the method includes matching road speed limits based on the real-time location of the heavy truck and dynamically setting the road condition perception distance window in conjunction with the driving response delay.

[0073] A110: Obtain the real-time location of the target heavy truck through a tightly coupled GNSS-IMU navigation system; A120: Based on the real-time location of the heavy truck, match the real-time lane and real-time road segment on a high-precision map, read the speed limit attribute, and obtain the speed limit of the road segment; A130: Calculate the reference perception distance based on the speed limit of the road segment and the driving response delay, and set it as the road condition perception distance window.

[0074] GNSS (Global Navigation Satellite System) and IMU (Inertial Measurement Unit) are technologies used for high-precision positioning. GNSS provides geographic coordinates, including longitude, latitude, and altitude, and can acquire vehicle position information over a large area, but it is sensitive to signal blockage. IMU provides vehicle attitude information, such as acceleration, angular velocity, and steering angle, through sensors like accelerometers, gyroscopes, and magnetometers, and can provide supplementary positioning information when GNSS signals are lost or interfered with. Combining these two technologies provides more stable and accurate positioning information, especially in environments with poor signal coverage, maintaining high positioning accuracy.

[0075] High-precision maps contain detailed road information, such as geographical and traffic attributes like lanes, road segments, intersections, and ramps, as well as dynamic road information, such as speed limit signs, road obstacles, and traffic markings. Based on real-time positioning information, the current location of the target heavy truck is mapped to the real-time lanes and road segments on the high-precision map. This means that it is possible to clearly identify which lane the heavy truck is currently in and which road segment it is traveling on, and to identify specific traffic restrictions and road condition information for that road segment based on map data. Each road segment has corresponding speed limit information in the high-precision map, which comes from traffic monitoring systems or road signs. Based on the matched real-time road segment, the speed limit information for that road segment is read from the high-precision map. The road segment speed limit includes both the maximum and minimum speed limits.

[0076] Driving response latency refers to the delay between the driver or autonomous driving system issuing a control command and the vehicle actually responding. This delay includes the processing time of the control signal, the response time of the transmission system, and the dynamic response time of the vehicle. For heavy trucks, the driving response latency is longer because their mass is greater, resulting in relatively slower steering and acceleration / deceleration responses. Driving response latency is calculated through actual measurements or empirical data.

[0077] Based on the current speed limit and driving response delay of the road segment, a baseline perception distance is calculated. This distance represents the minimum distance that the heavy truck needs to perceive before reacting at the current speed limit. The baseline perception distance equals the speed limit multiplied by the driving response delay. Setting the baseline perception distance as the road condition perception distance window aims to rationally set the vehicle's perception distance based on the real-time speed limit and driving response delay, providing the driver and control system with sufficient time to react and make corresponding control decisions, thereby improving driving safety and stability.

[0078] Example 2 is based on the same inventive concept as the heavy-duty truck optimal driving control method based on road and vehicle condition prediction in the previous examples, such as... Figure 2 As shown in the figure, this application provides an optimal driving control system for heavy trucks based on road condition and vehicle condition prediction. The system includes:

[0079] The window setting module 10 is used to match the road segment speed limit based on the real-time positioning of the heavy truck and dynamically set the road condition perception distance window in combination with the driving response delay; the road condition scanning module 20 is used to drive the lidar system to scan the forward driving corridor based on the road condition perception distance window and obtain real-time point cloud data; the fusion verification module 30 is used to load the twin road condition map based on the real-time positioning of the heavy truck, and then use the real-time point cloud data to perform road condition confidence fusion verification and output a local traffic cost map; the multi-objective optimization module 40 is used to optimize the target heavy truck's wheel-rail contact ratio... Using the envelope as the constraint boundary, finite-time multi-objective optimization is performed on the local traffic cost map to output a baseline longitudinal and lateral control spectrum; the disturbance compensation module 50 is used to interactively obtain the real-time vehicle condition characteristics of the target heavy truck, and then use the real-time vehicle condition characteristics to perform nonlinear disturbance compensation on the baseline longitudinal and lateral control spectrum to output an assisted driving control spectrum; the driving control module 60 is used to perform collaborative supervision and dynamic weight allocation of the driver's timing operation commands based on the assisted driving control spectrum until the target heavy truck reaches the boundary of the twin road condition map, triggering iterative driving control.

[0080] Furthermore, the fusion verification module 30 is used to perform the following operation steps:

[0081] Based on the real-time point cloud data, dense 3D reconstruction is performed to dynamically reconstruct the real-time road surface elevation profile; after spatial alignment and registration of the real-time road surface elevation profile and the twin road condition map, multimodal road condition correlation difference quantification is performed to extract multiple road condition change information; based on the multiple road condition change information, local traffic risk assessment of the twin road condition map is performed, and the local traffic cost map is output.

[0082] Furthermore, the fusion verification module 30 is used to perform the following operation steps:

[0083] After aligning and registering the real-time road surface elevation profile with the twin road condition map, a multi-scale sliding window partition is performed along the driving corridor to obtain multiple road cross-section analysis units. The slope variation vector field is calculated and output by traversing the multiple road cross-section analysis units. The roughness variation vector field is calculated and output by traversing the multiple road cross-section analysis units. New obstacle identification is performed and an obstacle semantic segmentation map is output. The slope variation vector field, roughness variation vector field, and obstacle semantic segmentation map are fused, and a three-dimensional spatial topological labeling of the road condition change area is performed to obtain the multiple road condition change information.

[0084] Furthermore, the fusion verification module 30 is used to perform the following operation steps:

[0085] Multiple static terrain-related variables are separated from the multiple road condition change information, including elevation change magnitude, slope variation intensity, and road surface roughness vector. A rollover risk index and a skid risk index are quantized and output based on the real-time vehicle condition characteristics. After matching the real-time driving scenario with the real-time location of the heavy truck, a real-time weight allocation is invoked based on the real-time driving scenario. Using the real-time weight allocation as a weighted quantization benchmark, and combining the rollover risk index and the skid risk index, a weighted fusion calculation of the multiple static terrain-related variables is performed to output multiple discrete risk levels. After performing a nonlinear mapping of the multiple discrete risk levels to the traffic cost value, the data is projected onto the two-dimensional raster coordinate system of the twin road condition map to generate the local traffic cost map.

[0086] Furthermore, the multi-objective optimization module 40 is used to perform the following operation steps:

[0087] Based on the target heavy truck's wheelbase, track width, and maximum steering angle dynamic parameters, the envelope boundary is calculated to obtain the wheel-rail contact geometric envelope. Using the driving direction vector and driving time domain as two-dimensional constraints, continuous trajectory fitting is performed on the twin road condition map using the wheel-rail contact geometric envelope to obtain multiple candidate trajectories. These candidate trajectories are projected onto the local traffic cost map, and control sequence fitting is performed to obtain multiple candidate longitudinal and lateral control spectra. Multiple sets of driving control-related targets are extracted by traversing these multiple candidate longitudinal and lateral control spectra. These multiple sets of driving control-related targets are weighted and quantized to obtain multiple scalarized performance values. Based on the ascending order of these multiple scalarized performance values, the baseline longitudinal and lateral control spectrum is located within the multiple candidate longitudinal and lateral control spectra.

[0088] Furthermore, each set of driving control-related objectives includes obstacle avoidance cost, expected energy consumption, driving efficiency, and control feasibility.

[0089] Furthermore, the disturbance compensation module 50 is used to perform the following operation steps:

[0090] The real-time vehicle condition features are subjected to disturbance source identification to obtain multiple real-time disturbance sources; the multiple real-time disturbance sources are input into a nonlinear disturbance observer to perform nonlinear mapping to obtain multiple disturbance quantity estimation vectors; multiple nonlinear feedforward compensation quantities are calculated based on the multiple disturbance quantity estimation vectors; after superimposing the multiple nonlinear feedforward compensation quantities onto the reference longitudinal and lateral control spectra, spectrum adaptive smoothing and shaping are performed to output the assisted driving control spectrum.

[0091] Furthermore, the driving control module 60 is used to perform the following operation steps:

[0092] After synchronizing the timing control commands and assisted driving control spectrum output by the driver, conflict detection is performed to obtain multiple real-time control conflicts. Based on the multiple control priorities of the multiple real-time control conflicts, real-time coupling arbitration of driving control is performed to generate an arbitration preference value. After dynamically adjusting the bias of the assisted driving control spectrum according to the arbitration preference value, the assisted driving control spectrum is executed in a closed loop until the target heavy truck reaches the boundary of the twin road condition map, triggering iterative driving control.

[0093] Furthermore, the window setting module 10 is used to perform the following operation steps:

[0094] The real-time location of the target heavy truck is obtained through a tightly coupled GNSS-IMU integrated navigation system; based on the real-time location of the heavy truck, real-time lanes and road segments are matched on a high-precision map, and speed limit attributes are read to obtain the speed limit of the road segment; based on the speed limit of the road segment and the driving response delay, a reference sensing distance is calculated and set as the road condition sensing distance window.

[0095] Through the foregoing detailed description of the optimal driving control method for heavy-duty trucks based on road condition and vehicle condition prediction, those skilled in the art can clearly understand the optimal driving control system for heavy-duty trucks based on road condition and vehicle condition prediction in this embodiment. Since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and relevant parts can be referred to in the method section.

[0096] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A method for optimal driving control of heavy-duty trucks based on road condition and vehicle condition prediction, characterized in that, The method includes: The road section speed limit is matched based on the real-time location of the heavy truck, and the road condition perception distance window is dynamically set in combination with the driving response delay. Based on the road condition perception distance window, the lidar system is driven to scan the road conditions of the forward driving corridor and acquire real-time point cloud data; After loading the twin road condition map based on the real-time positioning of the heavy truck, the real-time point cloud data is used to perform road condition confidence fusion verification and output a local traffic cost map. Using the wheel-rail contact geometric envelope of the target heavy truck as the constraint boundary, finite-time multi-objective optimization is performed on the local traffic cost map to output the benchmark longitudinal and lateral control spectrum; After interactively obtaining the real-time vehicle condition characteristics of the target heavy truck, the real-time vehicle condition characteristics are used to perform nonlinear disturbance compensation on the reference longitudinal and lateral control spectrum, and the assisted driving control spectrum is output. Based on the assisted driving control spectrum, the driver's timing operation commands are coordinated and dynamically weighted until the target heavy truck reaches the boundary of the twin road condition map, triggering iterative driving control; Using the wheel-rail contact geometry envelope of the target heavy truck as the constraint boundary, finite-time multi-objective optimization is performed on the local traffic cost map to output a baseline longitudinal and lateral control spectrum. The method includes: The wheel-rail contact geometric envelope is obtained by calculating the wheelbase, track width, and maximum steering angle dynamic parameters of the target heavy truck. Using the driving direction vector and driving time domain as two-dimensional constraints, the wheel-rail contact geometric envelope is used to perform continuous trajectory fitting on the twin road condition map to obtain multiple candidate trajectories; The multiple candidate trajectories are projected onto the local traffic cost map, and control sequence fitting is performed to obtain multiple candidate longitudinal and transverse control spectra. Multiple sets of driving control-related targets are extracted by traversing the multiple candidate longitudinal and lateral control spectra. The weighted quantization of the multiple sets of driving control-related objectives yields multiple scalarized performance values; Based on the ascending order of the multiple scalar performance values, the baseline longitudinal and transverse control spectrum is located among the multiple candidate longitudinal and transverse control spectra.

2. The optimal driving control method for heavy trucks based on road condition and vehicle condition prediction as described in claim 1, characterized in that, After loading the twin road condition map based on the real-time positioning of the heavy truck, the road condition confidence fusion verification is performed using the real-time point cloud data, and a local traffic cost map is output. The method includes: Dense 3D reconstruction is performed based on the real-time point cloud data to dynamically reconstruct the real-time road surface elevation profile. After aligning and registering the real-time road elevation profile and the twin road condition map, multimodal road condition correlation difference quantification is performed to extract multiple road condition change information. Based on the multiple road condition change information, a local traffic risk assessment is performed on the twin road condition map, and the local traffic cost map is output.

3. The optimal driving control method for heavy trucks based on road condition and vehicle condition prediction as described in claim 2, characterized in that, After spatially aligning and registering the real-time road elevation profile and the twin road condition map, multimodal road condition correlation difference quantification is performed to extract multiple road condition change information. The method includes: After aligning and registering the real-time road elevation profile with the twin road condition map, multi-scale sliding window partitioning is performed along the driving corridor to obtain multiple road cross-section analysis units. The slope variation vector field is output by traversing the multiple road cross-section analysis units. The roughness variation is calculated and output as a roughness variation vector field by traversing the multiple road cross-section analysis units. The multiple road cross-section analysis units are traversed to perform new obstacle identification and output an obstacle semantic segmentation map; By fusing the slope variation vector field, roughness variation vector field, and obstacle semantic segmentation map, three-dimensional spatial topological identification of the road condition change area is performed to obtain the multiple road condition change information.

4. The optimal driving control method for heavy trucks based on road condition and vehicle condition prediction as described in claim 2, characterized in that, Based on the multiple road condition change information, a local traffic risk assessment is performed on the twin road condition map, and the local traffic cost map is output. The method includes: Multiple static terrain-related variables are separated from the multiple road condition change information, wherein the static terrain-related variables include elevation change amplitude, slope change intensity and road surface roughness vector; Based on the real-time vehicle condition characteristics, the rollover risk index and skidding risk index are quantitatively output. After matching the real-time driving scenario with the real-time location of the heavy truck, the real-time weight allocation is invoked according to the real-time driving scenario; Using the real-time weight allocation as the weighted quantization benchmark, and combining the rollover risk index and the skid risk index, the weighted fusion calculation of the multiple static terrain-related variables is performed to output multiple discrete risk levels; After performing a nonlinear mapping of the multiple discrete risk levels to the traffic cost values, the data is projected onto the two-dimensional raster coordinate system of the twin road condition map to generate the local traffic cost map.

5. The optimal driving control method for heavy trucks based on road condition and vehicle condition prediction as described in claim 1, characterized in that, Each set of driving control-related objectives includes obstacle avoidance cost, estimated energy consumption, driving efficiency, and control feasibility.

6. The optimal driving control method for heavy trucks based on road condition and vehicle condition prediction as described in claim 1, characterized in that, After interactively obtaining the real-time vehicle condition characteristics of the target heavy truck, the real-time vehicle condition characteristics are used to perform nonlinear disturbance compensation on the reference longitudinal and lateral control spectra, and the assisted driving control spectrum is output. The method includes: Disturbance sources are identified from the real-time vehicle condition features to obtain multiple real-time disturbance sources; The multiple real-time disturbance sources are input into a nonlinear disturbance observer to perform nonlinear mapping, resulting in multiple disturbance quantity estimation vectors; Based on the multiple disturbance estimate vectors, multiple nonlinear feedforward compensation quantities are calculated; After superimposing the multiple nonlinear feedforward compensation quantities onto the reference longitudinal and lateral control spectra, adaptive spectrum smoothing and shaping are performed to output the assisted driving control spectrum.

7. The optimal driving control method for heavy trucks based on road condition and vehicle condition prediction as described in claim 1, characterized in that, Based on the assisted driving control spectrum, the method performs collaborative supervision and dynamic weight allocation of the driver's timing control commands until the target heavy truck reaches the boundary of the twin road condition map, triggering iterative driving control. The method includes: After synchronizing the timing control commands and assisted driving control spectrum output by the driver, conflict detection is performed to obtain multiple real-time control conflicts; Based on the multiple control priorities of the aforementioned multiple real-time control conflicts, real-time coupling arbitration of driving control is performed to generate an arbitration preference value; After dynamically adjusting the assisted driving control spectrum based on the arbitration preference value, the assisted driving control spectrum is executed in a closed loop until the target heavy truck reaches the boundary of the twin road condition map, triggering iterative driving control.

8. The optimal driving control method for heavy trucks based on road condition and vehicle condition prediction as described in claim 1, characterized in that, The method includes matching road speed limits based on real-time truck location and dynamically setting the road condition perception distance window based on driving response delay. The real-time positioning of the target heavy truck is obtained through a tightly coupled GNSS-IMU integrated navigation system; Based on the real-time positioning of the heavy truck, the real-time lane and real-time road segment are matched on a high-precision map, and the speed limit attribute is read to obtain the speed limit of the road segment. Based on the speed limit and driving response delay of the road segment, a baseline sensing distance is calculated and set as the road condition sensing distance window.

9. A heavy-duty truck optimal driving control system based on road condition and vehicle condition prediction, characterized in that, The system is used to implement the optimal driving control method for heavy trucks based on road condition and vehicle condition prediction as described in any one of claims 1-8, the system comprising: The window setting module is used to match the road section speed limit based on the real-time location of the heavy truck, and dynamically set the road condition perception distance window in combination with the driving response delay. The road condition scanning module is used to drive the lidar system to scan the road condition of the forward driving corridor based on the road condition perception distance window and obtain real-time point cloud data. The fusion verification module is used to load the twin road condition map based on the real-time positioning of the heavy truck, and then use the real-time point cloud data to perform road condition confidence fusion verification and output a local traffic cost map. The multi-objective optimization module is used to perform finite-time multi-objective optimization on the local traffic cost map with the wheel-rail contact geometry envelope of the target heavy truck as the constraint boundary, and output the benchmark longitudinal and transverse control spectrum. The disturbance compensation module is used to interactively obtain the real-time vehicle condition characteristics of the target heavy truck, and then use the real-time vehicle condition characteristics to perform nonlinear disturbance compensation on the reference longitudinal and lateral control spectrum, and output the assisted driving control spectrum. The driving control module is used to coordinate and dynamically allocate the driver's timing operation commands based on the assisted driving control spectrum until the target heavy truck reaches the boundary of the twin road condition map, triggering iterative driving control.

Images