High-precision beidou positioning driven adaptive multi-target path planning method

Abstract

The application discloses a high-precision Beidou positioning driving adaptive multi-target path planning method, relates to the technical field of navigation path planning, and comprises the following steps: step 1, a positioning quality field of a front road section is constructed, containing an error upper bound, a convergence residual time, an interruption risk, a correction effective residual period and a pre-charge success probability, and a continuity budget and a forbidden cutting window are set; step 2, a path and a channel are jointly solved on a road section-service channel grid, under the constraints of a budget and an opportunity, a service channel plan and pre-charge and entry time are generated; step 3, adaptive hysteresis switching and double-buffer pre-charge are implemented according to road semantics and vehicle states, and application time alignment and smooth transition are completed; and step 4, in rolling time domain control, the error upper bound and the budget are compiled into constraints and are hierarchically degraded. The method improves path positioning continuity, reduces switching frequency, avoids key road sections, reduces guidance mutation, and guarantees stable traffic under complex urban environments.

Description

Technical Field

[0001] This invention relates to the field of navigation path planning technology, specifically to an adaptive multi-objective path planning method driven by high-precision BeiDou positioning. Background Technology

[0002] In complex environments such as urban roads, highway interchanges, tunnel entrances and exits, dense high-rise buildings, and tree-lined areas, mobile vehicles such as road vehicles, unmanned delivery vehicles, park transport vehicles, and low-altitude drones are increasingly relying on high-precision positioning and path planning based on BeiDou for coordinated operation.

[0003] Existing engineering systems typically switch between satellite-based precise point positioning broadcasting, real-time carrier phase positioning or precise point positioning provided by ground-based continuous reference stations, satellite-based augmentation, and the fusion of inertial navigation, vision, and lidar. At the upper layer, maps and traffic information are used to generate paths and trajectories. However, satellite-based precise point positioning suffers from convergence on the order of minutes under dynamic conditions, and urban canyons can cause intermittent interruptions and jumps in correction messages due to obstruction of the geosynchronous direction. Ground-based augmentation is affected by communication latency and packet loss, and the solution type repeatedly switches between fixed, floating-point, and non-converged states.

[0004] The aforementioned instabilities are propagated to the path and control layers through the fusion link, triggering a chain reaction of consequences such as sudden location changes, passive triggering of speed and curvature constraints, and missed or mis-entered critical ramps. Existing solutions mostly determine service handover based on a single threshold or immediate utility comparison, and have not yet explicitly introduced quantitative indicators such as remaining convergence time, correction validity period, interruption probability, and handover cost at the path layer. They also lack handover prohibition constraints and handover schedule management for ramps, merging, and sharp curves. At the same time, the applicable time and reference frame of correction messages are not unified at the moment of handover, resulting in the inability to smoothly absorb de-jumping changes. Furthermore, existing route planning lacks continuous budgeting and upper quantile risk constraints across different routes, making it impossible to guarantee stable and high accuracy before critical travel nodes arrive. There are discrepancies in satellite-to-ground service coverage, low-orbit augmentation is still in the pilot stage, reference station network density is insufficient in some areas, and cellular networks experience delays or increased jitter under high load. The cloud and vehicle-side modules are disconnected, lacking a unified approach that incorporates positioning quality as an evaluation indicator into route planning and scheduling. This results in a decrease in the effective high-precision mileage ratio, increased manual takeover and safety intervention, greater fluctuations in arrival time, and increased energy consumption and maintenance costs.

[0005] Therefore, the current technical problem is that, under the conditions of multiple high-precision positioning services coexisting and the environment being significantly non-stationary, existing systems lack a computable description of positioning quality for the road segment ahead. They cannot simultaneously quantify the remaining convergence time, the effective period for correction, the probability of interruption and the cost of switching, and form consistent constraints with path selection, prohibition of switching on key road segments and switching schedules. As a result, it is difficult to obtain and maintain stable high precision before key nodes, and to resolve and control the trajectory of continuous disturbances caused by jumps. Summary of the Invention

[0006] (a) Technical problems to be solved To address the shortcomings of existing technologies, this invention provides an adaptive multi-objective path planning method driven by high-precision BeiDou positioning. By jointly solving for paths and channels on a road segment-service channel grid, and subject to budget and opportunity constraints, it generates service channel plans, pre-charge, and entry times. Based on road semantics and vehicle status, it implements adaptive hysteresis handover and double-buffer pre-charge, achieving applicable time alignment and smooth transition. In rolling time-domain control, the upper bound of error and budget are compiled into constraints and graded for degradation. This method improves path positioning continuity, reduces handover frequency, avoids critical road segments, minimizes guidance abrupt changes, ensures stable passage in complex urban environments, and solves the technical problems described in the background art.

[0007] (II) Technical Solution To achieve the above objectives, the present invention provides the following technical solution: An adaptive multi-objective path planning method driven by high-precision BeiDou positioning includes collecting multi-source positioning and fusion results, correction services and communication link status, map environment and historical logs; generating a positioning quality field covering the road segment ahead under a unified reference, the positioning quality field includes error upper bound, remaining convergence time, interruption risk and correction timeliness, and setting a continuity budget and no-cut window; Based on the positioning quality field, continuous budget, and no-switching window, the candidate path is jointly solved by road segment and service channel. The optimal path, service channel plan for each road segment, and corresponding pre-charging start time and entry time are output by comprehensively considering travel time, positioning quality cost, and switching cost. Based on the service channel plan and time, and combined with road semantics, curvature and longitudinal acceleration, the entry threshold and exit threshold are determined and the dwell time is set; in the allowed window, the first handshake and caching of buffer zone one are performed first, then the applicable time alignment of buffer zone two is performed, and then the channel switching is smoothly completed and recorded; Under the optimal path and reference trajectory, read the records, establish rolling time-domain control with the upper bound of error and continuous budget as constraints, and output control commands; when the constraints are triggered or the budget margin is insufficient, execute speed limit, distance increase, lane keeping or lane change, and write to the log.

[0008] Furthermore, by discretizing the road ahead with a fixed arc length step and establishing a unified coordinate and time reference, the fusion positioning covariance and solution type, the corrected service link latency and packet loss, the three-dimensional occlusion and historical carrier-to-noise ratio are aligned segment by segment to obtain the upper bound of error, the remaining convergence time, the interruption risk, the remaining effective correction period and the probability of successful pre-charging.

[0009] Furthermore, the continuity budget includes the cumulative duration of protection level exceeding limits, the cumulative duration of no correction, and the upper limit of switching times; the no-cut window is generated based on the road semantics of ramps, merging, and sharp curves, and the window length is adjusted according to vehicle speed; when the time conflict with the no-cut window, it is pushed back to the center of the nearest non-no-cut road segment for recalculation.

[0010] Furthermore, the joint solution of road segments and service channels includes: constructing a comprehensive objective on the road segment-channel grid, modeling the positioning quality cost consisting of travel time and comfort, error upper bound and interruption risk, and the switching cost of channel changes in a unified manner, prohibiting state transitions within the no-switching window, and outputting the service channel plan and time.

[0011] Furthermore, the joint solution introduces opportunity constraints to limit the tail expectation of protection level exceeding the limit, and incorporates the continuous budget into the cumulative constraints; during the solution process, the cumulative cost is forward-progressed according to the road segment sequence and the preceding channel is recorded, and the optimal path, service channel plan and compatible time are obtained by backtracking at the end.

[0012] Furthermore, the entry and exit thresholds are determined by a weighted mapping of the positioning quality field gradient, road curvature, vehicle longitudinal acceleration, and key road segment indications, and minimum dwell time and minimum hold time are set; the number of channel switching times per unit distance is set to an upper limit based on the continuity budget, and switching is not triggered within the no-switching window.

[0013] Furthermore, at the start of pre-charging, buffer one is triggered to handshake with the target service and cache correction to the minimum cache duration; subsequently, before the entry time, buffer two completes the correction of applicable time alignment and coordinate time frame alignment, and performs residual detection. Only after passing the detection can channel switching be entered.

[0014] Furthermore, to smoothly complete the channel switching, a fusion weight curve is used to perform a convex combination of the old and new solutions. The first and second derivatives of the fusion weight at the start and end times of the switching interval are zero. When the allowed window is insufficient or the alignment is not completed, the entry time is delayed and recalculated in the nearest non-restricted segment.

[0015] Furthermore, the rolling time-domain control introduces an integrity margin barrier term under the optimal path and reference trajectory, and writes the upper bound of error and safety boundary into the constraints; the upper bound of speed is adjusted monotonically based on the continuous budget surplus and the risk of in-service interruption, and control commands are issued periodically.

[0016] Furthermore, safety downgrades are triggered by a composite risk index, sequentially executing speed limits and distance increases, lane keeping and lane changing restrictions, lane rerouting, or requesting takeover; and are written to the log with fixed fields, including the downgrade level, trigger reason, start and end times, action sequence, speed upper limit changes, and whether the no-cut window was hit.

[0017] (III) Beneficial Effects This invention provides an adaptive multi-objective path planning method driven by high-precision BeiDou positioning, which has the following beneficial effects: By establishing a unified forward-looking quantitative framework that integrates the positioning quality field, continuous budget, and no-cut window, the remaining convergence time, interruption risk, effective remaining correction period, and pre-charge success probability are aligned by road segment. This allows subsequent stages to read inputs from the same reference, reducing caliber bias and redundant estimation in path and channel selection.

[0018] The joint solution of road segment and service corridor incorporates travel time and comfort, error upper bound and interruption risk, and switching cost into a comprehensive objective. With opportunity constraints and continuous budget as hard boundaries, it simultaneously provides the optimal path, service corridor plan, and pre-charging start time and entry time, reducing the frequency of temporary changes.

[0019] Adaptive hysteresis switching uses the location of mass field gradient, road curvature, vehicle longitudinal acceleration and key road segment indicators to construct entry and exit thresholds and set dwell time, so that the switching trigger is transformed from a single numerical comparison to a planned event coupled with semantics and state. Combined with a no-switching window, the switching timing avoids sensitive sections such as ramps and sharp bends.

[0020] The double-buffered pre-charge performs handshake and caching at the start of pre-charge, and then completes the correction of applicable time alignment and coordinate time frame alignment before entering the time. Within the allowed window, it performs a smooth transition based on the fusion weight, so that the old solution and the new solution are connected continuously, reducing the impact of solution jumps on trajectory generation and velocity planning, and improving the executability of channel switching.

[0021] Rolling time-domain control compiles the error upper bound and continuity budget into an integrity margin barrier term and a velocity upper bound convergence law, forming a well-constrained control problem on the reference trajectory. It is linked with the pre-charge and switching records that have been executed. When constraints are triggered or the budget is tight, it implements graded degradation based on composite risk indicators to maintain consistency between control and plan. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating the adaptive multi-objective path planning method driven by high-precision BeiDou positioning of the present invention. Detailed Implementation

[0023] 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.

[0024] Please see Figure 1This invention provides an adaptive multi-objective path planning method driven by high-precision BeiDou positioning, including: Step 1: Under a unified coordinate-time reference, form a positioning quality field covering several road segments and various service channels ahead, and simultaneously determine the continuity budget and no-switching window, providing a unique and traceable input for the subsequent path-channel joint solution and switching execution.

[0025] In urban canyons, elevated ramps, and complex intersections, satellite-based precise point positioning broadcast (PPP-B2b), ground-based carrier phase real-time dynamic positioning (RTK or PPP-RTK), satellite-based augmentation (BDSBAS), and low-orbit satellite augmentation exhibit significant spatial differences in availability and solution stability. The same vehicle has different requirements in different road sections regarding whether it can converge before the critical location, whether the correction will lose synchronization in a short time, and whether the handover will cause a jump in the solution.

[0026] Therefore, the upper bound of error, remaining convergence time, risk of interruption, remaining period of correction, and probability of successful pre-filling must be calculated together on the look-ahead scale and expressed in the same data object so that subsequent steps can directly reference them and maintain consistency.

[0027] Therefore, the distance ahead is first discretized into an ordered sequence of road segments according to the road centerline, and the positioning and link observations are summarized within a unified coordinate-time reference. Then, for each type of service channel, five core indicators are calculated segment by segment, and a continuity budget and no-cut window are given. Finally, a positioning quality field with timestamps is formed for direct use in the next step.

[0028] The road centerline-segment sequence serves as the carrier for all subsequent calculations, and the same coordinate-time reference is used to eliminate alignment errors between the positioning solution, link observation, and map features. Through this preliminary step, the calculation and placement of the five core indicators are all carried out around the same road segment-service channel grid, thereby ensuring that subsequent steps can be directly retrieved and reused.

[0029] Wherein, the initial look-ahead distance is given. According to the fixed arc length step length road segment step length Divide the road into road segment sequences along the center line. .

[0030] To avoid misalignment of data from different sources on the timeline and baseline framework, all observations are mapped to a unified reference coordinate system and a unified timeline. The coordinates use the ground-fixed coordinate reference currently set by the vehicle system, and the time uses the device's timing reference; discrete samples are mapped to the timestamps at the center of each road segment through interpolation.

[0031] Furthermore, to provide an upper bound for subsequent calculation errors, the support radius of the elliptical envelope is used to give the upper bound for the error. The analytical expression: Where: Upper bound of error Used to describe road sections Use service channel Maximum possible error bound of the two-dimensional plane; coverage factor : Used to expand the feature radius to the envelope radius; maximum eigenvalue operator The input is a symmetric positive definite matrix; the maximum directional energy along the principal axes of the ellipse is extracted; the position covariance matrix is ​​also included. : is a set of symmetric positive definite matrices, which characterizes the error ellipse of the fused localization solution in the horizontal plane; Select a driving route within the Third Ring Road network that includes straight sections, curves, and elevated ramps, and determine the route based on the segment length. The look-ahead interval is divided into several road segments, and the coordinates and timestamps of the current fused localization solution are aligned to the center of each road segment.

[0032] The upper limit of the error is visible on several road sections before the elevated ramp. The continuously varying numerical trajectory obtained under the support radius formula will be directly referenced in subsequent steps for path selection and constraint setting. This prevents coordinate-time alignment errors from being propagated further, avoiding repeated computational costs for alignment issues in later stages.

[0033] In service channel collection Internally, link latency is aggregated channel by channel. Packet loss rate Availability Correct the applicable time and validity period of the message, and convert the occlusion information of the three-dimensional road structure into an occlusion index. Mapping historical carrier-to-noise ratio statistics to carrier-to-noise ratio .

[0034] All data are uniformly recorded at the center of the road segment, forming a one-to-one correspondence between the observation set and the road segment sequence. To avoid common linear superposition distortion, an exponential mapping is used to integrate occlusion and link factors into a single interruption risk. Prior form: Where: interruption risk : Value Its function is to describe the road segment Use service channel Quantization of the probability of corrective out-of-synchronization or unusable observations; channel sensitivity coefficient , , The values ​​are positive real numbers, representing the intensity of the impact of occlusion, latency, and packet loss on the risk of interruption. Obstruction Index : This reflects the degree to which 3D buildings obstruct satellite visibility; link latency. : A non-negative real number representing the time cost of receiving corrective information; packet loss rate : Value This characterizes the stability of the correction information during transmission; In the tree-shaded areas along the main urban roads, link logs and building models are read to generate shading indices in one go. Link latency , The road segment alignment sequence is written in real time to identify the risk of interruption. The prior trajectory is accurately captured; before entering the weaving ramp area, the trend of the interruption risk increasing significantly with the shading index is accurately captured, thus providing a direct basis for the estimation of the remaining convergence time and the setting of the continuity budget in subsequent steps.

[0035] After completing the alignment and aggregation, the prior form of the upper bound of error and the risk of interruption needs to be expanded into five indicators that can serve subsequent path selection and channel switching, and a continuity budget and a no-cut window should be set on the same object.

[0036] For the convergence process, a remaining lifetime representation based on the hazard rate is adopted, mapping the occlusion, visible satellite count, link quality, and current solution type on the road segment side to channels. On the road section The convergence velocity field is used to obtain the remaining convergence time. : Where: Remaining convergence time : A positive real number that represents the remaining time required to reach the target solution type from the current state; Risk rate function : A nonnegative real function that describes the time-varying behavior of a function. The rate of change reaches the target solution type; its specific form in engineering implementation can be obtained by using exponential family regression or spline regression to obtain a continuous function shape, with the input containing... , , , and the current solution type; Convergence risk rate : In road segment index Service Channel Under conditions, internal time The instantaneous convergence rate is a positive real number; Convergence risk baseline : Baseline intensity that varies only with the road environment and access conditions, a positive real number; Time Gating : Characterizes the asymptotic coefficient from the initial to the stable phase, as Monotonically increasing and not exceeding 1; Coefficient scalar coefficient vector Offline data is identified by historical logs; real numbers, dimensions, and feature vectors are consistent. Feature vector : A combination of measurable factors aligned by road segment, including occlusion index Visible star count Link latency Packet loss rate Solution type indicator Applicable time deviation absolute value Historical stable duration ; Warm-up time constant : Feedback Channel The timescale from startup to reaching a stable solution; Road segment index Service Channel Internal time : These correspond to the internal time axes of the road segment discretization, channel set, and convergence process in step one, respectively.

[0037] Integral variables and internal time variables , : Non-negative real numbers; Meanwhile, regarding the timeliness of the correction message, the remaining validity period of the correction is given based on the applicable time of the message and the device buffering strategy. Engineering calculation: At the arrival time at the center of the road segment, read the applicable time of the message and the local retention time to directly obtain the remaining available time.

[0038] On the elevated section, the occlusion index is... With link latency As a hazard rate function The input, the remaining time to complete convergence Numerical integration (using a Gauss-Legend quadrature algorithm); simultaneously read the applicable time of the message and write the valid remaining period. Therefore, the driving process can obtain a clear basis for judging whether the target solution type can be reached before approaching the ramp and whether the correction can remain effective at the critical position.

[0039] Whether the pre-charge process before the switchover can be completed depends on the matching relationship between the handshake duration and the available window. Therefore, the probability of successful pre-charge is given in the form of a distribution function. The precharge window duration In relation to the time scale of a handshake: Where: Probability of successful pre-charge :, range of values Reflected in the road section Does the system have enough time to complete the handshake and caching? (Precharge window duration) : A positive real number representing the available time determined by vehicle speed, road speed limit, and planned switching window; scale parameter : is a positive real number, used to describe the service channel. Typical handshake timescale; shape parameters : Positive real number, describing the tail shape of the handshake duration; With a probability of successful pre-charging and Afterwards, combined , , Set a continuous budget set on the same object :in, The cumulative duration of protection level exceeding the limit per unit distance; This represents the allowable cumulative uncorrected time per unit distance. This represents the number of handovers allowed per unit distance.

[0040] At the same time, based on road semantics (ramps, merging, sharp curves) and curvature thresholds, a set of no-cut windows is automatically generated. And bind it to the same road segment object.

[0041] Before entering the tunnel on an urban expressway, the system calculates based on vehicle speed. Then, with the service channel of , Together, we provide the probability of successful pre-charge. If there are insufficient windows, the pre-charge action for that channel will not be scheduled for that segment. Then, the continuity budget is written. With Forbidden Window This process is completed at the data object level, ensuring that subsequent steps can be directly based on the location quality field after it is read in, without the need for further derivation.

[0042] Step 2: On the same road segment-service channel grid, jointly decide which path to take, which service channel to use for each road segment, and when to precharge and switch, and output the time-based results that can be directly executed, with continuous budget and opportunity constraints as hard boundaries.

[0043] The previous phase has already provided five core indicators for each road segment and each service corridor, namely the upper bound of the error. Remaining convergence time Interruption risk Correcting the remaining effective period Probability of successful pre-charge and continuous budget sets With the set of forbidden windows .

[0044] Therefore, the second stage no longer re-extracts observations, but instead, under this unified caliber, places three types of quantities—travel time and comfort, location continuity risk, and channel switching cost—into the same objective, so that decision-making does not sacrifice traffic efficiency or ignore location continuity; and by using opportunity constraints to limit the tail expectation of exceeding the protection level limit, the output results have an engineering-executable risk boundary.

[0045] First, construct a comprehensive objective, including travel time. Cost of comfort The location quality cost and handover cost are coupled into the same objective function; then, the opportunity constraint and continuity budget are transformed into hard constraints in the decision-making process; subsequently, a linked solution is performed within the road segment-service corridor grid to obtain the path. With channel sequence The paired solution; finally generating the time-stamped output of the precharge start time and entry time, and the set of forbidden windows. Perform a compatibility check so that the output can be directly passed down to the next stage of execution.

[0046] Traffic elements and location continuity elements are combined and expressed in the same function, with opportunity constraints and continuity budgets embedded simultaneously. Therefore, it is necessary to first give an explicit expression for the comprehensive objective, and then incorporate the tail risk of exceeding the protection level limit into the constraint in a closed form, so that the subsequent solution does not rely on temporary threshold splicing and avoids the situation where the objective and constraint calibers are inconsistent.

[0047] To ensure that route selection and channel selection are on the same scale, the costs of travel time and comfort are weighted and combined with the costs of positioning quality and channel switching within the same objective. The comprehensive objective function is defined as follows: In the formula: comprehensive objective : Balancing traffic and location continuity factors within the same expression; weighting coefficients for travel time and comfort. , : Positive real number, adjusting the weight of transportation elements in the objective according to task preferences; travel time : Depicting the path Arrival time; Comfort Cost Constraining ride comfort and vehicle stability; for candidate paths The quantitative cost of smoothness and maneuverability of the driving process over time is used to penalize sharp turns, sudden accelerations and frequent maneuvers, so that traffic factors in the path-channel joint objective and ride smoothness are comparable in the same dimension. Positioning quality weighting coefficient , : Positive real number, controlling the upper bound of the error and the impact of the risk of interruption on the target; upper bound of the error : on the road section Use channels Upper bound of horizontal error; risk of interruption : Range of values In the section of road Use channels Quantification of the probability of correcting out-of-step or loss of visibility occurring; Switching cost weighting coefficient : Positive real number, suppressing high-frequency switching; 1 : Indicator function, dimensionless, takes the value when the passage of adjacent road segments changes. Otherwise, the value is Switching cost Units can be set according to time equivalence or energy consumption equivalence, and quantification can be performed from the channel. Switch to channel The combined overhead of convergence time, solution jump, cost, and critical segment penalty; the number of segments in the look-ahead interval. : Determine the termination position of the summation; In practice, traffic elements and positioning continuity elements are expressed in the same domain with unified dimensions, avoiding the fragmented process of first determining the path and then changing the channel in subsequent stages. The selection of path and channel sequences is determined simultaneously at the same level. By introducing a handover cost term, frequent handovers per unit distance can be suppressed, making subsequent hysteresis handovers and double-buffer pre-charging easier to complete as planned.

[0048] To control the tail risk of exceeding protection level limits, the following formula is used to limit the expected tail loss of exceeding the limit to not exceed a given threshold; simultaneously, the duration without correction and the number of switching times per unit distance are limited in a cumulative form. Opportunity constraints are given in the form of conditional expected value: In the formula: expectation operator Confidence level is The condition is that the expected loss is a dimensionless operator that measures the expected loss beyond the quantile threshold. Out-of-limit loss random variable : in path With channel sequence Below, the cumulative duration of protection level exceeding the limit; tail threshold. : Real number, representing the baseline level of the segmentation loss; confidence level : Value Determines the tail range; Mathematical expectation operator : Averaging of randomness; Positive part operator Defined as taking the larger of the input and zero, truncating negative values; upper limit of tail expectation. : Positive real number, used to provide an acceptable upper bound for the expected value of tail overflow; In addition to the opportunity constraints mentioned above, continuous budgeting is directly embedded in the cumulative amount: the cumulative duration without correction does not exceed The number of switching times per unit distance does not exceed Forbidden window set No switching is allowed within the system; By embedding tail risks in a closed form and placing them alongside continuous budgeting, the joint decision-making of paths and channels has clear risk boundaries, avoiding sudden degradation caused by tail exposure during the execution phase; the synchronous embedding of the no-cut window allows subsequent time schedule generation to naturally avoid ramps and merging segments.

[0049] After the comprehensive objectives and constraints are defined, a coordinated solution needs to be performed within the road segment-service channel grid to ensure that the path and channel sequences are determined synchronously. During the solution process, the pre-charging start time and entry time are incorporated into the decision-making process to ensure that the output can be directly executed in the next stage. To this end, piecewise dynamic programming supplemented by column generation is used to handle large-scale candidate path sets, and a temporal representation of the pre-charging-switching process is provided after the solution is completed.

[0050] On the segment-service channel grid, progress is made sequentially by segment, maintaining the minimum cost for each segment under each channel, and accumulating switching costs and budget usage during state transitions. The state transition recursive formula is as follows: Where: cumulative cost The unit and the overall goal are consistent, in the first... Each section will use a passageway. Minimum reachability cost; service channel set It includes discrete elements such as satellite-based precise point positioning broadcasting, ground-based carrier phase real-time dynamic positioning or precise point positioning, satellite-based augmentation, and low-Earth orbit satellite augmentation, and the channel dimension of the state space; Current road segment cost item Includes interrupts The time-comfort increment of the path itself is the instantaneous cost accumulated segment by segment; switching cost item : equals switch The unit is consistent with the overall objective, representing the additional cost incurred when the passage changes at the road segment boundary; road segment index. : For the recursive stage number; service channel , The service channels selected for the current and previous road segments are derived from the set. ; During the recursive process, any transfer that causes a change in the channel within the no-cut window is directly eliminated; any transfer that causes the cumulative uncorrected time or number of handovers to exceed the budget is also excluded from the candidate set; after the terminal segment is completed, the path is backtracked. With channel sequence The paired solution; When in use, the driving cost, positioning cost, switching cost, prohibited window, and budget usage are all bound together in the state transition; the path and channel sequence are output once in the backtracking phase to avoid disassembly or secondary repair in subsequent phases, thereby making it easier to align the pre-charging and switching time arrangements with the road geometry.

[0051] After obtaining the channel sequence, it is necessary to express when to start pre-charging and when to perform the switch as explicit time values, and to combine this with the set of forbidden windows. Perform a compatibility check. The precharge start time is given using a window pushback approach: Where: Precharge start time : for in the section of road Entering the passage The previous handshake and caching time point; the entry time. : Entry time into this channel, in seconds; target time for switching execution; remaining convergence time. : A positive real number, used to ensure the target solution type is achieved after the switch is complete; Minimum cache duration Positive real numbers ensure sufficient data buffering for corrective actions to support the transition; operators Operators that take the larger value; For any time that triggers precharge Falling into the set of forbidden windows In such cases, the system will backtrack to the center of the nearest non-restricted lane and recalculate the entry time. Until it is combined with the forbidden window set No longer conflicting; if with continuous budget sets If a contradiction occurs, return to the previous sub-step to adjust the channel sequence to eliminate the contradiction.

[0052] By pushing back through the entry window, the pre-charge start time and entry time are explicitly bound to the segment-channel object, and compatibility checks are performed one by one with the set of prohibited cutting windows. This result can be directly transformed into specific actions in the next stage, such as establishing a handshake and caching corrections in buffer one, and aligning the applicable time with the reference frame in buffer two.

[0053] Step 3: Based on the established plan of entry time and precharge start time, formulate and execute a continuous action chain of hysteresis threshold - dwell time - double buffer precharge - restricted fusion, so that channel switching avoids the no-switching window in space, meets the continuity budget in time, and avoids control spikes caused by dejumping in dynamics.

[0054] The previous phase gave the start time of pre-charging. With entry time However, if the gradient changes of road curvature, vehicle longitudinal acceleration and positioning mass field are not considered, direct switching can easily cause unnecessary yaw and longitudinal fluctuations at ramps, merging points or sharp bends.

[0055] Furthermore, there are differences in the applicable time and coordinate / timeframe between satellite-based precise point positioning broadcasting and ground-based carrier phase real-time dynamic positioning / precise point positioning. Abrupt, unaligned handovers can amplify road geometry misjudgments. Therefore, it is necessary to filter invalid handover triggers with an adaptive hysteresis threshold, complete handshake and alignment before entering the window through double buffer pre-charging, and finally achieve smooth fusion with restricted polynomial weights, thereby transforming the plan into deterministic on-site actions.

[0056] First, on the road segment-service channel grid, entry hysteresis thresholds and exit hysteresis thresholds are constructed using the positioning quality field gradient norm, road alignment curvature, and vehicle longitudinal acceleration, and minimum dwell time and minimum hold time are set. Then, the start-up and completion of buffer zone one and buffer zone two are comprehensively determined by the pre-charge window duration, remaining convergence time, minimum buffer duration, and applicable time deviation. Finally, within the allowable window, smooth fusion is performed with restricted fifth-order polynomial weights, while recording the switching action and the status of the in-service service channel. If any constraint triggers an out-of-bounds error, the entry time is delayed until the budget and window boundary are met.

[0057] A calculable and verifiable threshold expression is provided for when entry into a new channel and when exit from an old channel, while ensuring that no switching is triggered within the no-switching window. To avoid erroneous triggering in areas of abrupt changes in road geometry, the gradient norm of the mass field, the road linear curvature, and the vehicle's longitudinal acceleration are coupled as inputs to the threshold, and minimum dwell time and minimum hold time are specified to control the number of channel switching per unit distance.

[0058] The entry and exit hysteresis thresholds are not constants, but vary with the semantics and motion state of the road segment. Therefore, a mass field gradient norm is defined. Road alignment curvature Longitudinal acceleration Key road section indication key indication key And construct the entry hysteresis threshold using logical mapping. : Where: Entrance hysteresis threshold : Positive real number; the minimum utility difference threshold for determining whether entry into the target channel from the current channel is allowed; the baseline threshold. : Positive real number, providing a benchmark for the threshold scale; mass field gradient norm : Non-negative real number; represents the positioning mass field in the road segment The intensity of the change comes from the gradient of the location quality field written in the previous stage to the road segment index; curvature : A non-negative real number representing the curvature of the road alignment; longitudinal acceleration Real numbers, representing the longitudinal dynamic state of a vehicle; key road segment indicators. : Values ​​are a set Take at ramps, merging points, sharp bends, etc. Take other sections Weighting coefficients The value range is positive real numbers, adjusting the influence of each input on the threshold; exponential function. : A smoothing factor that maps linear combinations to the saturation range; Once the threshold is determined, exit the hysteresis threshold. Calculated in the same manner, but the weighting coefficients can be increased in key road sections to extend the holding period. This is then combined with the no-cutting window set generated in the previous stage. All windows that fall into the forbidden window set Potential switching actions are marked as invalid, ensuring that subsequent access only occurs within the allowed window.

[0059] When in use, the threshold surface incorporates changes in road geometry and mass field in sync, so that switching triggers are no longer driven by a single value; by raising the threshold in key road sections, switching is pushed away from ramps and merging areas, avoiding yaw and longitudinal fluctuations on the control side; the threshold uses a closed function, which is convenient for direct calculation at the vehicle end.

[0060] After the threshold surface is determined, a calculable channel utility is needed for comparison with the entry and exit hysteresis thresholds. The upper bound of error, link delay and packet loss, and solution type confidence are mapped to channel utility. and the shortest stay With the shortest holding time Joint determination: Where: channel utility : is a positive real number, representing time Using channels Overall superiority; non-negative weighting coefficient : and for The proportions of adjustment error, link type, and solution type in the utility; attenuation coefficient. : A positive real number that maps the upper bound of the error to the link quality as an exponential confidence level; Upper bound of error : Derived from the positioning quality field of the previous stage; link delay :aisle Correction arrival time; packet loss rate : ,aisle Link stability; solution type reliability : This maps discrete states such as integer fixed, floating point, and non-converged states to continuous confidence quantities. Entry is determined when the target channel has poor channel utility. Greater than the entry hysteresis threshold And the current channel has met the dwell time requirement. Entry is permitted only at designated times; Refers to the target channel with the highest utility; exit is determined when the utility of the current channel drops to the exit hysteresis threshold. The following and the duration You can exit only when the goal has been achieved.

[0061] When used, the channel utility constructed with exponential mapping responds smoothly to errors and link changes, avoiding jitter caused by drastic fluctuations in a single indicator; when combined with dwell-hold dual time base, the number of handovers per unit distance is naturally controlled, and the timing of subsequent double buffer pre-charging is also more executable.

[0062] Provided the handshake is allowed, it must be ensured that the target channel handshake is completed and the correction data is cached locally before the entry window arrives. Simultaneously, the correction application time must be aligned with the reference frame, and an acceptable residual detection result must be provided for the deviation between the two solutions. Subsequently, fusion weights are deployed within the allowed window to ensure the continuity of the fusion curve in terms of velocity and acceleration, avoiding control-side spikes.

[0063] Provide a single indicator for pre-charge readiness The precharge window duration Remaining convergence time Minimum cache Deviation from applicable time Unified to In the executable metrics, the sequential actions used to trigger buffer one and buffer two are: Where: Pre-charge readiness :,Pick The feasibility of the pre-charging action is determined comprehensively based on the pre-charging window duration. : Derived from the entry window pushback of the previous stage, the continuous duration that can be used for pre-charging; remaining convergence time : Derived from the positioning mass field, the time requirement to reach the target solution type after switching; Minimum cache duration Ensure sufficient data cache for correction; applicable time deviation. Correct the time difference between the applicable time and the entry time of the message; align the time constant. : Controls the rate at which the applicable timing deviation decays in readiness; exponential function Map the deviation to Alignment factor; operator : Take the larger value operator and apply stricter conditions between convergence and caching requirements; When pre-charge readiness When the threshold is exceeded, Buffer 1 is activated, establishes a handshake with the target channel and caches the correction; Buffer 2 is then activated, completes the alignment of the applicable time for correction, the alignment of coordinates and time frame (unifying the ephemeris reference, time base and the current vehicle base), and performs residual detection on the deviation between the two solutions. If the residual fails, the entry time is delayed and a new precharge window duration is written back.

[0064] When in use, a single metric aligns time availability, convergence requirements, caching requirements, and applicable time into a single scale, with a clear triggering order, avoiding the squeezing of buffering actions and entry actions; the fallback path when alignment fails is clear, ensuring that the switch only enters the allowed window after alignment is completed.

[0065] To avoid spikes on the control side caused by the de-jump transformation within the window, a fifth-order polynomial is used to construct the fusion weights. This ensures that velocity and acceleration remain continuous within the fusion interval, and the start and end times of fusion and weight curve parameters are written to the execution log. Where: fusion weight :for The time weight for convex combination of old and new solutions within the fusion interval; normalized time. : Values , to the actual time Linear mapping to unit interval; current time The independent variable of the fusion curve; Fusion start time The left endpoint of the fusion interval, corresponding to the entry time or a time shortly thereafter; fusion duration. The length of the fusion interval must be compatible with the continuity budget and the allowable window. At the execution layer, weights are merged. The old and new solutions are combined in a convex manner, and the trajectory generator and velocity planner only perceive the reference value for smooth transition. At the same time, six items, namely the start and end times of fusion, the duration of fusion, the weight curve parameters, the in-service service channels, the residual detection results, and whether the no-cut window is hit, are written into the pre-charge and switching records that have been executed and submitted to the next stage.

[0066] If any constraint is triggered during fusion (e.g., insufficient continuity budget margin), weight growth is stopped early and the current channel is maintained. The entry time is then re-evaluated in a new allowable window. In practice, the endpoint derivatives of the fifth-degree polynomial are zero, ensuring continuity in speed and acceleration during the fusion process and preventing spikes on the control side. The execution log is structured and written to disk, enabling the next stage to update constraints and determine degradation based on the executed actions, in-service status, and fusion curve.

[0067] Step 4: Under the unified semantics of road segment-service channel-time, based on the upper bound of the error. With continuous budget set Construct a rolling time-domain control problem with a barrier term, and output control instructions that satisfy integrity and budget constraints. When the integrity margin and budget margin reach the threshold, enter the safety degradation according to the preset sequence, and write the entire process behavior into the traceability log to form a self-closed-loop budget feedback update.

[0068] The previous phase has completed adaptive hysteresis handover and double-buffer pre-charging, and provided in-service service channels and fusion weights. And the pre-charging and switching records that have been executed. However, if the vehicle enters a tunnel, crosses an elevated merging zone, or travels through densely built-up areas, the upper limit of the positioning error increases, availability decreases, or the budget margin is exhausted, trajectory tracking alone will be insufficient to maintain the safety boundary. Therefore, it is necessary to explicitly introduce integrity constraints, budget feedback, and degradation actions at the control layer to translate positioning uncertainty into engineering quantities for the upper limit of speed, the lower limit of distance between vehicles, and the lane-changing threshold, thereby implementing the upper-level plan into executable lower-level control.

[0069] This phase is divided into two main lines: constraint compilation and rolling solution, and degradation determination and log closure. First, a rolling time-domain control objective and constraints are constructed using a combination of integrity margin, barrier function, and budget convergence law. Second, degradation behavior is generated and logged back using a composite risk indicator, action sequence, and budget deduction sequence for subsequent road segment calls and budget updates. These two main lines are based on in-service service channels and fusion weights. This serves as a link to ensure that the control layer and the switching layer are aligned at the parameter level.

[0070] Based on reference trajectory, in-service channel, and fusion weights - To ensure consistent input for the five indicators of the positioning mass field, first construct an integrity margin, then write it into the rolling time-domain target in the form of a barrier, and combine it with velocity, curvature, and control increments to form a co-domain target; simultaneously, the continuous budget set... The remaining amount is transformed into an explicit convergence law of speed upper bound and lane change threshold, which makes the control more conservative when the budget is tight and the control closer to the reference when the budget is ample, thus completing a unified solution of safety-efficiency-smoothness within a time domain.

[0071] In the rolling time domain length Internally, define integrity margin. The safety boundary width is the difference between the upper bound of the in-service error, and a barrier function is used to suppress the behavior of the margin approaching zero; thus, together with velocity deviation, curvature, and control increment, it constitutes a comprehensive objective. First, let's define the margin: The overall objective with barriers is then given: Where: integrity margin Positive real number, measuring the time of prediction. The margin between the safety boundary and the upper bound of the in-service error; the width of the safety boundary. : Positive real number, the minimum usable width derived from lane geometry, vehicle width, and preset lateral safety zone; error amplification factor. Positive real numbers, converting the upper bound of the error into boundary occupancy; Upper bound of in-service error : Positive real number, the upper bound sequence of errors predicted by the in-service service channel and the positioning quality field, which can be taken as... The temporal unfolding; Overall Objectives Dimensionless or treated with internally unified dimensions, the solution objective of model predictive control; weighting coefficients. , , : Positive real numbers, representing the weights of velocity deviation, curvature penalty, and control increment smoothing, respectively; Predicting speed The range is non-negative real numbers; the velocity trajectory to be solved is represented by the reference velocity. : , unit meters per second; value range is non-negative real number, derived from the composite reference of upper-level trajectory and road speed limit; path curvature : A non-negative real number, indicating a smooth path; controlling the increment. : A real number that limits the range of changes in throttle, braking, and steering angle; barrier coefficient : A positive real number, representing the penalty intensity when the integrity margin approaches zero; Length of the rolling time domain. : A positive integer representing the time span between prediction and solution; Through the coupling of integrity margin and logarithmic barrier, the controller automatically increases the penalty when the margin narrows, and prioritizes the solution of reducing speed, curvature and control increment to avoid entering the state where the upper bound of error touches the boundary; when the margin is sufficient, the target automatically returns to the trade-off between speed and smoothness, thereby realizing the continuous behavior of tightening and relaxing positioning uncertainty according to the situation.

[0072] To integrate continuous budget sets The remaining amount is transformed into the behavior space of the control layer, and the excess budget is converted into the behavior space of the control layer. The bounded convergence law is applied to the upper bound of the velocity. When the budget is tight, the upper limit of speed limits will be gradually tightened; when the budget is ample, the upper limit will be relaxed to the road speed limit. : Where: upper bound of velocity : is a positive real number, in the th order. Steps are used for constraints Maximum permissible speed; road speed limit : A positive real number, representing the speed limit jointly determined by regulations and road conditions; convergence law strength coefficient. :for Adjusting the impact of the budget on the upper limit of speed; estimating the risk of exceeding the limit. : is a non-negative real number, representing the expected over-limit occupancy amount within the rolling time domain, which can be derived from... The function is generated together with the integral of the risk of service interruption; over-limit budget. : Positive real number, the maximum cumulative duration allowed for exceeding protection level limits per unit distance; The upper bound of speed converges smoothly with changes in budget surplus and over-limit risk estimates: when the budget is tight or the risk increases, the upper bound of speed... The downward convergence allows the vehicle to actively decelerate and provide more lateral and longitudinal buffer space; this is suitable when the budget is ample and the risk is low. Approaching the limit This maintains traffic efficiency. This mapping provides a unified speed convergence premise for subsequent degradation triggers and action sequences.

[0073] When the integrity margin, budget convergence law, and in-service interruption risk all point to an increased risk, a single threshold is needed to switch actions such as entering speed limits, increasing distance, maintaining lanes, or changing lanes. The sequence of action switching and budget deduction are written to the traceability log. To avoid conflicts between multiple thresholds, a composite risk indicator is defined, and the degradation level is determined by comparing this indicator with the threshold. Subsequently, after execution, the remaining budget is deducted according to three categories: exceeding limits, no correction, and switching, forming a closed loop.

[0074] Composite Risk Indicators The upper bound is taken from the reciprocal of the minimum integrity margin, the cumulative risk of service interruption in the time domain, and the reciprocal of the minimum pre-completion readiness, so that degradation can be triggered when any one of them is under stress: Where: (Formula omitted for brevity) ; Composite risk index : A positive real number, used to uniformly measure three types of risks: tight surplus, accumulated interruptions, and difficulties in pre-charging; weighting coefficient : Values ​​range from positive real numbers, adjusting for the relative impact of the three types of risks in the indicator; integrity margin Values ​​range from positive real numbers; risk of service interruption. :for The probability trajectory of service interruption in active service channels; pre-charge readiness. The pre-charge readiness level of in-service channels. ; Length of rolling time domain The range of integral and extremum operations; Downgrade level trigger: When composite risk indicators When the speed is below the first threshold, the system remains normal; when it exceeds the first threshold, it enters a speed limit-increase-distance mode; when it exceeds the second threshold, it enters a lane-keeping mode-no-lane-change mode; when it exceeds the third threshold, it enters a lane-change mode-requests manual intervention. In use, a single indicator unifies the three types of risk sources, avoiding conflicts between multiple thresholds; the hierarchical action makes the process of control from gentle convergence to forced conservatism predictable, making it easy for drivers to perceive and accept.

[0075] After the downgrade action is executed, the newly added usage due to over-limit usage, uncorrected usage, and switchover counts need to be deducted from the continuous budget set. The remaining amount is recorded and written to the trace log in an append-only manner.

[0076] Specifically, in each rolling cycle, periods with incorrect integrity margins are converted into over-limit usage; periods with service interruption risks exceeding the entry threshold and correction failures are converted into uncorrected usage; and switching events are included in the switching count. Simultaneously, the degradation level, trigger reason, start and end times, action sequence, speed upper bound change, and whether the forbidden switching window is hit are written. Formulas are no longer introduced; all quantities are directly recorded in the current and previous stages to ensure readability and usability for the next segment. During use, budget deduction and log writing employ fixed fields and a fixed order to ensure complete reconstruction of current decisions and actions during playback; the budget continues between segments, forming a dynamic balance of use-saving-replenishment, providing continuous constraints for subsequent long-distance driving.

[0077] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0078] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0079] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0080] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0081] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An adaptive multi-objective path planning method driven by high-precision BeiDou positioning, characterized in that: include, Collect multi-source positioning and fusion results, correction service and communication link status, map environment and historical logs; generate a positioning quality field covering the road segment ahead under a unified reference. The positioning quality field includes the upper bound of error, the remaining convergence time, the risk of interruption and the correction time limit, and sets the continuity budget and the no-cut window. Based on the positioning quality field, continuous budget, and no-switching window, the candidate path is jointly solved by road segment and service channel. The optimal path, service channel plan for each road segment, and corresponding pre-charging start time and entry time are output by comprehensively considering travel time, positioning quality cost, and switching cost. Based on the service channel plan and schedule, and combined with road semantics, curvature and longitudinal acceleration, the entry threshold and exit threshold are determined and the dwell time is set; First, perform the handshake and buffering in the allowed window, then perform the timing alignment in the second buffer, and then smoothly complete the channel switching and record it; Under the optimal path and reference trajectory, read the records, establish rolling time-domain control with the upper bound of error and continuous budget as constraints, and output control commands; when the constraints are triggered or the budget margin is insufficient, execute speed limit, distance increase, lane keeping or lane change, and write to the log.

2. The adaptive multi-objective path planning method according to claim 1, characterized in that: Discretize the road ahead by a fixed arc length step and establish a unified coordinate and time reference. Align the fusion positioning covariance with the solution type, the corrected service link latency and packet loss, the 3D occlusion and the historical carrier-to-noise ratio segment by segment to obtain the upper bound of the error, the remaining convergence time, the interruption risk, the remaining effective correction period and the probability of successful pre-charging.

3. The adaptive multi-objective path planning method according to claim 2, characterized in that: The continuity budget includes the cumulative duration of protection level exceeding limits, the cumulative duration of no correction, and the upper limit of switching times; the no-cut window is generated based on the road semantics of ramps, merging, and sharp curves, and the window length is adjusted according to vehicle speed; when the time conflict with the no-cut window, it is pushed back to the center of the nearest non-no-cut road segment for recalculation.

4. The adaptive multi-objective path planning method according to claim 3, characterized in that: The joint solution of road segments and service channels includes: constructing a comprehensive objective on the road segment-channel grid, modeling the positioning quality cost consisting of travel time and comfort, error upper bound and interruption risk, and the switching cost of channel changes in a unified manner, prohibiting state transitions within the no-switching window, and outputting the service channel plan and time.

5. The adaptive multi-objective path planning method according to claim 4, characterized in that: The joint solution further introduces opportunity constraints to limit the tail expectation of protection level exceeding the limit, and writes the continuous budget into the cumulative constraints; during the solution, the cumulative cost is recursively calculated in the order of road segments and the preceding channel is recorded, and the optimal path, service channel plan and compatible time are obtained by backtracking at the end.

6. The adaptive multi-objective path planning method according to any one of claims 1-5, characterized in that: The entry and exit thresholds are determined by a weighted mapping of the positioning quality field gradient, road curvature, vehicle longitudinal acceleration, and key road segment indications, and minimum dwell time and minimum hold time are set; the number of channel switching times per unit distance is set to an upper limit based on the continuity budget, and switching is not triggered within the no-switching window.

7. The adaptive multi-objective path planning method according to claim 6, characterized in that: At the start of pre-charging, Buffer 1 is triggered to handshake with the target service and cache correction to the minimum cache duration; then, before the entry time, Buffer 2 completes the correction of applicable time alignment and coordinate time frame alignment, and performs residual detection. Only after passing the detection can channel switching be entered.

8. The adaptive multi-objective path planning method according to claim 7, characterized in that: To smoothly complete the channel switching, a fusion weight curve is used to convexly combine the old and new solutions. The first and second derivatives of the fusion weight at the start and end times of the switching interval are zero. When the allowed window is insufficient or the alignment is not completed, the entry time is delayed and the calculation is performed on the nearest non-restricted switching segment.

9. The adaptive multi-objective path planning method according to claim 8, characterized in that: Rolling time-domain control introduces an integrity margin barrier term under the optimal path and reference trajectory, and writes the upper bound of error and safety boundary into the constraints; the upper bound of speed is adjusted monotonically based on the continuous budget surplus and the risk of in-service interruption, and control commands are issued periodically.

10. The adaptive multi-objective path planning method according to claim 9, characterized in that: Safety downgrades are triggered by a composite risk index, sequentially executing speed limits and distance increases, lane keeping and lane changing restrictions, lane rerouting, or requesting takeover. The downgrade is logged with fixed fields, including the downgrade level, trigger reason, start and end times, action sequence, speed upper limit changes, and whether the no-cut window was hit.

Images