Roadblock detection method, system and electric vehicle

Abstract

The present application relates to electric vehicle auxiliary driving technical field, especially to a kind of roadblock detection method, system and electric vehicle, comprising: the pitch angle, roll angle and wheel speed difference of vehicle are acquired, boundary envelope is generated after fusion processing;Forward projection detection beam, only collect echo signal in the boundary envelope limited area, generate situation spectrum after time-frequency domain fusion;Interframe timing solution is carried out to situation spectrum, signal noise component is separated from it and determines environment correction coefficient, signal characteristics reflecting target object relative motion trend are extracted, and collision approaching index is constructed based on signal characteristics;Combine boundary envelope shape parameter, environment correction coefficient and collision approaching index, avoid danger operation instruction is output through decision control logic, and finally execute instruction to complete vehicle active intervention.By the present application, it can adapt to two-wheeled electric vehicle roll, jolt, skid and other complex driving conditions, suppress detection false alarm and improve the adaptability of roadblock detection and active risk avoidance.

Description

Technical Field

[0001] This invention relates to the technical field of electric vehicle assisted driving, and more particularly to a road obstacle detection method, system, and electric vehicle. Background Technology

[0002] There is a large number of two-wheeled electric bicycles on the road, and riders generally lack systematic driving skills training, resulting in a higher risk of accidents in complex road conditions. To improve riding safety, the industry is attempting to adapt motor vehicle driver assistance systems to two-wheeled electric bicycles, using onboard sensors to detect obstacles ahead and issue warnings to the rider.

[0003] However, the above solutions do not consider the driving characteristics of two-wheeled vehicles. When a vehicle corners, it needs to actively tilt, while the onboard sensors detect with a fixed field of view. This makes lane edge guardrails or road markings easily misjudged as threats, generating false alarms. In addition, on unpaved roads or in rainy or snowy weather, vehicle bumps and tire slippage introduce interference into the sensor signals. Existing methods lack means to handle such interference, resulting in a decrease in the reliability of obstacle detection.

[0004] In summary, existing methods lack an active risk avoidance mechanism that integrates vehicle motion state with driving environment disturbances into decision-making, making it difficult to accurately identify road obstacles and implement effective intervention in complex two-wheeled vehicle conditions. Summary of the Invention

[0005] This invention provides a road obstacle detection method, system, and electric vehicle, which can effectively solve the problems in the background art.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A roadblock detection method, comprising: The vehicle's pitch angle, roll angle, and wheel speed difference are obtained and then fused to generate a boundary envelope. The boundary envelope has morphological parameters that dynamically change with the pitch angle, roll angle, and wheel speed difference. Project a detection beam forward, collect reflected echo signals within the area defined by the boundary envelope, and perform time-frequency domain fusion processing on them to generate the situation spectrum of the current frame; Inter-frame temporal analysis is performed on the situation spectrum to separate the signal and noise components caused by driving environment excitation, and the environmental correction coefficient is determined based on the statistical characteristics of the signal and noise components. From the situation spectrum after separating the signal and noise components, the signal features reflecting the relative motion trend of the target object are extracted, and a collision approach index is constructed based on the signal features; The morphological parameters of the boundary envelope, the environmental correction coefficient, and the collision proximity index are combined and applied to the preset decision control logic to output the expected risk avoidance operation command. Execute the expected risk avoidance operation command to trigger active intervention on the vehicle.

[0007] Furthermore, the boundary envelope is generated through fusion processing, including: The pitch angle, roll angle, and wheel speed difference are coupled by dynamic parameters. Based on the coupling results, a three-dimensional virtual boundary is constructed with the vehicle body center of mass as the origin and dynamically changes with the pitch angle and roll angle, which serves as the boundary envelope. Among them, the lateral extension range of the boundary envelope narrows as the roll angle increases, the longitudinal extension range narrows as the pitch angle increases, and the overall coverage area of ​​the boundary envelope shrinks when the wheel speed difference exceeds the preset slip threshold.

[0008] Furthermore, the detection beam includes a millimeter-wave radar beam and an infrared structured light beam; the reflected echo signal is fused in the time and frequency domain to generate the situation spectrum of the current frame, including: aligning the timestamps and fusing the spatial coordinates of the echo signals from the millimeter-wave radar beam and the infrared structured light beam to form a composite data matrix including distance information and three-dimensional contour information, which serves as the situation spectrum of the current frame.

[0009] Furthermore, inter-frame temporal analysis is performed on the situation spectrum to separate the signal-noise components caused by driving environment excitations, including: Take N consecutive frames of situation spectrum and extract the time series signal of each frame at the corresponding spatial location. The time series signal is separated to obtain low-frequency and high-frequency components. The high-frequency components are regarded as signal noise components caused by driving environment excitation.

[0010] Furthermore, environmental correction coefficients are determined based on the statistical characteristics of the signal and noise components, including: Calculate the root mean square value of the signal noise components; Based on the preset range of values ​​to which the root mean square value belongs, the corresponding environmental correction coefficient is determined. The larger the root mean square value, the smaller the environmental correction coefficient, indicating that the current driving environment has a higher degree of interference with the sensor signal.

[0011] Furthermore, signal features reflecting the relative motion trend of the target object are extracted from the situation spectrum after separating the signal and noise components. Based on these signal features, a collision approach index is constructed, including: For the area located within a preset angle range directly in front of the vehicle in the situation spectrum after separating the signal and noise components, time-frequency transformation processing is performed to obtain the range Doppler spectrum. Extract the energy distribution of the preset target approach frequency band from the distance-Doppler spectrum, calculate the energy attenuation coefficient based on the energy distribution, and use the normalized energy attenuation coefficient as the collision approach index.

[0012] Furthermore, the morphological parameters of the boundary envelope, the environmental correction coefficient, and the collision proximity index are combined and applied to the preset decision control logic to output the expected risk avoidance operation instructions, including: The coverage volume and asymmetry coefficient of the boundary envelope are used as morphological parameters and input into the decision control logic along with the environmental correction coefficient and collision proximity index. The decision control logic is configured as follows: when the environmental correction coefficient indicates that the interference of the driving environment on the sensor signal exceeds the preset interference threshold, the response weight of the collision approach index in the decision is increased; when the morphological parameters of the boundary envelope indicate that the vehicle stability is lower than the preset stability threshold, the output weight of steering avoidance commands is reduced while the output weight of braking commands is increased.

[0013] Furthermore, executing the anticipated risk avoidance maneuver command triggers active intervention on the vehicle, including at least one of the following: Trigger the vehicle's audible and visual alarm device to output an alarm signal; Send a torque limiting command to the vehicle motor controller to reduce the output power of the vehicle drive motor; Send braking commands to the vehicle's braking system to establish braking torque on the vehicle's wheels.

[0014] Secondly, the present invention also provides a road obstacle detection system, comprising: The attitude acquisition module is used to acquire the vehicle's pitch angle, roll angle, and wheel speed difference. After fusion processing, a boundary envelope is generated. The boundary envelope has morphological parameters that dynamically change with the pitch angle, roll angle, and wheel speed difference. The detection fusion module is used to project a detection beam forward, collect reflected echo signals within the area defined by the boundary envelope, and perform time-frequency domain fusion processing on them to generate the situation spectrum of the current frame. The noise correction module is used to perform inter-frame temporal analysis on the situation spectrum, separate the signal noise components caused by driving environment excitation, and determine the environment correction coefficients based on the statistical characteristics of the signal noise components. The feature extraction module extracts signal features reflecting the relative motion trend of the target object from the situation spectrum after separating the signal and noise components, and constructs a collision approach index based on the signal features; The decision control module combines the morphological parameters of the boundary envelope, the environmental correction coefficient, and the collision proximity index with the preset decision control logic to output the expected risk avoidance operation command. The execution module is used to execute the expected risk avoidance operation instructions, triggering active intervention on the vehicle.

[0015] Thirdly, the present invention also provides an electric vehicle that employs the obstacle detection method described above.

[0016] The technical solution of this invention can achieve the following technical effects: it adapts to complex driving conditions such as cornering and tilting of two-wheeled electric vehicles, bumpy unpaved roads, and tire slippage in rainy or snowy weather; it avoids false alarms caused by fixed field-of-view detection by dynamically generated boundary envelopes; and it improves the reliability of sensor signals by separating environmental excitation noise components and determining environmental correction coefficients. By combining collision proximity indicators and multi-parameter collaborative decision-making, it achieves accurate identification and graded active intervention of road obstacles, improves the accuracy of road obstacle detection and the adaptability of active risk avoidance for two-wheeled electric vehicles, and ensures riding safety.

[0017] 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

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating a road obstacle detection method according to the present invention; Figure 2 This is a schematic diagram of the structure of a road obstacle detection system according to the present invention. Detailed Implementation

[0020] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0022] like Figure 1 As shown, a road obstacle detection method of the present invention specifically includes the following steps: Step S100: Obtain the vehicle's pitch angle, roll angle, and wheel speed difference, and generate a boundary envelope through fusion processing. The boundary envelope has morphological parameters that dynamically change with the pitch angle, roll angle, and wheel speed difference. Step S200: Project a detection beam forward, collect the reflected echo signal within the area defined by the boundary envelope, and perform time-frequency domain fusion processing on it to generate the situation spectrum of the current frame. Step S300: Perform inter-frame temporal analysis on the situation spectrum, separate the signal and noise components caused by driving environment excitation, and determine the environment correction coefficient based on the statistical characteristics of the signal and noise components; Step S400: Extract signal features reflecting the relative motion trend of the target object from the situation spectrum after separating the signal and noise components, and construct a collision approach index based on the signal features; Step S500: The morphological parameters of the boundary envelope, the environmental correction coefficient, and the collision proximity index are applied together to the preset decision control logic to output the expected risk avoidance operation command. Step S600: Execute the expected risk avoidance operation command to trigger active intervention on the vehicle.

[0023] In this embodiment, a dynamic boundary envelope is generated by pitch angle, roll angle, and wheel speed difference to limit the detection range and eliminate false alarm sources such as guardrails when the vehicle is tilted. The echo signal is fused in the time and frequency domains to form a situation spectrum, and then environmental noise is separated and the degree of interference is quantified through inter-frame time-series calculation to distinguish whether the signal attenuation is caused by real obstacles or road bumps. The relative motion trend of the target is extracted from the situation spectrum after noise removal to construct a collision approach index, avoiding misjudging stationary targets as threats. The boundary envelope shape, environmental correction coefficient, and collision approach index are input into the decision logic to output a risk avoidance command. The combination of the three can adaptively narrow the lateral detection when cornering and adaptively reduce the sensor weight when there are bumps.

[0024] In a specific implementation, as one example, given that the pitch angle, roll angle, and wheel speed difference of a two-wheeled electric bicycle change in real time during operation, directly altering the vehicle's attitude and motion controllability in space, and that existing sensors have a fixed field of view, a mismatch occurs between the detection area and the actual stable state of the vehicle. Therefore, it is necessary to construct a detection boundary that dynamically adjusts with the real-time attitude and driving state of the vehicle, matching the sensor signal acquisition range with the actual spatial area where a collision may occur under the current operating conditions. This embodiment achieves real-time tracking of the vehicle's stability state by fusing pitch angle, roll angle, and wheel speed difference to generate a boundary envelope with dynamic morphological parameters, as detailed below: Step S110: Obtain the vehicle pitch angle, roll angle, and wheel speed difference; The pitch angle is defined as the angle between the vehicle's longitudinal axis and the horizontal plane. This angle is zero when the vehicle is stationary on a horizontal surface. During acceleration, the rear of the vehicle dips and the front of the vehicle rises, resulting in a positive pitch angle. During braking, the front of the vehicle dips and the rear of the vehicle rises, resulting in a negative pitch angle. The roll angle is defined as the angle between the vehicle's transverse plane and the vertical plane. This angle is zero when the vehicle is upright. When cornering to the right, the vehicle tilts to the right, resulting in a positive roll angle. When cornering to the left, the roll angle is negative. Both are obtained through an inertial measurement unit installed at the vehicle's center of gravity. The inertial measurement unit uses a combination of a three-axis accelerometer and a three-axis gyroscope. The sampling frequency can be set to 50Hz. This frequency can capture the dynamic changes in the vehicle's attitude in real time while avoiding signal redundancy and increased power consumption caused by excessively high sampling frequencies. The wheel speed difference is defined as the difference between the speed of the rear wheel and the speed of the front wheel, and is obtained through two Hall effect wheel speed sensors. The two sensors are installed on the inner side of the two wheel hubs respectively, and collect the speed of the front wheel and the speed of the rear wheel. The wheel speed difference is obtained by subtracting the two. The sampling frequency of the wheel speed sensors is consistent with that of the inertial measurement unit to ensure that the wheel speed difference data is synchronized with the pitch angle and roll angle data in time. Step S120: Obtain the pre-stored boundary generation rule. This rule is represented in the form of a three-dimensional spatial coordinate function. The origin of the coordinate system is set at the center of mass of the vehicle body, the X-axis points directly in front of the vehicle, the Y-axis points to the right side of the vehicle, and the Z-axis is vertically upward. The boundary envelope is defined as a region enclosed by a closed surface. This region extends a distance L along the positive X-axis, extends half a width W symmetrically along the Y-axis to the left and right, and extends a height H upward along the Z-axis. Step S130: Obtain the reference longitudinal extension distance , baseline horizontal half width and reference vertical height Among them, the benchmark longitudinal extension distance The maximum distance at which a vehicle can reliably trigger a warning under standard operating conditions, defined by the baseline lateral half-width. The reference vertical height is the maximum lateral offset distance that the vehicle can stably detect obstacles under standard operating conditions. The maximum height at which the vehicle can stably detect obstacles under standard operating conditions is set; the standard operating conditions are set with pitch angle of zero, roll angle of zero, and wheel speed difference of zero. All benchmark parameters are pre-determined and stored through actual vehicle calibration tests. Step S140: Based on the real-time pitch angle Determine the vertical scaling factor Vertical scaling factor Used to adjust the longitudinal extension distance of the boundary envelope Its value is related to the real-time pitch angle. The absolute value is positively correlated, ensuring that the longitudinal extension range narrows as the absolute value of the pitch angle increases; the calculation formula is:

[0025] in, The pitch angle influence coefficient is determined by simulating different pitch conditions, collecting and analyzing data, and then determining its value through actual vehicle calibration tests, taking into account vehicle length, braking and acceleration characteristics. Based on real-time roll angle Determine the horizontal scaling factor Horizontal scaling factor Used to adjust the horizontal half-width of the boundary envelope Its value is related to the real-time roll angle. The absolute value is positively correlated, ensuring that the lateral extension range narrows as the absolute value of the roll angle increases; the calculation formula is:

[0026] in, The roll angle influence coefficient is determined by simulating different roll conditions, collecting and analyzing data, and then determining its value through actual vehicle calibration tests, taking into account vehicle width and cornering characteristics. Based on wheel speed difference Determine the overall scaling factor Overall scaling factor Longitudinal extension distance used for synchronous adjustment of boundary envelope Horizontal half width and vertical height Its value is determined by the difference in wheel speeds. With preset sliding threshold The size relationship determines that when the wheel speed difference exceeds the preset slip threshold, all dimensions of the boundary envelope are reduced proportionally, resulting in a smaller overall coverage area; the calculation formula is:

[0027] in, The preset slip threshold is pre-set based on vehicle tire characteristics and road conditions through real vehicle calibration, and is used to determine whether the wheels are slipping significantly. The slippage scaling factor is determined by real-vehicle calibration tests simulating different road slippage conditions, taking into account the wheel grip performance. Step S150: Apply the scaling factors together to the baseline parameters to obtain the actual shape parameters of the boundary envelope at the current time:

[0028]

[0029]

[0030] Based on the above parameters, the spatial region corresponding to the boundary envelope is represented as follows:

[0031] This region is the three-dimensional spatial range enclosed by the boundary envelope with dynamic morphological parameters.

[0032] In this embodiment, the generated boundary envelope (L, W, H) changes in real time with the vehicle's attitude and driving state, directly limiting the signal acquisition range of the sensor detection beam. When the absolute value of the pitch angle increases, the front suspension travel is compressed or stretched, and the relative approach speed of the vehicle to the obstacle in front changes abruptly. At this time, the longitudinal scaling factor narrows L to avoid misjudging distant stationary targets as immediate collision threats. When the absolute value of the roll angle increases, the lateral acceleration of the vehicle increases, reducing the physical feasibility of steering or lane changing. At this time, the lateral scaling factor narrows W to exclude stationary targets such as lane edge guardrails or road markings. When the wheel speed difference exceeds the preset slip threshold, the rear wheels spin, causing the vehicle's trajectory to become uncontrollable and the sensor signal to be mixed with high-frequency noise. At this time, the overall scaling factor reduces L, W, and H proportionally to filter out interference signals caused by tire slippage.

[0033] In some embodiments of the present invention, existing methods for detecting obstacles ahead typically employ a single type of sensor. For example, millimeter-wave radar can acquire target distance and velocity information, but its ability to distinguish target contour details is limited, making it unable to differentiate between obstacles and low-lying ground structures. Infrared structured light can acquire the target's three-dimensional contour information, but it is easily affected by changes in ambient lighting, resulting in insufficient reliability when used alone. To address this issue, a detection method combining millimeter-wave radar beams and infrared structured light beams is required. This leverages the complementary advantages of both to compensate for the shortcomings of a single sensor. Furthermore, time-frequency domain fusion processing is performed on the two types of signals to generate a composite data matrix containing both distance and three-dimensional contour information as a situational spectrum. Specifically, the following operations are performed: Step S210: Simultaneously project a millimeter-wave radar beam and an infrared structured light beam forward; both the millimeter-wave radar module and the infrared structured light module are installed at the front of the two-wheeled electric bicycle, in front of the vehicle's center of gravity, with the installation position fixed relative to the vehicle's center of gravity; the millimeter-wave radar beam is generated by the vehicle-mounted millimeter-wave radar transmitting module, and the infrared structured light beam is generated by an infrared laser projector; the projection angle range of the infrared structured light beam is consistent with the field of view of the millimeter-wave radar beam, ensuring that the two beams overlap in space to cover the same detection area; Step S220: Receive the reflected echo signal and perform spatial filtering on the received echo signal based on the boundary envelope; for the echo signal of millimeter-wave radar, the radar receiver collects the original complex signal, calculates the target distance corresponding to each echo point according to the echo propagation time, retains the echo points located within the boundary envelope space area, and discards the remaining echo points; for infrared structured light images, the infrared camera collects the infrared spot grayscale image modulated by the target surface, backprojects each pixel to three-dimensional space according to the camera calibration parameters, retains the pixels located within the boundary envelope space area, and sets the pixels located outside the envelope to invalid values; Step S230: Perform time-frequency domain processing on the spatially filtered millimeter-wave radar echo signal and infrared structured light image, including timestamp alignment and spatial coordinate fusion. Timestamp alignment adopts a unified timing reference, using the sampling timing of the inertial measurement unit as the reference, and adds a unified timestamp to the two signals. For cases where the sampling frequencies are inconsistent, interpolation is used to supplement the signal data with missing timing, ensuring that the two signals correspond to the same detection time under the same timestamp, and avoiding fusion distortion caused by timing deviation. Spatial coordinate fusion uses the vehicle body centroid coordinate system as the unified coordinate system. The distance information of the millimeter-wave radar echo signal and the three-dimensional contour coordinates of the infrared structured light echo signal are converted into three-dimensional coordinates under this unified coordinate system through coordinate transformation formulas, ensuring that the coordinate systems of the two signals are consistent. Step S240: Integrate the data of the two signals in the current frame to form a composite data matrix of the current frame as a situation spectrum; the composite data matrix is ​​defined as follows: within the spatial region defined by the boundary envelope, all valid echo points after spatial filtering and time alignment are arranged into a two-dimensional array. The rows of the array correspond to each valid target echo point, and the columns of the array store the three-dimensional coordinates, three-dimensional contour parameters and amplitude of the echo signal of the echo point in a unified coordinate system.

[0034] In this embodiment, the millimeter-wave radar can still output reliable distance information under adverse lighting conditions such as rain, snow, and strong light, while the infrared structured light supplements the fine geometric features of the target under good lighting conditions. The two are fused in the time and frequency domains to form a composite data matrix, which allows for obstacle identification based on the situation spectrum using both the distance and contour dimensions. By limiting the acquisition range of the detection beam to within the boundary envelope, echo signals from non-interested spatial areas can be eliminated at the source, thereby improving the accuracy of obstacle detection for two-wheeled electric vehicles under complex conditions.

[0035] In a specific implementation, as one example, given that a single-frame situation spectrum only contains spatial distribution information at the current moment, it is impossible to distinguish whether the amplitude fluctuation of a certain echo point in that frame originates from a real obstacle or environmental noise. Since real obstacles manifest as low-frequency components in consecutive frames, and environmental excitations manifest as high-frequency components, the temporal information of consecutive frames can be used to separate noise components and quantify the interference level. This embodiment establishes a time-series signal of the situation spectrum at corresponding spatial locations across consecutive frames, and uses low-pass filtering to separate it into low-frequency and high-frequency components. The specific implementation steps are as follows: Step S310: Obtain N consecutive frames of situation spectrum, arrange them in order of timestamps to form a multi-row and multi-column composite data sequence; the value of N is determined by actual vehicle calibration. Based on the vehicle speed and detection frequency, the higher the speed and detection frequency, the higher the value of N will be, to ensure that the random mutation characteristics of noise components and the continuous change characteristics of target signals can be captured. Step S320: Select multiple spatial feature points that are uniformly distributed within the boundary envelope, extract the signal amplitude data of each spatial feature point in N consecutive frames, and form the time series signal of the spatial feature point. Step S330: Use a low-pass filter to perform low-pass filtering on the time series signal corresponding to each spatial feature point. The cutoff frequency of the low-pass filter is determined according to the frequency range of vehicle body vibration caused by road bumps in the driving environment of the two-wheeled electric vehicle. The low-pass filter retains frequency components below the cutoff frequency as low-frequency components, which correspond to the echo contribution of real obstacles and the slow attitude change of the vehicle body over a long time scale. Subtract the low-frequency components from the original time series signal to obtain high-frequency components, which correspond to the signal noise components caused by driving environment excitation. Step S340: For each spatial feature point, take the values ​​of each frame in the high-frequency component time series at that point, calculate the sum of squares divided by the sequence length, and then take the square root to obtain the local root mean square value of that point; perform an arithmetic mean of the local root mean square values ​​of all spatial feature points within the boundary envelope to obtain the global root mean square value; the global root mean square value quantifies the overall interference degree of the current driving environment on the sensor signal. The larger the value, the stronger the bump or slip excitation in the environment, and the higher the proportion of noise components mixed into the sensor signal. Step S350: Determine the corresponding environmental correction coefficient based on the preset numerical interval to which the root mean square (RMS) value belongs. The preset numerical interval is determined through actual vehicle calibration. The calibration process simulates common driving environments for two-wheeled electric bicycles, including typical working conditions such as smooth roads, slightly bumpy roads, severely bumpy roads, and rainy / snowy roads. Under each working condition, the RMS values ​​of the signal and noise components are collected, and the numerical interval is divided according to the distribution range of the RMS values. For each global RMS value interval, an initial value for the environmental correction coefficient is set, and the coefficient value is iteratively adjusted according to the test results so that the false alarm rate and false negative rate under the corresponding working condition of the interval reach a preset balance point. The finally determined coefficient value is bound to the interval and stored as a lookup table. In real-time operation, the environmental correction coefficient of the current frame is obtained by querying the lookup table through the global RMS value of the current frame. In this embodiment, N consecutive frames of situation spectrum are selected. Utilizing the temporal differences between the target signal and noise components, a low-pass filter is used to separate the noise components. Then, the root mean square (RMS) value is used to quantify the degree of environmental interference, and an environmental correction coefficient is determined, achieving synergy between noise separation and environmental interference quantification. On unpaved roads or in rainy or snowy weather, vehicle bumps cause the amplitude of high-frequency components to increase, the global RMS value to increase, and the environmental correction coefficient to decrease. The decision logic can thereby reduce the confidence level of the sensor signal and avoid misjudging echo fluctuations caused by road bumps or tire slippage as obstacle approaching signals. On smooth roads, the amplitude of high-frequency components is lower, and the environmental correction coefficient remains at a higher value, maintaining sensitivity to real obstacles.

[0036] In a specific implementation, as one example, given that the situation spectrum after separating the signal and noise components has eliminated interference caused by environmental excitation and retained the effective signal of the target object, but this situation spectrum only contains the spatial coordinates, contour parameters, and signal amplitude information of the target object, it cannot directly reflect the relative motion state between the target object and the vehicle; if the collision risk is judged solely based on the distance parameter, misjudgment is likely to occur. Therefore, it is necessary to focus on the core detection area directly in front of the vehicle, extract the signal features corresponding to the relative motion of the target object through time-frequency transformation, quantify the approach degree of the target object using energy distribution, construct a collision approach index, and realize the prediction of collision risk. The specific implementation steps are as follows: Step S410: Extract a spatial region located within a preset angle range directly in front of the vehicle from the situation spectrum after separating the signal and noise components; the preset angle range is determined based on the horizontal field of view of the millimeter-wave radar beam, and this range covers the main risk areas in the direction of vehicle travel, while excluding interference from stationary targets on the side to collision detection. Step S420: Perform time-frequency transformation processing on the intercepted situation spectrum data to obtain the range Doppler spectrum; the time-frequency transformation adopts short-time Fourier transform or fast Fourier transform to transform the echo signal in each range gate from the time domain to the frequency domain; in the transformed range Doppler spectrum, the horizontal axis represents the Doppler frequency, corresponding to the radial velocity of the target relative to the vehicle, and the vertical axis represents the distance; Step S430: Extract the energy distribution of the preset target approach frequency band from the range Doppler spectrum; set the frequency range where the positive Doppler frequency shift is located as the target approach frequency band, which corresponds to the motion state of the target in front of the vehicle remaining stationary or continuously approaching. Step S440: Take multiple range slices from the front of the vehicle to the boundary envelope in the range Doppler spectrum, extending longitudinally over a distance L. Calculate the cumulative energy in the target approach frequency band on each range slice, fit the energy attenuation curve as a function of distance, and take the absolute value of the attenuation slope of the attenuation curve as the energy attenuation coefficient. The larger the energy attenuation coefficient, the more concentrated the target is at a relatively close distance and the more obvious the approach trend is. Step S450: The energy attenuation coefficient is normalized and used as a collision approach index. The normalization process uses linear mapping. The lower and upper limits of the energy attenuation coefficient are determined through actual vehicle calibration. The value of the energy attenuation coefficient in this range is linearly mapped to a dimensionless value between 0 and 1. When the normalized collision approach index is close to 1, it indicates that there is a target with a strong approaching trend ahead. When the index is close to 0, it indicates that there is no threatening target ahead or the target is moving slowly.

[0037] In this embodiment, by focusing on the core detection area directly in front of the vehicle, the echo signal is converted from the time domain to the range Doppler domain using time-frequency transformation, allowing the relative motion characteristics of the target to be explicitly expressed. Furthermore, the energy distribution within the preset target approach frequency band can be extracted, an energy attenuation curve as a function of distance can be fitted, and the energy attenuation coefficient can be calculated. After normalization, a collision approach index is obtained, used to distinguish approaching targets from stationary or distant targets, avoiding misjudging stationary road markings or vehicles moving away from oncoming lanes as collision threats. This mechanism, based on a situation spectrum that has already filtered out environmental noise components, further utilizes frequency domain information to extract relative motion trends, thereby improving the accuracy of collision detection.

[0038] In some embodiments of the present invention, in order to incorporate the real-time stability of the vehicle body and the degree of disturbance in the driving environment into the risk avoidance decision, the morphological parameters of the boundary envelope, the environmental correction coefficient, and the collision proximity index need to be input into the decision control logic. By dynamically adjusting the response weights and command priorities under different operating conditions, risk avoidance operation commands adapted to the current vehicle body state and the degree of environmental disturbance are output, specifically performing the following operations: Step S510: Extract the coverage volume and asymmetry coefficient from the morphological parameters of the boundary envelope; the coverage volume is obtained by multiplying the longitudinal extension distance L, the lateral half-width W, and the vertical height H of the boundary envelope of the current frame, which represents the size of the detection area. The smaller the volume, the lower the vehicle stability; the asymmetry coefficient is calculated by obtaining the lateral extension range of the boundary envelope on the left and right sides of the vehicle, and taking the ratio of the absolute value of the difference between the left and right half-widths to the sum of the left and right half-widths, which represents the vehicle's cornering state and the degree of sharpness. The sharper the cornering, the larger the value. Step S520: Input the coverage volume, asymmetry coefficient, environmental correction coefficient and collision proximity index into the decision control logic; the decision control logic is stored in the form of a preset judgment rule table. The rule table is established through real vehicle calibration. The calibration process simulates various typical working conditions, records relevant data and determines the optimal avoidance strategy, and solidifies and stores the mapping relationship between parameter range and output command. Step S530: The decision control logic judges the environmental correction coefficient: When the environmental correction coefficient is lower than the preset interference threshold, it indicates that the current driving environment has a high degree of interference with the sensor signal and the reliability of the collision approach indicator has decreased. The response weight of the collision approach indicator in the decision is increased. Step S540: The decision control logic judges the coverage volume and asymmetry coefficient: when the coverage volume is lower than the preset volume threshold or the asymmetry coefficient is higher than the preset asymmetry threshold, it indicates that the vehicle stability is insufficient, and the output weight of steering avoidance commands is reduced while the output weight of braking commands is increased. Step S550: The decision control logic outputs the expected risk avoidance operation instruction based on the weight adjustment result.

[0039] In this embodiment, the coverage volume and asymmetry coefficient of the boundary envelope are used as quantitative representations of the vehicle stability state, and the environmental correction coefficient is used as a quantitative representation of the reliability of the sensor signal. These are input together with the collision approach index into the decision control logic to achieve multi-condition adaptive decision-making: when the environmental interference is large, the weight of the collision approach index is increased to avoid missed detection; when the vehicle stability decreases, the weight of the steering command is reduced and the weight of the braking command is increased to avoid rollover accidents caused by improper steering.

[0040] In some embodiments of the present invention, in order to convert the output expected risk avoidance operation command into actual control of the vehicle actuators, it is necessary to trigger corresponding active intervention actions according to the command type. Different types of commands correspond to different actuators and control methods. The alarm command acts on the audible and visual alarm device to remind the rider to pay attention to the risks ahead; the torque limiting command acts on the motor controller to reduce the output power of the drive wheel to reduce the approach speed; the braking command acts on the braking system to establish braking torque on the wheels to achieve deceleration or stopping, specifically performing the following operations: Step S610: Receive the expected risk avoidance operation command and parse the command type code; the command type code includes three independent types: alarm command, torque limit command, braking command, and any combination of the three; Step S620: If the instruction type code contains an alarm instruction, the audible and visual alarm device is triggered to output an alarm signal. The audible and visual alarm device includes a buzzer and an LED indicator. The installation location is selected to ensure that the alarm signal can be clearly perceived by the cyclist and surrounding vehicles and pedestrians at the same time. The buzzer uses a medium frequency band to emit sound, and the LED indicator uses a red light flashing mode to remind the cyclist to pay attention to the risks ahead, and at the same time warn surrounding vehicles and pedestrians, so as to buy time for the cyclist to take the initiative to avoid danger. Step S630: If the instruction type code includes a torque limiting instruction, a torque limiting instruction is sent to the vehicle motor controller to reduce the output power of the vehicle drive motor, thereby reducing the speed at which the vehicle approaches the obstacle in front, giving the rider more reaction time, and reducing the impact force when a collision occurs; The vehicle drive motor is a hub-type brushless DC motor, and the matching motor controller receives the torque request instruction through the communication bus; Reads the original torque request value output by the current throttle throttle, and sets the limited torque request value to a preset ratio of the original torque request value; The ratio is determined so that the drive wheel can still maintain the vehicle's constant speed on a flat road, but cannot provide enough driving force to continue accelerating towards the obstacle; At the same time, a lower limit protection is set. If the limited torque request value calculated according to this ratio is lower than the minimum torque value corresponding to the motor's no-load loss, the limited torque request value is forcibly set to the minimum torque value to prevent the motor controller from entering the protection mode due to the torque request being too low; Step S640: If the instruction type code includes a braking instruction, a braking instruction is sent to the vehicle braking system to establish a braking torque on the vehicle wheels, thereby decelerating or stopping the vehicle and fundamentally preventing collision accidents. The braking system includes front wheel disc brakes and rear wheel drum brakes, each equipped with an electromagnet-driven brake cable actuator. When the electromagnet is energized, it pulls the brake cable, which drives the brake arm to rotate. The brake arm pushes the brake pads against the brake disc or brake shoes to tighten the brake drum. The electromagnet drive current is fixedly set to a preset ratio of the rated operating current. This ratio is determined based on the principle that the braking system generates a perceptible but insufficient braking torque to cause wheel lock-up. Step S650: Handle resource conflicts when multiple commands exist simultaneously; when alarm commands and torque limiting commands exist simultaneously, the two commands are executed independently in parallel; when alarm commands and braking commands exist simultaneously, the two commands are executed independently in parallel, but during the execution of the braking command, the buzzer drive duty cycle and LED indicator flashing frequency in the alarm command are increased to a second level higher than the preset fixed value; when torque limiting commands and braking commands exist simultaneously, the two commands are executed independently in parallel, but during the execution of the braking command, the torque limit request value in the torque limiting command is further reduced from the preset ratio to a smaller ratio; when alarm commands, torque limiting commands, and braking commands exist simultaneously, the above-mentioned alarm increase and torque limit decrease adjustments are applied simultaneously.

[0041] In this embodiment, all parameter values ​​are pre-calibrated and solidified based on physical principles through bench testing; the three commands independently correspond to their respective actuators, and only necessary adjustments are made at points of physical resource conflict; when braking and alarm occur simultaneously, the alarm intensity is increased to overcome the increase in the rider's perception threshold during braking; when braking and torque limiting occur simultaneously, the torque limiting ratio is further reduced to adapt to the physical characteristics of reduced vertical load on the rear wheel and decreased upper limit of the driving force that can be carried during braking.

[0042] Based on the same inventive concept as the road obstacle detection method in the foregoing embodiments, the present invention also provides a road obstacle detection system, such as... Figure 2 As shown, the system includes: The attitude acquisition module is used to acquire the vehicle's pitch angle, roll angle, and wheel speed difference. After fusion processing, a boundary envelope is generated. The boundary envelope has morphological parameters that dynamically change with the pitch angle, roll angle, and wheel speed difference. The detection fusion module is used to project a detection beam forward, collect reflected echo signals within the area defined by the boundary envelope, and perform time-frequency domain fusion processing on them to generate the situation spectrum of the current frame. The noise correction module is used to perform inter-frame temporal analysis on the situation spectrum, separate the signal noise components caused by driving environment excitation, and determine the environment correction coefficients based on the statistical characteristics of the signal noise components. The feature extraction module extracts signal features reflecting the relative motion trend of the target object from the situation spectrum after separating the signal and noise components, and constructs a collision approach index based on the signal features; The decision control module combines the morphological parameters of the boundary envelope, the environmental correction coefficient, and the collision proximity index with the preset decision control logic to output the expected risk avoidance operation command. The execution module is used to execute the expected risk avoidance operation instructions, triggering active intervention on the vehicle.

[0043] The system described above in this invention can effectively implement a road obstacle detection method, and the technical effects it can achieve are as described in the above embodiments, which will not be repeated here.

[0044] It is understood that electric vehicles, including those with the obstacle detection system described in the above embodiments, possess all the advantages and technical effects of the obstacle detection system, which will not be elaborated here.

[0045] The obstacle detection method described in the above embodiments can be used for different types of electric vehicles. Compared with electric vehicles using existing obstacle detection methods, this embodiment can combine collision proximity indicators with multi-parameter collaborative decision-making to achieve accurate obstacle identification and graded active intervention, improve the accuracy of obstacle detection and the adaptability of active risk avoidance for two-wheeled electric vehicles, and ensure riding safety.

[0046] Although this application has been described in conjunction with specific features and embodiments, it is apparent that various modifications and combinations can be made thereto without departing from the spirit and scope of this application. Accordingly, this specification and accompanying drawings are merely exemplary illustrations of the application as defined herein, and are to be considered as covering any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Thus, if such modifications and modifications fall within the scope of this application and its equivalents, this application is intended to include such modifications and modifications.

Claims

1. A method for detecting road obstructions, characterized in that, include: The vehicle's pitch angle, roll angle, and wheel speed difference are obtained and fused to generate a boundary envelope. The boundary envelope has morphological parameters that dynamically change with the pitch angle, roll angle, and wheel speed difference. Project a detection beam forward, collect the reflected echo signal within the area defined by the boundary envelope, and perform time-frequency domain fusion processing on it to generate the situation spectrum of the current frame; Inter-frame temporal analysis is performed on the situation spectrum to separate the signal and noise components caused by driving environment excitation, and environmental correction coefficients are determined based on the statistical characteristics of the signal and noise components. From the situation spectrum after separating the signal and noise components, extract signal features that reflect the relative motion trend of the target object, and construct a collision approach index based on the signal features; The morphological parameters of the boundary envelope, the environmental correction coefficient, and the collision proximity index are combined and applied to a preset decision control logic to output the expected risk avoidance operation command. Execute the expected risk avoidance operation command to trigger active intervention on the vehicle.

2. The obstacle detection method according to claim 1, characterized in that, The boundary envelope is generated after fusion processing, including: The pitch angle, roll angle, and wheel speed difference are coupled by dynamic parameters. Based on the coupling result, a three-dimensional virtual boundary is constructed with the vehicle body center of mass as the origin and dynamically changes with the pitch angle and roll angle, which serves as the boundary envelope. The lateral extension range of the boundary envelope narrows as the roll angle increases, the longitudinal extension range narrows as the pitch angle increases, and the overall coverage area of ​​the boundary envelope shrinks when the wheel speed difference exceeds a preset slip threshold.

3. The obstacle detection method according to claim 1, characterized in that, The detection beam includes a millimeter-wave radar beam and an infrared structured light beam; The reflected echo signal is fused in the time and frequency domain to generate the situation spectrum of the current frame, including: aligning the timestamps and fusing the echo signals of the millimeter-wave radar beam and the infrared structured light beam to form a composite data matrix including distance information and three-dimensional contour information, which serves as the situation spectrum of the current frame.

4. The obstacle detection method according to claim 1, characterized in that, Inter-frame temporal analysis is performed on the aforementioned situation spectrum to separate the signal-noise components caused by driving environment excitations, including: Take N consecutive frames of the situation spectrum and extract the time series signal at the corresponding spatial location of each frame; The time series signal is separated to obtain low-frequency components and high-frequency components, and the high-frequency components are regarded as signal noise components caused by the driving environment excitation.

5. The obstacle detection method according to claim 4, characterized in that, The environmental correction coefficient is determined based on the statistical characteristics of the signal and noise components, including: Calculate the root mean square value of the signal-noise component; The corresponding environmental correction coefficient is determined based on the preset numerical range to which the root mean square value belongs. The larger the root mean square value, the smaller the value of the environmental correction coefficient, indicating that the current driving environment has a higher degree of interference with the sensor signal.

6. The obstacle detection method according to claim 1, characterized in that, From the situation spectrum after separating the signal and noise components, signal features reflecting the relative motion trend of the target object are extracted, and a collision approach index is constructed based on the signal features, including: For the area located within a preset angle range directly in front of the vehicle in the situation spectrum after separating the signal and noise components, time-frequency transformation processing is performed to obtain a range Doppler spectrum. Extract the energy distribution of the preset target approach frequency band from the distance Doppler spectrum, calculate the energy attenuation coefficient based on the energy distribution, and use the energy attenuation coefficient as the collision approach index after normalization.

7. The obstacle detection method according to claim 1, characterized in that, The morphological parameters of the boundary envelope, the environmental correction coefficient, and the collision proximity index are applied together to a preset decision control logic to output the expected risk avoidance operation command, including: The coverage volume and asymmetry coefficient of the boundary envelope are used as the morphological parameters, and are input into the decision control logic along with the environmental correction coefficient and the collision proximity index. The decision control logic is configured to: increase the response weight of the collision approach index in the decision-making process when the environmental correction coefficient indicates that the interference of the driving environment on the sensor signal exceeds a preset interference threshold; and decrease the output weight of steering avoidance commands and increase the output weight of braking commands when the morphological parameters of the boundary envelope indicate that the vehicle stability is lower than a preset stability threshold.

8. The obstacle detection method according to any one of claims 1-7, characterized in that, Executing the expected risk avoidance maneuver command triggers active intervention on the vehicle, including at least one of the following: Trigger the vehicle's audible and visual alarm device to output an alarm signal; Send a torque limiting command to the vehicle motor controller to reduce the output power of the vehicle drive motor; Send braking commands to the vehicle's braking system to establish braking torque on the vehicle's wheels.

9. A road obstacle detection system, characterized in that, include: The attitude acquisition module is used to acquire the vehicle's pitch angle, roll angle, and wheel speed difference, and generate a boundary envelope through fusion processing. The boundary envelope has morphological parameters that dynamically change with the pitch angle, roll angle, and wheel speed difference. The detection fusion module is used to project a detection beam forward, collect the reflected echo signals within the area defined by the boundary envelope, and perform time-frequency domain fusion processing on them to generate the situation spectrum of the current frame. The noise correction module is used to perform inter-frame temporal calculations on the situation spectrum, separate the signal noise components caused by driving environment excitation, and determine the environment correction coefficients based on the statistical characteristics of the signal noise components. The feature extraction module extracts signal features reflecting the relative motion trend of the target object from the situation spectrum after separating the signal and noise components, and constructs a collision approach index based on the signal features; The decision control module combines the morphological parameters of the boundary envelope, the environmental correction coefficient, and the collision proximity index with a preset decision control logic to output the expected risk avoidance operation command. The execution module is used to execute the expected risk avoidance operation command and trigger active intervention on the vehicle.

10. An electric vehicle, characterized in that, The obstacle detection method according to any one of claims 1 to 8 shall be adopted.

Images