Vehicle close-range blind area anti-collision alarm system

Abstract

The invention discloses a vehicle close-range blind area anti-collision alarm system, and particularly relates to the field of vehicle driving installation assistance, which comprises a multi-mode sensing unit, a space-time fusion processing unit, a self-adaptive decision unit, a multi-channel interaction unit, an execution linkage interface unit and a dynamic blind area risk situation assessment and intention prediction unit. According to the invention, through innovative multi-mode sensing unit combination and hierarchical processing architecture, a full blind area of 0.2-10 meters behind the side of a vehicle can be accurately covered, and high-reliability target identification and classification can be realized in a complex environment and an extreme scene; in combination with intuitive multi-channel interactive design, non-interference directional risk prompts are provided for a driver; and through a limited execution linkage mechanism, active safety bottom support guarantee is provided at a key danger moment, a protection system from perception to prevention is finally constructed, and blind area collision anxiety is reduced.

Description

Technical Field

[0001] The present invention relates to the field of vehicle driving assistance, and more particularly to a vehicle close-range blind spot anti-collision alarm system. Background Art

[0002] As urban traffic density continues to increase, vehicle near-field blind spot collision accidents are on the rise. According to statistics, about 23% of side collisions and lane change accidents in low-speed scenarios are caused by the driver's lack of perception of moving targets in the blind spot.

[0003] Existing technologies for avoiding collisions in close-range blind spots generally require drivers to walk around the vehicle before driving, much like taking a driver's license test. However, many drivers fail to maintain this habit after obtaining their licenses. This provides a research and development direction for automakers in the market to develop intelligent collision avoidance warning systems. Although some high-end models have introduced sensor fusion technology, traditional ultrasonic radars have detection blind spots at distances less than 0.5m, millimeter-wave radars have insufficient close-range clutter suppression, and cameras have a high risk of failure at night / in backlight. Single-sensor alarm strategies cannot adapt to complex scenarios (such as accidental triggering in rainy days and false alarms in narrow parking spaces), and lack real-time coordination with vehicle dynamics (steering intentions, gear position). Summary of the Invention

[0004] In order to overcome the above-mentioned defects of the prior art, the present invention provides a vehicle close-range blind spot anti-collision alarm system.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a vehicle close-range blind spot collision avoidance warning system, comprising a multimodal perception unit, a spatiotemporal fusion processing unit, an adaptive decision-making unit, a multi-channel interaction unit, an execution linkage interface unit, and a dynamic blind spot risk situation assessment and intention prediction unit, wherein:

[0006] The multimodal sensing unit is a short-range, wide-angle millimeter-wave radar array, a sector-focused ultrasonic sensor array, and an ultra-wide-angle camera module integrated on the side and rear of the vehicle;

[0007] The input end of the spatiotemporal fusion processing unit is connected to the multimodal perception unit, and the output end generates a blind spot target list with confidence;

[0008] The input end of the dynamic blind spot risk situation assessment and intention prediction unit is connected to the output end of the spatiotemporal fusion processing unit and the vehicle navigation / high-precision positioning unit (such as access to HD Map data) and the V2X communication unit (such as access to surrounding vehicle / facility information). The output end generates a situation assessment report including the blind spot target predicted trajectory, collision time estimation (TTC), target behavior intention probability and comprehensive risk rating;

[0009] The input end of the adaptive decision unit receives the situation assessment report, the blind spot target list and the vehicle CAN bus signal, and the output end generates a graded warning instruction;

[0010] The multi-channel interaction unit includes a spatialized sound field generator, a steering wheel tactile feedback device and an AR-HUD projector, and the input end is connected to the adaptive decision unit;

[0011] The execution linkage interface unit establishes two-way communication with the vehicle AEB controller, ESC controller and EPS controller;

[0012] The dynamic blind spot risk situation assessment and intention prediction unit includes:

[0013] a. Target trajectory predictor: This uses a kinematic model or machine learning model to predict the target's short-term (e.g., 3-5 seconds) trajectory based on the target's historical state (position, velocity, acceleration), the vehicle's own state, and environmental constraints (lane lines, curbs).

[0014] b. Intent Recognizer: This combines target type (from the target classification unit), trajectory prediction, vehicle signals (such as turn signals), V2X information (such as the intentions of surrounding vehicles), and a traffic rules knowledge base to predict the target's behavioral intention (such as cutting in, crossing, or staying).

[0015] c. Collision Risk Calculator: This calculates a dynamic collision risk value by integrating factors such as the target predicted trajectory, the vehicle predicted trajectory (based on vehicle status and driver input), time to collision (TTC), target type, and intent probability.

[0016] d. Situation Comprehensive Evaluator: Integrates the above information to generate a situation report that includes risk hotspot areas, key threat target identification, and predicted collision points.

[0017] As a further improvement to the technical solution of the present invention, the operating frequency band of the short-range, wide-angle millimeter-wave radar array is 77-81 GHz, the detection angle covers a ±150° sector area, and the minimum detection distance is ≤0.2 m; the beam angle of the sector-focused ultrasonic sensor array is ≤30°, and the longitudinal resolution is ≤5 cm; the ultra-wide-angle camera module supports 120 dB HDR imaging and 940 nm infrared night vision dual modes.

[0018] As a further improvement of the technical solution of the present invention, the spatiotemporal fusion processing unit includes a multi-sensor clock synchronization module, a dynamic calibration compensation module and an environment-adaptive weight distributor, wherein the multi-sensor clock synchronization module aligns the timestamps of each sensor based on the PTP protocol; the input end of the dynamic calibration compensation module is connected to the vehicle IMU and wheel speed sensor, and the output end corrects the coordinate system offset; the environment-adaptive weight distributor has a built-in light / rainfall sensor interface.

[0019] As a further improvement of the technical solution of the present invention, the adaptive decision unit includes a scene recognition engine, a driver status input interface and a collision point predictor, wherein the scene recognition engine is used to receive turn signal signals, gear signals and steering wheel angles; the driver status input interface is connected to the DMS camera in the cabin; and the output end of the collision point predictor is connected to the AR-HUD projector.

[0020] As a further improvement of the technical solution of the present invention, the multi-channel interaction unit includes a sound field positioning controller, a tactile encoder and an AR graphics generator, wherein the sound field positioning controller is used to drive the multi-channel vehicle audio to generate directional prompt sounds; the tactile encoder is used to independently control the vibration motors on the left and right sides of the steering wheel; and the AR graphics generator is used to project a dangerous area light strip aligned with the real space position on the windshield.

[0021] As a further improvement of the technical solution of the present invention, the execution linkage interface unit includes a limited braking command interface and a steering intervention command interface, wherein the limited braking command interface is used to output a braking request with a deceleration of ≤0.3g to the AEB controller; the steering intervention command interface is used to output a micro-torque request of ≤2Nm to the EPS controller.

[0022] As a further improvement of the technical solution of the present invention, it also includes an anti-interference unit, which is composed of a millimeter-wave radar frequency hopping controller and an ultrasonic waveform analyzer, wherein the millimeter-wave radar frequency hopping controller supports 77-81GHz pseudo-random frequency switching; the input end of the ultrasonic waveform analyzer is connected to the original echo signal of the ultrasonic sensor.

[0023] As a further improvement of the technical solution of the present invention, it also includes a target classification unit, which includes a metal characteristic identifier and a micro-motion feature extractor, wherein the input end of the metal characteristic identifier is connected to the ultrasonic resonant frequency analysis circuit; the input end of the micro-motion feature extractor is connected to the millimeter wave radar micro-Doppler spectrum data.

[0024] As a further improvement of the technical solution of the present invention, it also includes a degradation fault-tolerant unit, which includes a sensor failure detector and a virtual blind spot reconstruction module, wherein the sensor failure detector is used to monitor the validity of each sensor data; the input end of the virtual blind spot reconstruction module is connected to the vehicle wheel speed and steering angle sensor, and the virtual blind spot reconstruction module is used to simulate blind spot risks through vehicle dynamic data when the perception ability is reduced due to sensor failure or environmental interference.

[0025] As a further improvement of the technical solution of the present invention, the millimeter-wave radars in the multimodal perception unit are installed on both sides of the rear of the vehicle, the ultrasonic sensors are arranged at equal intervals along the side skirts, and the camera module is fixed to the bottom of the exterior rearview mirror housing; the execution linkage interface unit communicates with the AEB / ESC / EPS controller through the vehicle Ethernet bus.

[0026] Beneficial effects of the present invention:

[0027] Through an innovative combination of multimodal perception units and a layered processing architecture, combined with a newly added dynamic blind spot risk situation assessment and intention prediction unit, this invention achieves high-precision, intelligent coverage of the entire blind spot zone (0.2-10 meters) to the side and rear of the vehicle. This not only enables highly reliable target recognition and classification in complex environments and extreme scenarios, but also deeply analyzes the dynamic situation in the blind spot, predicts the trajectory and intention of potential collision objects, and assesses forward-looking collision risks. Based on this more comprehensive situational information, the adaptive decision-making unit combines real-time vehicle dynamics and driver status to achieve more accurate and timely scenario-based risk classification. An intuitive multi-channel interaction unit provides the driver with non-interference directional guidance based on risk situation. A restricted execution linkage interface unit provides active safety protection in critical moments of danger. Furthermore, an anti-interference unit, a target classification unit, and a degradation fault tolerance unit ensure robustness. In summary, this invention establishes a closed-loop protection system from "environmental perception - situation understanding - risk prediction - intelligent decision-making - intuitive interaction - controlled execution." This significantly surpasses the static or simple dynamic perception of blind spots in existing technologies, effectively reducing drivers' blind spot collision anxiety and significantly improving the intelligence level and proactive prevention capabilities of vehicle near-field safety protection. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 This is a system framework diagram of the present invention.

[0029] The accompanying drawings are marked as follows: 1. Multimodal perception unit; 2. Spatiotemporal fusion processing unit; 3. Adaptive decision-making unit; 4. Multi-channel interaction unit; 5. Execution linkage interface unit; 6. Dynamic blind spot risk situation assessment and intention prediction unit. DETAILED DESCRIPTION

[0030] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0031] As attached Figure 1The vehicle close-range blind spot collision avoidance warning system shown in the figure includes a multimodal perception unit 1, a spatiotemporal fusion processing unit 2, an adaptive decision unit 3, a multi-channel interaction unit 4, an execution linkage interface unit 5, and a dynamic blind spot risk situation assessment and intention prediction unit 6, wherein:

[0032] The multimodal sensing unit 1 is a short-range, wide-angle millimeter-wave radar array, a sector-focused ultrasonic sensor array, and an ultra-wide-angle camera module integrated on the side and rear of the vehicle.

[0033] The input end of the spatiotemporal fusion processing unit 2 is connected to the multimodal perception unit 1, and the output end generates a blind spot target list with confidence;

[0034] The input end of the dynamic blind spot risk situation assessment and intention prediction unit 6 is connected to the output end of the spatiotemporal fusion processing unit 2 and the vehicle navigation / high-precision positioning unit (such as access to HD Map data) and the V2X communication unit (such as access to surrounding vehicle / facility information). The output end generates a situation assessment report including the blind spot target predicted trajectory, collision time estimation (TTC), target behavior intention probability and comprehensive risk rating;

[0035] The input end of the adaptive decision unit 3 receives the situation assessment report, the blind spot target list and the vehicle CAN bus signal, and the output end generates a graded warning instruction;

[0036] The multi-channel interaction unit 4 includes a spatialized sound field generator, a steering wheel tactile feedback device and an AR-HUD projector, and its input end is connected to the adaptive decision unit 3;

[0037] The execution linkage interface unit 5 establishes two-way communication with the vehicle's AEB controller, ESC controller, and EPS controller;

[0038] The dynamic blind spot risk situation assessment and intention prediction unit 6 includes:

[0039] a. Target trajectory predictor: This uses a kinematic model or machine learning model to predict the target's short-term (e.g., 3-5 seconds) trajectory based on the target's historical state (position, velocity, acceleration), the vehicle's own state, and environmental constraints (lane lines, curbs).

[0040] b. Intent Recognizer: This combines target type (from the target classification unit), trajectory prediction, vehicle signals (such as turn signals), V2X information (such as the intentions of surrounding vehicles), and a traffic rules knowledge base to predict the target's behavioral intention (such as cutting in, crossing, or staying).

[0041] c. Collision Risk Calculator: This calculates a dynamic collision risk value by integrating factors such as the target predicted trajectory, the vehicle predicted trajectory (based on vehicle status and driver input), time to collision (TTC), target type, and intent probability.

[0042] d. Situation Comprehensive Evaluator: Integrates the above information to generate a situation report that includes risk hotspot areas, key threat target identification, and predicted collision points.

[0043] Preferably, the operating frequency band of the short-range, wide-angle millimeter-wave radar array is 77-81GHz, the detection angle covers a ±150° sector area, and the minimum detection distance is ≤0.2m; the beam angle of the sector-focused ultrasonic sensor array is ≤30°, and the vertical resolution is ≤5cm; the ultra-wide-angle camera module supports 120dB HDR imaging and 940nm infrared night vision dual modes.

[0044] Preferably, the spatiotemporal fusion processing unit 2 includes a multi-sensor clock synchronization module, a dynamic calibration compensation module and an environment-adaptive weight distributor, wherein the multi-sensor clock synchronization module aligns the timestamps of each sensor based on the PTP protocol; the input end of the dynamic calibration compensation module is connected to the vehicle IMU and wheel speed sensor, and the output end corrects the coordinate system offset; the environment-adaptive weight distributor has a built-in light / rainfall sensor interface.

[0045] Preferably, the adaptive decision unit 3 includes a scene recognition engine, a driver status input interface and a collision point predictor, wherein the scene recognition engine is used to receive turn signal signals, gear signals and steering wheel angles; the driver status input interface is connected to the DMS camera in the cabin; and the output end of the collision point predictor is connected to the AR-HUD projector.

[0046] Preferably, the multi-channel interaction unit 4 includes a sound field positioning controller, a tactile encoder and an AR graphics generator, wherein the sound field positioning controller is used to drive the multi-channel vehicle audio to generate directional prompt sounds; the tactile encoder is used to independently control the vibration motors on the left and right sides of the steering wheel; and the AR graphics generator is used to project a dangerous area light strip aligned with the real space position on the windshield.

[0047] Preferably, the execution linkage interface unit 5 includes a limited braking command interface and a steering intervention command interface, wherein the limited braking command interface is used to output a braking request with a deceleration of ≤0.3g to the AEB controller; the steering intervention command interface is used to output a micro-torque request of ≤2Nm to the EPS controller.

[0048] Preferably, the vehicle's close-range blind spot collision avoidance alarm system also includes an anti-interference unit, which is composed of a millimeter-wave radar frequency hopping controller and an ultrasonic waveform analyzer, wherein the millimeter-wave radar frequency hopping controller supports 77-81GHz pseudo-random frequency switching; the input end of the ultrasonic waveform analyzer is connected to the original echo signal of the ultrasonic sensor.

[0049] Preferably, the vehicle close-range blind spot collision avoidance alarm system also includes a target classification unit, which includes a metal property identifier and a micro-motion feature extractor, wherein the input end of the metal property identifier is connected to the ultrasonic resonant frequency analysis circuit; the input end of the micro-motion feature extractor is connected to the millimeter wave radar micro-Doppler spectrum data.

[0050] Preferably, the vehicle's close-range blind spot collision avoidance alarm system also includes a degradation fault-tolerant unit, which includes a sensor failure detector and a virtual blind spot reconstruction module, wherein the sensor failure detector is used to monitor the validity of each sensor data; the input end of the virtual blind spot reconstruction module is connected to the vehicle wheel speed and steering angle sensor, and the virtual blind spot reconstruction module is used to simulate blind spot risks through vehicle dynamic data when the sensor fails or the environmental interference causes the perception ability to decline.

[0051] Preferably, the millimeter-wave radars in the multimodal sensing unit 1 are installed on both sides of the rear of the vehicle, the ultrasonic sensors are arranged at equal intervals along the side skirts, and the camera module is fixed to the bottom of the exterior rearview mirror housing; the execution linkage interface unit 5 communicates with the AEB / ESC / EPS controller through the vehicle Ethernet bus.

[0052] Working principle: The present invention designs a vehicle close-range blind spot collision avoidance warning system, which consists of a multimodal perception unit 1, a spatiotemporal fusion processing unit 2, an adaptive decision unit 3, a multi-channel interaction unit 4 and an execution linkage interface unit 5. It also includes an anti-interference unit, a target classification unit and a degradation fault tolerance unit. The multimodal perception unit 1 (millimeter wave radar, ultrasound, camera) is used to achieve centimeter-level high-precision environmental detection, and the spatiotemporal fusion processing unit 2 is used to perform synchronous fusion and dynamic compensation of multi-source data to generate a reliable target list; the adaptive decision unit 3 combines vehicle dynamics and driver status to achieve scenario-based risk classification; the multi-channel interaction unit 4 provides spatialized sound field, tactile and AR- Multi-dimensional precise warnings such as HUD projection; the execution linkage interface unit 5 supports controlled collaborative intervention with the vehicle's braking / steering system, and is equipped with anti-interference, target classification and degradation fault tolerance mechanisms, ultimately achieving all-weather, highly robust perception of blind spot targets, intelligent graded warnings and limited active safety control, significantly improving the vehicle's near-field safety protection capabilities; the dynamic blind spot risk situation assessment and intention prediction unit 6 achieves high-precision, intelligent coverage of the entire blind spot of 0.2-10 meters to the rear of the vehicle, not only achieving highly reliable target recognition and classification in complex environments and extreme scenarios, but also deeply analyzing the dynamic situation of the blind spot, predicting the trajectory and intention of potential collision objects, and evaluating forward-looking collision risks.

[0053] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A vehicle close-range blind spot anti-collision alarm system, characterized by: The system comprises a multimodal perception unit (1), a spatiotemporal fusion processing unit (2), an adaptive decision-making unit (3), a multi-channel interaction unit (4), an execution linkage interface unit (5), and a dynamic blind spot risk situation assessment and intention prediction unit (6), wherein: The multimodal sensing unit (1) is a short-range wide-angle millimeter-wave radar array, a sector-shaped focused ultrasonic sensor array, and an ultra-wide-angle camera module integrated on the side and rear of the vehicle; The input end of the spatiotemporal fusion processing unit (2) is connected to the multimodal perception unit (1), and the output end generates a blind spot target list with confidence; The input end of the dynamic blind spot risk situation assessment and intention prediction unit (6) is connected to the output end of the spatiotemporal fusion processing unit (2) and the vehicle navigation / high-precision positioning unit and the V2X communication unit, and the output end generates a situation assessment report including the blind spot target prediction trajectory, collision time estimation, target behavior intention probability and comprehensive risk rating; The input end of the adaptive decision unit (3) receives the situation assessment report, the blind spot target list and the vehicle CAN bus signal, and the output end generates a graded warning instruction; The multi-channel interaction unit (4) includes a spatialized sound field generator, a steering wheel tactile feedback device and an AR-HUD projector, and an input end is connected to the adaptive decision unit (3); The execution linkage interface unit (5) establishes two-way communication with the vehicle AEB controller, ESC controller and EPS controller; The dynamic blind spot risk situation assessment and intention prediction unit (6) includes: a. Target trajectory predictor: This predicts the target's future short-term trajectory using a kinematic model or machine learning model based on the target's historical state, the vehicle's own state, and environmental constraints. b. Intent Recognizer: This combines target type, trajectory prediction, vehicle signals, V2X information, and a knowledge base of traffic rules to predict target behavior intentions. c. Collision Risk Calculator: This calculates the dynamic collision risk value by integrating the target predicted trajectory, the vehicle predicted trajectory, collision time, target type, and intention probability factors. d. Situation Comprehensive Evaluator: Integrates the above information to generate a situation report that includes risk hotspot areas, key threat target identification, and predicted collision points.

2. The vehicle close-range blind spot collision avoidance warning system according to claim 1, characterized in that: The operating frequency band of the short-range, wide-angle millimeter-wave radar array is 77-81 GHz, the detection angle covers a ±150° sector area, and the minimum detection distance is ≤0.2m; the beam angle of the sector-focused ultrasonic sensor array is ≤30°, and the vertical resolution is ≤5cm; the ultra-wide-angle camera module supports 120dB HDR imaging and 940nm infrared night vision dual modes.

3. The vehicle close-range blind spot collision avoidance warning system according to claim 1, characterized in that: The spatiotemporal fusion processing unit (2) comprises a multi-sensor clock synchronization module, a dynamic calibration compensation module and an environment adaptive weight distributor, wherein the multi-sensor clock synchronization module aligns the timestamps of each sensor based on the PTP protocol; the input end of the dynamic calibration compensation module is connected to the vehicle IMU and the wheel speed sensor, and the output end corrects the coordinate system offset; the environment adaptive weight distributor has a built-in light / rainfall sensor interface.

4. The vehicle close-range blind spot collision avoidance warning system according to claim 1, characterized in that: The adaptive decision unit (3) comprises a scene recognition engine, a driver state input interface and a collision point predictor, wherein the scene recognition engine is used to receive a turn signal, a gear signal and a steering wheel angle; the driver state input interface is connected to a DMS camera in the cabin; and the output end of the collision point predictor is connected to an AR-HUD projector.

5. The vehicle close-range blind spot collision avoidance warning system according to claim 1, characterized in that: The multi-channel interaction unit (4) comprises a sound field positioning controller, a tactile encoder and an AR graphic generator, wherein the sound field positioning controller is used to drive the multi-channel vehicle-mounted audio system to generate directional prompt sounds; the tactile encoder is used to independently control the vibration motors on the left and right sides of the steering wheel; and the AR graphic generator is used to project a dangerous area light strip aligned with the real space position on the windshield.

6. The vehicle close-range blind spot collision avoidance warning system according to claim 1, characterized in that: The execution linkage interface unit (5) comprises a limited braking command interface and a steering intervention command interface, wherein the limited braking command interface is used to output a braking request with a deceleration of ≤0.3g to the AEB controller; and the steering intervention command interface is used to output a micro-torque request of ≤2Nm to the EPS controller.

7. The vehicle close-range blind spot collision avoidance warning system according to claim 1, characterized in that: It also includes an anti-interference unit, which is composed of a millimeter-wave radar frequency hopping controller and an ultrasonic waveform analyzer. The millimeter-wave radar frequency hopping controller supports 77-81GHz pseudo-random frequency switching; the input end of the ultrasonic waveform analyzer is connected to the original echo signal of the ultrasonic sensor.

8. The vehicle close-range blind spot collision avoidance warning system according to claim 1, characterized in that: It also includes a target classification unit, which includes a metal characteristic identifier and a micro-motion feature extractor, wherein the input end of the metal characteristic identifier is connected to the ultrasonic resonance frequency analysis circuit; the input end of the micro-motion feature extractor is connected to the millimeter wave radar micro-Doppler spectrum data.

9. The vehicle close-range blind spot collision avoidance warning system according to claim 1, characterized in that: It also includes a degradation fault-tolerant unit, which includes a sensor failure detector and a virtual blind spot reconstruction module, wherein the sensor failure detector is used to monitor the validity of each sensor data; the input end of the virtual blind spot reconstruction module is connected to the vehicle wheel speed and steering angle sensor, and the virtual blind spot reconstruction module is used to simulate blind spot risks through vehicle dynamic data when the sensor failure or environmental interference causes the perception ability to decline.

10. The vehicle close-range blind spot collision avoidance warning system according to claim 1, characterized in that: The millimeter wave radars in the multimodal sensing unit (1) are installed on both sides of the rear of the vehicle, the ultrasonic sensors are arranged at equal intervals along the side skirts, and the camera module is fixed to the bottom of the exterior rearview mirror housing; the execution linkage interface unit (5) communicates with the AEB / ESC / EPS controller via the vehicle Ethernet bus.

Images