CN-122028272-A - Multi-source vehicle state sensing method, system, equipment and medium for tunnel vehicle-mounted linkage illumination

CN122028272ACN 122028272 ACN122028272 ACN 122028272ACN-122028272-A

Abstract

The invention relates to the technical field of signal analysis and discloses a method, a system, equipment and a medium for multi-source sensing vehicle state of tunnel vehicle-mounted linkage illumination, wherein the method comprises the steps of uniformly calibrating a multi-source sensor and evolving the vehicle state into a discrete time linear state space model; the method comprises the steps of converting vehicle motion parameters and identity characteristic information into standardized observation vectors, carrying out state prediction based on a vehicle kinematic model to generate continuous motion tracks, respectively calculating joint likelihood of each sensor output and each vehicle state-identity hypothesis for a plurality of observation targets, carrying out Bayesian inference to obtain associated posterior probability of observation-vehicle, fusing motion and identity consistency, and calculating and generating linkage illumination control instructions of current and predicted coverage areas of the vehicle in real time according to a spatial mapping relation between the vehicle and lamps. The invention can effectively inhibit cross-equipment error association and identity drift, realize lighting of the vehicle running light, darkness of the vehicle running light and running of the light along with the vehicle, and reduce ineffective illumination energy consumption.

Inventors

WEI LIXING
HUANG JINHAI
LI ZHENGJIE
WANG DONGYUN

Assignees

桂林信息科技学院

Dates

Publication Date: 20260512
Application Date: 20260113

Claims (10)

1. A method for multi-source perception of vehicle status for tunnel onboard linkage illumination, comprising: Unified space-time calibration is carried out on the tunnel entrance and the sensors in the tunnel, and the vehicle state is evolved into a discrete time linear state space model; Based on the linear state space model, converting the vehicle motion parameters and identity characteristic information acquired by the multi-source sensor into standardized observation vectors, and carrying out state prediction based on the vehicle kinematic model so as to generate a continuous motion track; based on the motion trail, calculating the joint likelihood of each sensor output and each vehicle state-identity hypothesis and carrying out Bayesian inference on a plurality of observation targets respectively to obtain the associated posterior probability of the observation-vehicle, and fusing motion and identity consistency to realize joint updating of the probability relay tracking of multiple sensors and the vehicle state-identity; Based on the continuous position and the identity stability estimation of the vehicle obtained after the joint updating, the linkage illumination control instruction of the current and the predicted coverage area of the vehicle is calculated and generated in real time according to the space mapping relation between the vehicle and the lamp.
2. The method for multi-source perception of vehicle state for tunnel on-vehicle linkage illumination of claim 1, wherein unified space-time calibration is performed on sensors at tunnel entrance and inside the tunnel, and vehicle state is evolved into a discrete time linear state space model, comprising: establishing a unified one-dimensional arc length coordinate system in the longitudinal direction of the tunnel; Defining the accumulated running distance of the vehicle along the real running path of the tunnel based on the one-dimensional arc length coordinate system, and equivalently expanding the nonlinear tunnel into a linear section; under the one-dimensional arc length coordinate system, the tunnel entrance is taken as a zero point, and the motion state of the vehicle at any moment is defined as a motion state vector consisting of longitudinal position, speed and acceleration; Based on a preset sampling period and vehicle kinematics assumption, a discrete time linear state space equation describing the evolution of the motion state vector along with time is established.
3. The method for multi-source sensing vehicle state of tunnel onboard linkage illumination according to claim 2, wherein based on the linear state space model, converting vehicle motion parameters and identity feature information acquired by a multi-source sensor into standardized observation vectors comprises: An entrance vision system is arranged at an entrance of the tunnel, and radars are arranged in the tunnel along the driving direction; The entrance vision system and each radar locally observe the vehicle state; performing recursive estimation on the motion state of the vehicle by adopting Kalman filtering, and optimally fusing the state prediction at the last moment with the current radar observation to obtain continuous position, speed and acceleration estimation of the vehicle in the whole tunnel range; and combining the estimated values of the entrance vision system and radar observation to obtain the integral standardized observation vector of the sensor.
4. A method for multi-source perception of vehicle status for tunnel-mounted linked lighting as claimed in claim 3, wherein based on said linear-state spatial model, converting vehicle motion parameters and identity information obtained by multi-source sensors into standardized observation vectors, further comprising: Defining an identity characteristic observation vector, wherein the identity characteristic observation vector specifically comprises a vehicle length characteristic output by a radar, a vehicle license plate number corresponding to a vehicle output by a vision system and a corresponding vehicle type output by the vision system; defining a joint random variable for candidate vehicles in the tunnel by combining the identity feature observation vector and the motion state vector; an association event is defined for each sensor observation that represents a probabilistic association hypothesis that the observation data originated from a particular vehicle.
5. The method for multi-source perception of vehicle status for tunnel on-vehicle coordinated illumination of claim 4, wherein for each of a plurality of observed targets output by each sensor based on said motion profile, a joint likelihood with each vehicle status-identity hypothesis is calculated, respectively, comprising: Establishing a likelihood function for the joint random variable for any sensor to the vehicle observation set, wherein the likelihood function is decomposed into a product of motion consistency likelihood and identity consistency likelihood; Wherein the motion consistency likelihood is measured based on a gaussian distribution between a vehicle state prediction value and a sensor observation value; wherein the identity likelihood is measured based on a probability of matching between the vehicle identity feature prediction value and the sensor identification value.
6. The method for multi-source perception of vehicle status for tunnel-mounted linked lighting of claim 5, wherein said performing bayesian inference to obtain an observation-vehicle associated posterior probability comprises: calculating the associated posterior probability of each sensor observation and each candidate vehicle by using a Bayesian inference formula based on the likelihood function; Fusing all associated observations from a plurality of sensors, and obtaining an optimal estimation of the motion state by maximizing the joint posterior probability distribution; and outputting a fusion result which simultaneously comprises the optimal motion state estimation and the optimal identity characteristic estimation.
7. The method for multi-source perception of vehicle state of tunnel vehicle-mounted linkage illumination according to claim 6, wherein based on the vehicle continuous position and identity stability estimation obtained after the joint update, the linkage illumination control command of the current and predicted coverage area of the vehicle is calculated and generated in real time according to the spatial mapping relation between the vehicle and the lamp, comprising: Defining an illumination interval of each lamp in the tunnel based on the central position of each lamp in the central line direction of the tunnel and the effective illumination length of the lamp along the driving direction; based on the optimal estimation of the vehicle motion state, predicting the running track of the target vehicle in a future time window by using a vehicle uniform acceleration model; Combining the predicted track and the identity characteristics of the vehicle, introducing the lighting advance and retard buffer distances, and calculating the space area required to be covered by the target vehicle for lighting in a current period and a future period; and generating an instruction for controlling the corresponding lamp to be started or stopped according to the matching result.
8. A multi-source aware vehicle state system for tunnel onboard linkage lighting, applying the method of any of claims 1-7, comprising: the linear state space model construction module is used for carrying out unified space-time calibration on the tunnel entrance and the sensors in the tunnel and evolving the vehicle state into a discrete time linear state space model; The motion track generation module is used for converting the vehicle motion parameters and the identity characteristic information acquired by the multi-source sensor into standardized observation vectors based on the linear state space model, and carrying out state prediction based on the vehicle kinematic model so as to generate a continuous motion track; the calculation module is used for respectively calculating the joint likelihood of each vehicle state-identity hypothesis and the joint likelihood of each vehicle state-identity hypothesis for a plurality of observation targets output by each sensor based on the motion trail, carrying out Bayesian inference to obtain the associated posterior probability of the observation-vehicle, fusing the motion and identity consistency, and realizing the joint update of the probability relay tracking of a plurality of sensors and the vehicle state-identity; The command generation module is used for calculating and generating linkage illumination control commands of the current and predicted coverage areas of the vehicle in real time according to the spatial mapping relation between the vehicle and the lamps based on the continuous position and the identity stability estimation of the vehicle obtained after the joint update.
9. A computer device, comprising: A memory and a processor; the memory is configured to store computer-executable instructions that, when executed by the processor, perform the steps of a method for multi-source perception of vehicle status for tunnel onboard linkage illumination as recited in any one of claims 1 to 7.
10. A computer readable storage medium storing computer executable instructions which when executed by a processor perform the steps of a method for multi-source perception of vehicle status for tunnel onboard linkage illumination according to any one of claims 1 to 7.

Description

Multi-source vehicle state sensing method, system, equipment and medium for tunnel vehicle-mounted linkage illumination Technical Field The invention relates to the technical field of vehicle state prediction, in particular to a method, a system, equipment and a medium for multi-source sensing vehicle state of tunnel vehicle-mounted linkage illumination. Background The existing tunnel illumination and vehicle perception system mostly adopts a control mode of single sensor threshold judgment, but the entrance visual recognition, the radar detection in the tunnel and illumination control decoupling design lack a unified mathematical modeling and probability inference framework. In practical engineering, lighting control is often based on fixed partition or coarse-grained vehicle detection, and once a certain section detects a vehicle, the vehicle is integrally lightened, so that the number, the position and the driving direction of the vehicle are difficult to accurately distinguish, and the phenomena that multiple vehicles are indistinguishable, multiple lamps are simultaneously lightened are common. Under the conditions of a tunnel nonlinear structure, multi-camera relay and dense traffic of vehicles, the problems of vehicle cross-equipment error association and identity drift are further amplified, so that a conservation strategy has to be adopted by an illumination system to ensure traffic safety, such as early lighting, delayed extinction or full-section constant lighting, thereby causing a great deal of unnecessary energy consumption waste. Meanwhile, the visual system of the existing method lacks of joint depiction of vehicles on a probability level, so that under the conditions of invisible license plates, abrupt illumination change or shielding and the like, the vehicle track is unstable, the illumination following is inaccurate, and the real fine and energy-saving illumination control according to the vehicle level is difficult to realize. Disclosure of Invention The present invention has been made in view of the above-described problems occurring in the prior art. Therefore, the invention provides a multi-source vehicle state sensing method and system for tunnel vehicle-mounted linkage illumination, which solve the problem that energy-saving illumination control cannot be accurately performed according to the vehicle level. In order to solve the technical problems, the invention provides the following technical scheme: In a first aspect, the present invention provides a method for sensing vehicle states by multiple sources for tunnel vehicle-mounted linkage illumination, including: Unified space-time calibration is carried out on the tunnel entrance and the sensors in the tunnel, and the vehicle state is evolved into a discrete time linear state space model; Based on the linear state space model, converting the vehicle motion parameters and identity characteristic information acquired by the multi-source sensor into standardized observation vectors, and carrying out state prediction based on the vehicle kinematic model so as to generate a continuous motion track; based on the motion trail, calculating the joint likelihood of each sensor output and each vehicle state-identity hypothesis and carrying out Bayesian inference on a plurality of observation targets respectively to obtain the associated posterior probability of the observation-vehicle, and fusing motion and identity consistency to realize joint updating of the probability relay tracking of multiple sensors and the vehicle state-identity; Based on the continuous position and the identity stability estimation of the vehicle obtained after the joint updating, the linkage illumination control instruction of the current and the predicted coverage area of the vehicle is calculated and generated in real time according to the space mapping relation between the vehicle and the lamp. As a preferable scheme of the multi-source perception vehicle state method for tunnel vehicle-mounted linkage illumination, the invention comprises the following steps of carrying out unified space-time calibration on a tunnel entrance and sensors in the tunnel, and evolving the vehicle state into a discrete time linear state space model, wherein the method comprises the following steps: establishing a unified one-dimensional arc length coordinate system in the longitudinal direction of the tunnel; Defining the accumulated running distance of the vehicle along the real running path of the tunnel based on the one-dimensional arc length coordinate system, and equivalently expanding the nonlinear tunnel into a linear section; under the one-dimensional arc length coordinate system, the tunnel entrance is taken as a zero point, and the motion state of the vehicle at any moment is defined as a motion state vector consisting of longitudinal position, speed and acceleration; Based on a preset sampling period and vehicle kinematics assumption, a discrete time linear state space