Rearward white line inference device, target recognition device, and method
An object is to provide a rearward white line inference device, a target recognition device, and a method capable of inferring a rearward white line with high accuracy. A rearward white line inference device includes: a white line detection unit that detects a white line ahead of a host vehicle; an azimuth-relative-to-white-line generation unit that obtains an azimuth of the host vehicle relative to the white line every predetermined time interval based on a detection result of the white line detection unit; and a relative azimuth inference unit that infers a relative azimuth from the azimuth of the host vehicle at a certain point based on the azimuth of the host vehicle obtained by the azimuth-relative-to-white-line generation unit.
1 . A rearward white line inference device comprising a processor configured to: detect a white line ahead of a host vehicle; obtain a default azimuth of the host vehicle relative to the white line every predetermined time interval based on a white line detection result; select: (i) the default azimuth by default; (ii) an azimuth based on a Global Navigation Satellite System (GNSS), in response to satisfying a first condition that (a) a positioning state of the GNSS is within a predetermined accuracy, and (b) a speed is above a predetermined level or a change in cumulative azimuth is below a predetermined level: (iii) an azimuth based on a group of point sequences of forward white lines, in response to satisfying a second condition that (a) the first condition is not satisfied, (b) the change in cumulative azimuth is above the predetermined level, and (c) at least one white line exists; and (iv) an azimuth based on a yaw rate, in response to satisfying a third condition that the first condition is not satisfied, (b) the change in cumulative azimuth is below the predetermined level, and (c) no white line exists; and determine a relative azimuth of the host vehicle at a certain point based on the selected azimuths of the host vehicle.
2 . The rearward white line inference device according to claim 1 , wherein the white line detection result is a function expressing white line point sequences during a predetermined cycle, and the default azimuth is obtained from positions of the white line point sequences between different time points.
3 . A target recognition device comprising a processor configured to: detect a white line ahead of a host vehicle; obtain a white-line azimuth of the host vehicle relative to the white line every predetermined time interval based on a white line detection result; select: (i) the white-line azimuth; (ii) an azimuth based on a Global Navigation Satellite System (GNSS), in response to satisfying a first condition that (a) a positioning state of the GNSS is within a predetermined accuracy, and (b) a speed is above a predetermined level or a change in cumulative azimuth is below a predetermined level: (iii) an azimuth based on a group of point sequences of forward white lines, in response to satisfying a second condition that (a) the first condition is not satisfied, (b) the change in cumulative azimuth is above the predetermined level, and (c) at least one white line exists; and (iv) an azimuth based on a yaw rate, in response to satisfying a third condition that the first condition is not satisfied, (b) the change in cumulative azimuth is below the predetermined level, and (c) no white line exists; determine a relative azimuth of the host vehicle at a certain point based on the selected azimuths of the host vehicle; detect a target behind the host vehicle; determine a moving distance of the host vehicle; determine a moving trajectory of the host vehicle relative to the white line based on the relative azimuth at every predetermined time interval and the moving distance at every predetermined time interval; determine a position of a white line outside a current detection range based on a past detection result and the moving trajectory; and identify a positional relationship between the white line outside the detection range and the target based on the position of the white line outside the detection range, and a position of the detected target.
4 . The target recognition device according to claim 3 , wherein the white line detection result is a function expressing white line point sequences that are periodically output, and the white-line azimuth is obtained from positions of the white line point sequences between different time points.
5 . The target recognition device according to claim 3 , wherein the processor is further configured to determine the moving trajectory of the host vehicle relative to the white line using a change in relative azimuth.
6 . The target recognition device according to claim 3 , wherein the processor is further configured to: select the azimuth based on the GNSS by default; and select either the white-line azimuth or the azimuth based on the yaw rate, when the positioning state of the GNSS is not within the predetermined accuracy.
7 . The target recognition device according to claim 3 , wherein a position and the white-line azimuth of the host vehicle are stored in association with each other.
8 . The target recognition device according to claim 3 , wherein the processor is further configured to record a detected past position of the white line, using an approximate parameter for the recording.
9 . The target recognition device according to claim 3 , wherein the processor is further configured to capture the white line closest to the target using a white line recorded as a position of the host vehicle on the moving trajectory at a time point.
10 . A rearward white line inference method using a computer, the rearward white line inference method comprising: detecting a white line ahead of a host vehicle; obtaining a default azimuth an azimuth of the host vehicle relative to the white line every predetermined time interval based on a white line detection result; select: (i) the default azimuth by default; (ii) an azimuth based on a Global Navigation Satellite System (GNSS), in response to satisfying a first condition that (a) a positioning state of the GNSS is within a predetermined accuracy, and (b) a speed is above a predetermined level or a change in cumulative azimuth is below a predetermined level; (iii) an azimuth based on a group of point sequences of forward white lines, in response to satisfying a second condition that (a) the first condition is not satisfied, (b) the change in cumulative azimuth is above the predetermined level, and (c) at least one white line exists; and (iv) an azimuth based on a yaw rate, in response to satisfying a third condition that the first condition is not satisfied, (b) the change in cumulative azimuth is below the predetermined level, and (c) no white line exists; and determining a relative azimuth of the host vehicle at a certain point based on the selected azimuths of the host vehicle.
11 . A target recognition method using a computer, the target recognition method comprising: detecting a white line ahead of a host vehicle; obtaining a white-line azimuth of the host vehicle relative to the white line every predetermined time interval based on a white line detection result; selecting: (i) the white-line azimuth; (ii) an azimuth based on a Global Navigation Satellite System (GNSS), in response to satisfying a first condition that (a) a positioning state of the GNSS is within a predetermined accuracy, and (b) a speed is above a predetermined level or a change in cumulative azimuth is below a predetermined level; (iii) an azimuth based on a group of point sequences of forward white lines, in response to satisfying a second condition that (a) the first condition is not satisfied, (b) the change in cumulative azimuth is above the predetermined level, and (c) at least one white line exists; and (iv) an azimuth based on a yaw rate, in response to satisfying a third condition that the first condition is not satisfied, (b) the change in cumulative azimuth is below the predetermined level, and (c) no white line exists; determining a relative azimuth of the host vehicle at a certain point based on the selected azimuths of the host vehicle; detecting a target behind the host vehicle; determining a moving distance of the host vehicle; determining a moving trajectory of the host vehicle relative to the white line based on the relative azimuth at every predetermined time interval and the moving distance at every predetermined time interval; determining a position of a white line outside a current white line detection range based on a past white line detection result and the moving trajectory; and identifying a positional relationship between the white line outside the detection range and the target based on the position of the white line outside the inferred detection range and a position of the detected target.