JP4277717B2 - Vehicle position estimation device and driving support device using the same - Google Patents

Description

  The present invention relates to a driving support apparatus that supports driving by estimating the position of a vehicle.

  In recent years, various equipment has been attached to vehicles for the purpose of improving safety and comfort in driving a car. And, automobiles are equipped with a function to prevent accidents and are evolving to continuously improve safety. In order to prevent the occurrence of an accident, a warning is issued to the driver before reaching the accident, and the vehicle is decelerated and steered at the discretion of the driver, and the throttle, brake and steering are bypassed by the driver. There is a method to avoid it by mechanical control.

  Here, if it is going to improve the function of a warning or control, a high precision will be needed with respect to the current position of the vehicle and the map information referred to by the position information. For example, the throttle and brake are controlled to stop at the position of the stop line, the throttle and the steering are controlled not to deviate from the lane on a sharp curve, or to control the vehicle to travel in the center of the lane by steering when the lane departs. It can be said that this is an application that requires accuracy.

  With regard to position information, in current automobiles, current position estimation using GPS positioning is used as a main application of navigation systems. However, since the accuracy of GPS varies greatly depending on the location and time, there is a limit to applying it to a highly accurate alarm or control using this as the current position.

  On the other hand, a technology that extracts the lane boundary line such as a white line from the video captured by the in-vehicle camera and estimates the current position in the lateral direction and uses it for warning and control has been put into practical use, but only information on the captured range is used. Therefore, for example, when traveling at a high speed, it may be difficult to decelerate before the curve or to stop at the stop line.

As a method for solving these problems and obtaining a vehicle position with high accuracy, a method for extracting road shape information from a video taken by an in-vehicle camera and collating it with map data is disclosed in JP-A-6-94470. Yes. Similarly, a method of extracting information such as traffic lights, signs, the number of lanes, road width, etc. from a video photographed by an in-vehicle camera and correcting the current position by comparing it with map information is disclosed in Japanese Patent Laid-Open No. 9-9
It is disclosed in Japanese Patent No. 152348.

Japanese Patent Laid-Open No. 6-94470 Japanese Patent Laid-Open No. 9-152348

  However, in the method disclosed in Japanese Patent Application Laid-Open No. 6-94470, if an attempt is made to find a point to be collated using only the in-vehicle camera image and road information, for example, there are a plurality of points to be collated on a straight road, and the current position is unique. There is a problem that it is not found in.

  In the method disclosed in Japanese Patent Laid-Open No. 9-152348, when there is a characteristic object that can be collated with the image of the in-vehicle camera, it is possible to estimate the current position to some extent accurate. There is a problem that accuracy is not guaranteed.

  The present invention extracts a lane boundary line and a landmark (such as a road surface display or a road sign) from an image taken by an in-vehicle camera in order to estimate a high-accuracy position of an automobile, and at a lane level with detailed map information on the road. These two pieces of information are used in combination for the consistency check, and the vehicle speed information is used at a place where these pieces of information do not exist. At this time, the target position in the detailed map information for performing the consistency check is narrowed down using the approximate position information obtained by GPS.

  Then, alarm / control of the vehicle is executed using the estimated high-accuracy position and the extracted partial detailed map information. In the alarm and control determination, the target value is set using not only accurate current position information but also route information to the destination and detailed map information on the route. The alarm timing is set in consideration of a driver's alarm recognition delay, device operation delay, and the like. For a car that can be used by switching between two power sources, such as a hybrid car in recent years, a target for realizing an optimum driving mode in consideration of a road gradient is set.

  Here, the route information necessary for setting a control target that assumes a certain amount of driving process is outlined from the current location and destination (assuming a map with the level of detail used in the current navigation system) The detailed map information existing on the calculated route is extracted and sent to the function for executing the alarm / control.

  In addition, it is considered that the detailed map information of the range that can be traveled changes every day due to new road construction, traffic regulation, traffic closure, road construction, and the like. In order to reflect this in driving support by alarm / control, a function for updating the in-vehicle map DB information, a function for receiving a map update request from the in-vehicle device, and a map server for holding the latest map information are provided.

  On the other hand, in order to generate a detailed map DB to be used for referring to information, images are continuously acquired by a camera that is installed in a vehicle and images the front, and a partial lane boundary line sequence from the acquired images Information is extracted, and each extracted image is converted into data when viewed from the road surface vertical direction to superimpose the data so as to be most closely matched, thereby generating detailed lane boundary data. At this time, in order to prevent the accumulation of overlay errors, the exact position of landmarks such as road markings and road signs are separately measured and stored in a database so that the exact position of the landmarks in the captured image can be referenced. Let it be a point. Further, lane point sequence information expressed by an axis along the road surface is converted into three-dimensional elevation information with reference to elevation data at representative points on the map measured separately.

  According to the present invention, there is an advantage that driving assistance can be stably implemented in many scenes of a vehicle traveling scene.

  In addition, by providing a map DB update function, it is possible to realize appropriate driving support even when lane traffic is temporarily restricted.

  A high-accuracy position of the vehicle is estimated by identifying a location where the local road alignment acquired by the in-vehicle camera matches in the detailed map information DB. And the driving assistance using this position information is realized, while the function of updating the map information is also provided.

  In order to realize driving support based on the vehicle position detected with high accuracy, the device of the present invention most closely matches the local road alignment information obtained from the image information captured by the in-vehicle camera and the detailed road alignment information separately acquired. A position estimation device that determines the current position by identifying the location to be used, a route information generation device that extracts detailed map information on the vehicle's travel route and the route, and driving assistance based on accurate position and route / detail map information Further, the configuration includes a map update vehicle-mounted device for updating a detailed map to the latest information, a map update base station that accepts a map update request from the vehicle, and a latest map server.

FIG. 1 shows the functional configuration of a position estimation device 100 that is one of the components of the device of the present invention. The position estimation apparatus 100 first includes a GPS positioning unit 101 that obtains the current position by GPS, a detailed map DB 102 that includes information such as the position type of lane alignment and landmarks (road markings, signs, etc.), and the GPS positioning unit 101. And a partial detailed map extraction unit 103 that extracts a map near the position obtained by GPS positioning using information from the detailed map DB 102. Partial map information is extracted in order to quickly and accurately identify locations that match local lane-level road alignment information acquired by an in-vehicle camera (described later) from a large amount of detailed map information. is there.

  Further, the current position identification is performed by a white line extraction unit 106 that extracts road line information of a lane level such as a white line from image information captured by the in-vehicle camera 105 and a landmark extraction unit 107 that extracts landmarks from the same captured image information. Ask for the information you need.

  The current position estimation unit 108 uses the vehicle speed information obtained by the vehicle speed detection unit 104 to determine the amount of movement of the vehicle and estimates the current position of the vehicle from the vehicle position information. Using the information obtained from the white line extraction unit 106 and the landmark extraction unit 107 in the detailed map obtained from the map extraction unit 103, a point where the scene obtained by the in-vehicle camera matches the detailed map information is identified. Then, it is obtained as a new highly accurate vehicle position.

  The current position / local map information obtained by the current position estimating unit 108 is sent to the alarm / control device 200 and used for driving support, while the position information obtained by the GPS positioning unit 101 and the detailed map DB 102 are routed. The information is sent to the information generation device 300 and used for generating a route to be a driving assistance target and a detailed map on the route.

Here, a road on which a vehicle for driving assistance by the device of the present invention travels will be described with reference to FIG. In this example, it is assumed that the road network 400 forms a road network arranged in a lattice pattern, and detailed road information exists only for roads surrounded by broken lines on the road network. And, from the current location 401 to the destination 402, the vehicle is at point A403, point B
It is assumed that a route 406 passing through 404 and the point C405 is passed. In this example, the vehicle has detailed map information only in the section from the point A403 to the point C405 on the route 406, and only the navigation map information in the section from the current location 401 to the point A403 and the section from the point C405 to the destination 402. Indicates that it does not exist.

  Also, with regard to white line information collected from images captured by the in-vehicle camera 105, there are places that are not necessarily clearly drawn, such as in intersections, residential areas, and parking lots outside the road, and are also collected from images of the in-vehicle camera 105. Some landmark information, such as road markings and signs, exist only discretely on the road. Therefore, it is necessary to fully consider the processing when the detailed map, white line, and landmark information are missing.

The operation flow of each function in the position estimation apparatus 100 in FIG. 1 will be described with reference to FIG. First, a processing flow including the processing from step 604 to step 608 in FIG. 6A will be described (hereinafter, this will be referred to as process 6-1). The process of estimating the approximate position by the GPS in step 604 in the GPS positioning unit 101 is a process of obtaining the approximate position using a GPS receiver, and includes single positioning, differential positioning, and further kinematic positioning such as RTK-GPS. . Further, the GPS positioning includes not only GPS satellites in the United States but also satellites such as GLONASS and GALILEO and pseudo-satellite installed on the ground. The accuracy of GPS positioning varies depending on the method, and the positioning accuracy varies depending on the location and time because it is affected by the shielding and reflection of GPS signals by structures. Therefore, here, the position information obtained by GPS is used as a key for extracting data in the vicinity of the detailed map DB having a large capacity.

  In this embodiment, the approximate position when extracting data from the detailed map DB is determined by GPS positioning. However, if a marker representing the position on the road is painted separately from the landmarks described later, It is possible to detect the approximate position from the video by the in-vehicle camera, and the GPS positioning unit can be omitted.

  The process of setting the detailed map extraction range based on the GPS positioning accuracy information in step 605 includes receiving the information on the integrity included in the positioning signal from the GPS satellite and the GPS satellite signal separately measured or calculated for each point on the road. This is processing for setting an extraction range using situation information. Basically, the extraction range is narrow when the reception accuracy is high, and the extraction range is set wide when the reception accuracy is low. As a setting method, there are a method of determining a range according to available information, a method of determining a range based on error information based on actual measurement of a GPS positioning position, and the like.

  Next, the process of determining the presence / absence of a detailed map within the range of Step 606 in the partial detailed map extraction unit 103 is detailed in the detailed map DB for the range set at Step 605 based on the approximate position obtained at Step 604. Whether the map information exists is determined. If the detailed map information does not exist, the process returns to step 604 in the determination process of step 607. On the other hand, if the detailed map information exists, the partial map information within the corresponding range is extracted from the detailed map DB 102, and the information thus obtained is stored in a place that can be referred to from other processing processes (step 608). The processing in the partial detailed map extraction unit 103 is repeatedly executed during traveling. In this case, as shown in FIG. 9, assuming the circular detailed map extraction range 1203 centered on the GPS positioning position 1202 and having the set range as the radius, the presence / absence determination and the detailed map information are extracted. . Also, as an option, using map matching technology, the GPS position is rearranged to a position on the road, and then judgment and extraction are performed, or judgment and extraction are performed according to the type and importance of the road included in the circular area The range of detailed roads to be changed may be changed.

  Next, with reference to FIG. 6B, a flow of processing in the white line extraction unit 106 constituted by the processing from step 609 to step 612 and processing in the landmark extraction unit 107 constituted by the processing from step 614 to step 616 will be described. (In the following, these are referred to as process 6-2). The process of acquiring video from the in-vehicle camera in step 609 is a process of reading video data captured by the in-vehicle camera 105 in FIG. FIG. 9 shows an image of the in-vehicle camera imaging range 1204 captured from the in-vehicle camera 105 mounted on the vehicle 1205. The process of extracting the lane boundary line such as a white line from the video data in step 610 is a process of extracting information indicating the lane boundary such as a white line or a yellow line from the image information photographed by the in-vehicle camera 105. Various image processing technologies can be applied to the device itself, but there are locations where the lane boundary cannot be clearly extracted while traveling on the road, so it is determined whether the lane boundary can be extracted ( Step 611) If the boundary line cannot be extracted, the process proceeds to Step 614. If the lane boundary line can be extracted, the lane boundary line information is stored in a storage area that can be referred to by other processing processes (step 612). At this time, in order to determine the consistency between the extracted boundary line information and the information in the road detail DB, the camera parameters (installation height, angle of view, depression angle, focal length, etc.) of the in-vehicle camera are taken into consideration. Expressed as 3D or 3D point sequence information.

Next, in the process of extracting landmark candidates such as road markings and signs from the video data in step 614, the markings painted on the road surface from the image information taken by the in-vehicle camera 105 captured in step 609 ( Road speed facilities such as maximum speed display 1207, stop line 1209, traveling direction 1208, or pedestrian crossing as shown in FIG. 9) and road signs 1206 (regulatory signs, instruction signs, warning signs, guidance signs, etc.) This is a process of extracting what is called a landmark. Specifically, in order to extract a road marking, a pattern using a pattern pre-learning function for landmark candidates extracted through processing such as binarization focusing on the contrast between the road road surface and the road marking. A matching method can be applied. On the other hand, for the extraction of road signs, a pattern matching method that makes use of the characteristics of sign shapes such as circles and squares can be applied. Alternatively, it is possible to adopt a method of discriminating by classification focusing on each feature amount.

However, since there are many places where there are no landmarks while traveling on the road, it is determined whether or not there are landmarks in the image (step 615). The processing returns to step 609. If a landmark exists, the landmark information is stored in a storage area that can be referred to by other processing processes (step 616). Then, the processing from step 609 is repeatedly executed. The extracted landmark information is expressed together with its type as two-dimensional or three-dimensional position information in consideration of the camera parameters of the in-vehicle camera 105 described above.

  With reference to FIG. 7, the processing flow in the current position estimation unit 108 configured by the processing from step 617 to step 623 will be described. First, it is determined whether or not the partial detailed map information has been extracted by the process 6-1 (step 617). If not extracted, the current position calculated based on the detected vehicle speed is output (step 623). If partial detailed map information has been extracted, it is next determined whether or not landmark information has been extracted in process 6-2 (step 618). If landmark information has not been extracted, it is next determined whether or not lane boundary information has been extracted (step 621). If not, the current position calculated based on the detected vehicle speed is output as before. (Step 623), the process returns to the beginning. Here, in the process of detecting the vehicle speed in the vehicle speed detection unit 104, for example, rotation information of the axle is detected, and the vehicle speed is obtained in consideration of the tire diameter and tire characteristics. In the process of calculating the current position, the vehicle movement amount is obtained using the vehicle speed and processing time interval information thus obtained, the vehicle position at that time is obtained, and the current position is stored in a predetermined storage area. deep. Basically, the vehicle position in the direction of the movement vector of the vehicle is obtained, but it is also possible to calculate the position on the route by referring to the route information obtained by the route information generation device 300. Information obtained in a place that can be referred to from other processing processes is stored. The current position is updated by repeatedly executing the above processing.

  However, if the lane boundary information has been extracted, a process of calculating a point where the lane boundary extracted in step 622 and the partial detailed map are most consistent is executed. This is the lane information for in-vehicle camera imaging obtained by the process 6-2 in the partial detailed map obtained by the process 6-1 (for example, lane information in the circular extraction range 1203 in FIG. 9). This is a process of performing a consistency check for identifying a location where the lane information (for example, lane information in the in-vehicle camera imaging range 1204 in FIG. 9) matches, and these two pieces of information can be executed together. As a specific method, the lane information in the detailed map DB 102 and the lane information extracted from the video captured by the in-vehicle camera 105 are both composed of a plurality of point sequences expressed in two-dimensional or three-dimensional coordinates. Is assumed. That is, it is assumed that the lane 1201 in FIG. 9 is represented by a point sequence.

  FIG. 9 shows a situation where the GPS positioning position 1202 obtained by GPS has an error and is thus deviated from the actual position of the vehicle 1205. Therefore, the consistency of the lane point sequence measured by the in-vehicle camera is obtained for the detailed lane point sequence within a predetermined range such as 1203, and this is repeatedly calculated while gradually shifting within the range, and the highest consistency is obtained. Find the point.

  Consistency check using lane information expresses lane boundary lines extracted from in-vehicle camera video as point series information expressed in two-dimensional or three-dimensional coordinate values, while road lane boundaries in detailed map information Similarly, the line is also expressed as point series information, and the degree of coincidence between the point series information is obtained.

  For example, the distance between two point sequences at several representative coordinate values may be obtained, and the least mean square may be used as an index of consistency. Then, within the narrowed matching target range, the index is calculated by changing the number of matching points, and as a result, the point where the smallest index value matches the vehicle-mounted camera image in the detailed map information, that is, the vehicle Accurate current position. The current position of the vehicle in the detailed map information is, for example, the origin of the image taken from the in-vehicle camera installation position or the point sequence information of the lane boundary line in the local coordinate system. Can be obtained by determining where the is located by coordinate transformation. In addition, in order to obtain the current position of the vehicle using the landmark information, the local landmark in the in-vehicle camera image is extracted by registering the attribute and position information in advance in the detailed map information. The current position can be obtained by searching the detailed map information, referring to the exact position, and aligning the point series information of the lane boundary line of the in-vehicle camera in the same manner as described above to determine the direction of the vehicle. In addition, the current position can be obtained more accurately by matching using a three-dimensional coordinate value using road elevation information. That is, the video obtained with the in-vehicle camera is a two-dimensional image. Therefore, when matching with the point sequence information of the lane boundary line of the detailed map information, the road surface is expressed as an aggregate of planes from the elevation information, for example, and the coordinates from the image from the in-vehicle camera are mapped before matching. The point sequence information that has been converted, coordinated, and coordinate-converted is matched with the point sequence information of the detailed map.

  Specifically, as described above, by the process 6-2, the left and right two lane boundary line point sequences (hereinafter referred to as in-vehicle detection point sequences) in front of the running lane are obtained from the image of the in-vehicle camera. On the other hand, all lane point sequences included in the range are obtained from the entire detailed map DB 102. In the example of FIG. 9, two lane point sequences (hereinafter referred to as “extracted lane point sequences”) in the extraction range 1203 of the detailed map correspond to that.

  First, set the base point of the right vehicle detection point sequence to the base point of the right extracted lane point sequence, and use the least square method etc. as the axis point, the right lane detection point sequence is the right extracted lane point sequence The direction in which the degree of overlap is the largest is obtained. At that time, the position and orientation of the left vehicle-mounted detection point sequence are also moved together with the right vehicle-mounted detection point sequence. Then, the degree of overlap between the moved left-side vehicle detection point sequence and the left-side extracted lane point sequence is obtained using an index such as a mean square, and the thus obtained index is used as the consistency index at this point. This process is repeatedly determined while moving the base point of the right lane detection point sequence on the right extracted lane point sequence by a predetermined interval, so that the most consistent point is set as the vehicle position. Here, the consistency can be calculated in the same manner by using the left in-vehicle detection point sequence, and the consistency index may be obtained by combining both the left and right overlapping degrees. Furthermore, as a method of searching for the most consistent point, in addition to a method of obtaining at every predetermined interval, a convergence calculation that goes back and forth on the extracted lane point sequence may be set. The vehicle position information calculated using the method described above and the in-vehicle detection point sequence information, which is the extracted local map information, are output to the outside (step 620), and the process returns to step 604 and is repeatedly executed. The

Here, when determining the consistency, the accuracy is increased by using the altitude information. This will be described with reference to FIG. 1501 shows a situation in which a vehicle 1502 equipped with a vehicle-mounted camera is photographing the road surface in the forward visual field 1503, but the current position of the vehicle and the photographing range 1504 are not necessarily the same because of a bank of slopes and curves. Sometimes it is not on a flat surface. In this case, the consistency determination includes an error without taking such altitude information into consideration. In order to avoid this, when obtaining a consistency index between the lane detection point sequence from the in-vehicle camera and the extracted lane point sequence from the detailed map information, a plurality of road surfaces are obtained from the altitude information in the extracted lane point sequence. It is expressed by a plane (for example, a plane triangle), and is obtained after a coordinate transformation for mapping a lane detection point sequence thereon. As a result, an accurate consistency index with the error corrected can be obtained.

  Next, when it is determined in step 618 that the landmark information has been extracted, processing for calculating the most consistent point of the partial detailed map is executed using the landmark position in step 619. This is the in-vehicle camera obtained in the process 6-2 in the partial detailed map obtained by the process 6-1 (for example, lane information in the circular area of the extraction range 1203 of the detailed map shown in FIG. 9). This is to identify a location where information of imaging landmark candidates (for example, the road sign 1206, the maximum speed display 1207, the stop line 1209, the traveling direction 1208, etc. in FIG. 9) matches, and these two pieces of information are executed together It becomes possible. Here, it is assumed that the position and attribute of the landmark to be referred to are separately measured by a method such as surveying and stored in the detailed map DB 102.

  Specifically, when the attributes and positions of the landmarks are extracted by the process 6-2, it is detected from the landmarks matching the attributes in the detailed map information extracted by the process 6-1. Select the one closest to the location. Then, the current position of the vehicle is obtained using the position information of the selected landmark recorded in the detailed map information. The vehicle position information calculated using the method described above and the in-vehicle detection point sequence information that is the extracted local map information are output to the outside (step 620), and the process returns to step 604 and executed.

FIG. 2 shows the functional configuration of the alarm / control device 200, which is one of the components of the device of the present invention. The alarm / control device 200 first includes a travel plan generation unit 201 that receives current position / local map information from the position estimation device 100 and further receives route / detailed map information from the route information generation device 300. The alarm determination unit 202 uses the processing result of the travel plan generation unit 201 to determine whether or not to issue an alarm to the driver, and the alarm display unit 203 actually issues an alarm to the driver. As a specific method of alarming, sound output, warning lamp lighting, steering wheel vibration, minute movement of the seat belt, and the like are assumed.

  On the other hand, when the position information / detailed map information is used for vehicle control, the real-time control unit 204 refers to the operation plan / control target value information obtained by the travel plan generation unit 201, and the throttle 205, brake A control signal is issued to 206 and steering 207. Furthermore, when there are a plurality of power systems in the vehicle, such as a hybrid vehicle, the real-time control unit 204 issues a control signal to the power switching unit 208 to select an appropriate power to control the traveling.

  Here, the processing flow of the alarm / control apparatus 200 will be described with reference to FIG. First, referring to FIG. 8A, a processing flow configured by the processing from step 701 to step 704 executed in the travel plan generation unit 201 will be described (hereinafter referred to as process 7-1). In the process of determining the travel lane on the route in step 701, the information sent from the position estimation device 100 and the route information generation device 300 is read, and the vehicle travels when traveling along the route from the currently traveling lane. The lane line to be obtained is obtained. In the area where the number of lanes increases or decreases and before the intersection, an appropriate lane is selected in consideration of the route information and the lane direction attribute.

  Next, in the process of setting constraint information such as curvature and stop position in step 702, the maximum speed information and stop line position that can travel without departing from the lane on the curve, and in some cases legally regulated speed information in step 701 With reference to the determined lane information, it is given as a constraint condition on the route.

  Then, a travel plan for how the vehicle is to be traveled up to a predetermined future time using the current position and speed of the vehicle as initial values is stored, and can be referenced from other processing processes. Save in the area (step 703). The travel plan here is expressed as a combination of a plurality of sets up to a predetermined time using information on a combination of future times, positions, and speeds of the vehicle as an information series. At this time, the constraint condition given in step 702 is considered. It can be said that the travel plan is a target expression on the system side that the vehicle wants to travel in this way.

  Further, in step 704, the process of estimating and saving the predicted travel locus with the current travel state as the initial value is, for example, the vehicle traveling when the vehicle travels as it is on the premise of constant acceleration motion. This is a process for obtaining information on a combination of time, position, and speed in the future as an information series.

  Next, with reference to FIG. 8B, a processing flow constituted by the processing from step 705 to step 707 in the alarm determination unit 202 will be described (hereinafter referred to as process 7-2). This process 7-2 has a part of the function aiming at preventing the occurrence of an accident by issuing an alarm to the driver. In step 705, in the process of setting the alarm timing in consideration of the driver's operation characteristics from the travel plan obtained by the process 7-1 and the predicted travel locus, the driver said that the driver has made a recognition error such as road surface conditions or signs. Assuming that an error will occur, the driver predicts the predicted travel path as a travel pattern drawn when the driver drives the vehicle due to a human error, and warns the driver that the predicted travel path does not deviate from the travel plan. Find the timing to issue. In the prediction of the travel locus, the assumption of constant velocity motion and the influence of road alignment and gradient are taken into consideration. In other words, when traveling at high speed, a traveling locus deviating from the lane is predicted on a steep curve, and a traveling locus on which the vehicle speed is decelerated on an uphill is predicted. Here, taking into account the delay time from when the alarm is issued until the driver starts operating the vehicle and the amount of vehicle operation by the driver when receiving the alarm (the amount of depression of the throttle and brake, the steering angle, etc.) Decide what level of alarm is to be issued at which point.

  In the alarm timing determination in step 706, if the timing is not the timing to issue an alarm, the process returns to step 705 again. If the timing is to issue an alarm, the alarm display unit 203 in the next step 707 is alerted. A process of transmitting the generated signal is executed. Similarly, the processing returns to step 705.

  Further, with reference to FIG. 8C, a processing flow constituted by the processing from step 708 to step 710 in the real-time control unit 204 will be described (hereinafter referred to as process 7-3). This process 7-3 has a part of function for the purpose of preventing the occurrence of an accident by directly controlling the throttle 205, the brake 206, the steering 207, the power switching unit 208, and the like. In the process of calculating the control target value for each control target from the travel plan obtained by the process 7-1 in step 708, the throttle 205, the brake 206, the steering 207, and the power switching unit are used to realize the travel plan of the vehicle. A control target for 208 is defined. The speed target needs to be realized by a combination of control of the throttle 205 and the brake 206, and the position target needs to be realized by a combination of the control of the steering 207, the throttle 205, and the brake 206. How to decide is conceivable.

  Then, by the process of transmitting the control signal to the control device in step 709, the control signal is output so as to reach the target value determined in step 708, and the current state quantity is received from the control device in step 710. A response to the output control signal is received, and the processing returns to step 708 to feed back the response.

  Here, it has been described that these processes 7-1, 7-2, and 7-3 operate in parallel, but it is also possible to realize only one of the process 7-2 and the process 7-3. .

  FIG. 3 shows the functional configuration of a route information generation device 300 that is one of the components of the device of the present invention. The route information generation device 300 first receives the position and detailed map information obtained by the GPS positioning from the position estimation device 100, refers to the destination information from the destination setting unit 301 and the data in the navigation map DB 302, A route calculation unit 303 for obtaining a route to the destination is included. The detailed map information extraction unit 304 on the route obtains detailed map information existing on the obtained route and transmits it to the external alarm / control device 200.

  Here, the process of the route information generation device 300 will be described. First, by setting the destination input by the user by the destination setting unit 301 in FIG. 3, the destination to be the target of driving support is set. This can be realized by using the method used in the navigation system. Next, the route calculation unit 303 calculates a route from the destination, GPS position information, and navigation map information, and obtains a route to the destination. The system currently used for the navigation system can also be used for this. Next, the detailed map information extraction unit 304 extracts the detailed map information on the route calculated above from the navigation map DB 302 to obtain information used in the alarm / control device 200. Finally, the previously calculated route And the extracted detailed map information is transmitted to the alarm / control device 200.

  FIG. 5 is a diagram showing a map update vehicle-mounted device 500, a map update base station 510, a latest map server 520 and functional configurations in each device, which are components of the device of the present invention. When using high-accuracy position information and detailed map information for driving support, a system for keeping the map information referred to always up-to-date is necessary. The updated map information includes newly established roads, closed road sections, construction sections, new and abolished landmarks, and changes in attributes and positions.

  The map update in-vehicle device 500 is mounted on a vehicle and plays a role of updating information in the detailed map DB 102 and the navigation map DB 302 in the vehicle. The functions included therein include a map update setting unit 501, a map information update unit 502, a local map information transmission unit 503, and a new map information reception unit 504.

  Here, the process of the map update vehicle-mounted device 500 will be described. First, the map update process is started by setting the map update setting unit 501 to update the latest map information. It is also possible to omit the setting by the user by always turning on this map update setting. Next, the map information update unit 502 extracts information related to the map information currently in use, and obtains the range and version information of the detailed map held by the in-vehicle device at that time. The local map information transmission unit 503 sends information regarding the currently used map information to the external map update base station 510. Transmission to the map update base station 510 is basically performed using wireless communication, but it is also possible to use wired communication. Then, the new map information receiving unit 504 receives the latest map information from the map update base station 510. Then, the map information update unit 502 updates the information in the detailed map DB 102 and the navigation map DB 302 using the received new map information.

  The map update base station 510 is installed on the ground side, receives a map update request from the vehicle, and obtains necessary map information from the latest map server and transmits it to the vehicle. The functions included therein include a local map information receiving unit 511, a map information selection requesting unit 512, and a new map information transmitting unit 513.

  Here, the processing of the map update base station 510 will be described. First, the local map information receiving unit 511 receives information on the map information currently in use transmitted from the map update vehicle-mounted device 500, receives information on the map held by the vehicle that transmitted the request, and then maps The information selection requesting unit 512 requests the latest information of a necessary part from the latest map server 520, and transmits the information transmitted from the map update in-vehicle device 500 to the latest map server 520. The contents to be transmitted to the latest map server 520 are the detailed map range and map version information transmitted from the local map information transmitting unit 503. Then, the map information selection request unit 512 receives the map information from the latest map server 520, the map update vehicle-mounted device 500 receives the latest map information to be updated, and finally, the new map information transmission unit 513 updates the map. New map information is transmitted to the in-vehicle device 500. Here, the map update base station 510 generally processes requests from a plurality of vehicles, but in order to efficiently deal with them, a detailed map information cache function is set to the latest map server 520. Before issuing a request, it is possible to search from information in the cache and to request the latest map server 520 for information not in the cache. In addition, as a method for updating based on the contents of specific detailed map information, several methods are used such as a method for replacing the entire map information, for example, a method for replacing a mesh-shaped designated area, and a method for replacing only a difference portion of information. Is feasible.

  The latest map server 520 plays a role of extracting and transmitting map information to be updated based on a request from the map update base station 510. The functions included in the latest map server 520 include a detailed map latest DB 522 and a navigation map latest DB 523 in addition to the map information extraction unit 521.

  Here, the process of the latest map server 520 will be described. First, the map information extraction unit 521 accepts a map information transmission request from the map update base station 510, and then extracts the latest map information corresponding to the request from the base station, so that the range and version of the detailed map can be obtained. The consistency is confirmed, and the latest information is taken out from the detailed map latest DB 522 and the navigation map latest DB 523 as necessary. Then, the extracted latest map information is transmitted to the map update base station 510.

  Here, in order to exchange information between the above-described map update vehicle-mounted device 500 and the map update base station 510 or the latest map server 520, not only wireless and wired communication but also a removable recording medium is used. 500 stores the entire map information or the detailed map range and version information described above in the recording medium, and obtains update information directly from the map update base station 510 or the latest map server 520 having a function capable of reading and writing to the medium. It is also possible to do.

  On the other hand, the description has been made on the assumption that the detailed map DB that can be referred to exists even partially in advance, but here, the detailed map is based on the processing used for matching the video by the in-vehicle camera and the detailed map. A process for generating information will be described. The detailed map information here mainly refers to information expressing lane boundary lines as point sequence information.

FIG. 10 shows a state in which the imaging range moves with the movement of the vehicle by the in-vehicle camera 105 mounted on the vehicle and photographing the front. An imaging range 1301 indicates a range captured by a vehicle-mounted camera at that time from a vehicle position at a certain time t1. Similarly, an imaging range 1302 and an imaging range 1303 indicate the imaging range at times t2 and t3. At this time, the images acquired by the in-vehicle camera become images 1304, 1305, and 1306, respectively. In the video 1305, a road sign as a landmark is photographed. Then, as will be described later, in order to extract partial lane boundary information from the obtained image and align it at the most consistent location, first, the image is transformed as seen from the road surface vertical direction. . The converted images are an image 1307, an image 1308, and an image 1309, respectively. Partial lane boundary line information is extracted from each of these images as point sequence information.

  At this time, images are continuously photographed with an overlap necessary for the alignment process. The largest amount of recording occurs when the camera shoots in accordance with the period in which video is acquired. However, the cost of the mounted device may increase. Although it is possible to reduce the cost of the processing apparatus by setting a longer processing cycle, for example, when the operation is stopped, the same scene is recorded many times. In essence, it is most efficient to take an image based on the amount of movement of the vehicle. Therefore, for example, a method of obtaining and recording the imaging timing from the vehicle speed information obtained by the vehicle speed detection unit 104 is effective.

Next, processing for connecting partial lane boundary information will be described with reference to FIG. As described with reference to FIG. 10, in the partial lane boundary information extracted from the images (1307, 1308, 1309) obtained by converting the image captured by the in-vehicle camera 105, first, information about the time t 1 is used as a base point. And Next, using the position at time t2 estimated from the vehicle speed information obtained by the vehicle speed detection unit 104 as an initial position, a part where the partial lane boundary information at time t2 is most consistent is searched. The matching process used in the description of the current position estimating unit 108 in FIG. 1 can be applied to this process. Here, since the landmark is photographed in the video at time t2, the absolute position (latitude and longitude) is determined with reference to the landmark position DB 1401, and the point sequence information of the partial lane boundary before that is corrected. . A method of correcting the lane boundary information between two landmarks so that the error becomes uniform can be applied. Similarly, the partial lane boundary information for time t3 is overlaid at the most consistent location. For the overlapped lane boundary line point sequence information, partial lane boundary line information at either time t2 or time t3 may be selected, or an average value may be used as a representative point.

  In this way, a continuous lane boundary line sequence can be finally obtained. In the example of FIG. 11, two pieces of point sequence information are obtained. The position of the landmark can be obtained by a conventional survey.

  Furthermore, if the map elevation information can be referred to, it can be converted into lane boundary information as a three-dimensional coordinate point sequence. For example, the Geographical Survey Institute has published altitude data of 5m, 50m, 250m mesh. Therefore, the obtained elevation information is used as the representative point elevation DB 1402, and the mesh points (= representative points) are linearly or smoothly interpolated, and there is a link level map (a level mainly used in the current navigation system). Map), and then, by applying a transformation that attaches the point sequence data of the lane boundary along the road traveling direction obtained by the above-described method, each point of the point sequence has a three-dimensional coordinate value. Will have. In addition, the altitude information includes more accurate three-dimensional coordinate values by including information on the structure provided for the functioning of the road, such as bank information on the curve and embankment provided for creating the road. Can be.

Finally, the obtained point sequence information of the lane boundary line is stored in the lane boundary line DB 1404. The detailed map construction DB 1405 includes information on the landmark position DB 1401 and the lane boundary line DB 1404, and is the original data of the detailed map DB used in the vehicle. That is, the detailed map construction DB 1405 in FIG. 11, the detailed map latest DB 522 in FIG. 5, and the detailed map DB 102 in FIG. 1 have the same information items, but since each is updated at any time, the information version is different at some point. There is. Basically, the data update operation is performed using the detailed map construction DB 1405 as the original data, and a version that can be released at a certain point of time is disclosed as the detailed map latest DB 522 on the latest map server 520. The vehicle detailed map DB 102 is updated when accessed by the map update base station 510.

  If high-accuracy position estimation based on a detailed map is realized in this way, it will be more difficult than GPS positioning (including correction by map matching) and simple navigation maps, which have been insufficient in accuracy. There is an effect that quantitative driving support can be realized. Specifically, it can be realized by using a quantitative expression such as “the stop position is XXm later” from the support that has been limited to a qualitative level such as “it is a stop position soon”. Furthermore, it can be said that this quantitativeness is indispensable when realizing automatic control in which throttle, brake, and steering are directly operated.

  By preparing two navigation maps and detailed maps as reference maps in this way, even if there is only a part of the detailed map information that generally requires cost and time to maintain all There is an effect that driving support can be realized continuously.

  By applying this device to automobiles, it can contribute to improving safety and can also be applied to railways.

It is a block diagram centering on a position estimation apparatus. It is a block diagram centering on an alarm / control device. It is a block diagram centering on a route information generation device. It is a figure which shows the example of a road network. It is a block diagram in a map update vehicle-mounted apparatus, a map update base station, and the latest map server. It is a processing flow in a position estimation apparatus. It is a processing flow in a position estimation apparatus. It is a processing flow in the alarm / control device. It is an example of the road which expressed the detailed map information and the range imaged with the vehicle-mounted camera. It is explanatory drawing for producing | generating detailed map information from a continuous image. It is a block diagram for producing | generating detailed map information. It is explanatory drawing of the vehicle-mounted camera visual field range in case there exists a gradient in a road surface.

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 ... Position estimation apparatus, 101 ... GPS positioning part, 102 ... Detailed map DB, 103 ... Partial detailed map extraction part, 104 ... Vehicle speed detection part, 105 ... Car-mounted camera, 106 ... White line extraction part, 107 ... Landmark extraction part, DESCRIPTION OF SYMBOLS 108 ... Current position estimation part, 200 ... Warning / control apparatus, 201 ... Travel plan production | generation part, 202 ... Alarm determination part, 203 ... Alarm display part, 204 ... Real time control part, 205 ... Throttle, 206 ... Brake, 207 ... Steering , 208 ... power switching unit, 300 ... route information generating device, 301 ... destination setting unit, 302 ... navigation map DB, 303 ... route calculating unit, 304 ... detailed map information extracting unit on route, 500 ... map update in-vehicle device, 501 ... Map update setting unit 502 ... Map information update unit 503 ... Local map information transmission unit 504 ... New map information reception unit 510 ... Map update base station 51 ... the current map information receiving unit, 512 ... map information selection requesting unit, 513 ... new map information transmission unit, 520 ... latest map server, 521 ... map information extraction unit, 522 ... More Map latest DB, 523 ... navigation map latest DB.

Claims (4)

External imaging means mounted on the vehicle, lane boundary extraction means for extracting white line information based on image information acquired by the external imaging means, equipment extraction means for extracting road-attached equipment information, and positioning results by GPS positioning In the vehicle position estimation device provided with the GPS positioning means to be obtained and the map information database,
Map extraction means for extracting map information around the positioning result obtained by the GPS positioning means from the map information database;
When the road ancillary equipment information is extracted by the equipment extraction means, the position of the closest road ancillary equipment whose attribute of the road ancillary equipment matches as a point most consistent with the extracted road ancillary equipment is extracted from the map extraction means. When the lane boundary information is extracted by the lane boundary extraction means, the lane point sequence of the lane boundary line is selected as a point most consistent with the extracted lane boundary information. Means for searching the position of the most overlapping lane point sequence from the map information extracted by the map extraction means ;
A vehicle position estimation apparatus characterized in that the most consistent point is set as the current position of the vehicle. External imaging means mounted on the vehicle, lane boundary extraction means for extracting white line information based on image information acquired by the external imaging means, equipment extraction means for extracting road-attached equipment information, and positioning results by GPS positioning In the vehicle control apparatus provided with the GPS positioning means to be obtained and the map information database,
Map extraction means for extracting map information around the positioning result obtained by the GPS positioning means from the map information database;
When the road ancillary equipment information is extracted by the equipment extraction means, the position of the closest road ancillary equipment whose attribute of the road ancillary equipment matches as a point that most matches the extracted road ancillary equipment is extracted from the map extraction means. When the lane boundary information is extracted from the map information extracted by the lane boundary extraction means, the lane point sequence of the lane boundary line as a point most consistent with the extracted lane boundary information Means for searching the position of the most overlapping lane point sequence from the map information extracted by the map extraction means ;
It has a function of setting the most consistent point as the current position of the vehicle, presenting support information to the driver based on the set current position information of the vehicle and the map information, and generating a vehicle control signal A vehicle control device.   3. The vehicle control device according to claim 2, further comprising a navigation map information database storing road information, and obtaining a route to a set destination from the position information obtained by the GPS positioning means and the navigation map information database. A vehicle control device that extracts map information corresponding to the route from the map information database.   2. The vehicle position estimation apparatus according to claim 1, further comprising: a position database storing pre-measured position information of road accessory facilities; a lane information database storing lane information; and an altitude database storing altitude data; The partial lane information is extracted by extracting partial lane information from the road images continuously captured by the means so as to overlap the shooting ranges, and referring to the road ancillary equipment imaged in the road image. Is generated in the overlapping portion to generate continuous lane information, obtain altitude information from an altitude database based on the positioning result by the GPS positioning means, and use the lane information corrected with reference to the altitude information. A vehicle position estimating apparatus, wherein a map information database is updated.

Images