Search

EP-3582135-B1 - METHOD AND APPARATUS FOR DETECTING, IN AN ENVIRONMENT OF AN OBJECT, THE PRESENCE OF A LOCATION WHERE ROAD BRANCHES OF AN AREA WITHIN THE ENVIRONMENT CONVERGE OR DIVERGE

EP3582135B1EP 3582135 B1EP3582135 B1EP 3582135B1EP-3582135-B1

Inventors

  • Onkarappa, Naveen
  • Chavali, Pothuraju

Dates

Publication Date
20260513
Application Date
20180614

Claims (8)

  1. A method for detecting, in an environment of a vehicle (V), the presence of a location (200; 100) where branches (41', 42'; 41, 42) of an area (A) within the environment converge or diverge, corresponding to the location of an entry (5) to or an exit (3) from a road (1), a multi-lane road (1), a highway (1) or freeway (1), in a driver assistance system, comprising steps of: providing (S1'; S1) an occupancy grid, by means of an occupancy detector, comprising a plurality of grid cells, each grid cell comprising a value for a probability of an occupancy in the environment of the vehicle (V), wherein the occupancy grid is computed based on Dempster-Shafer Theory; detecting (S2'-S6'; S2-S5), by means of a boundary detecting and processing device (302), boundaries (21', 22', 23', 24'; 21, 22, 23, 24) of the area (A) based on values of grid cells of the occupancy grid, wherein detecting the boundaries (21'-24'; 21-24) includes detecting a first boundary (21'; 21) on a first side (31) of the vehicle (V) in a direction (7) orthogonal, to a direction of movement (2) of the vehicle(V), and detecting a second boundary (22'; 22) on a second side (32) of the vehicle (V) opposite the first side (31) and wherein the step of detecting the boundaries (21-24) includes obtaining a distance (9), in the direction (7) orthogonal to the direction of movement (2) of the vehicle (V), between the first and second boundaries (21, 22) detected, at a plurality of points (10) spaced from each other along the direction of movement (2) of the vehicle (V), and further includes initiating a detection of at least one further boundary (23, 24) of the area (A) if the distance (9) between the first and second boundaries (21, 22) at the plurality of points (10) continuously increases as a distance of the point (10) from the object (V) increases; identifying (S7'; S7) the number of boundaries (21'-24'; 21-24) detected; depending on the number of boundaries (21'-24'; 21-24) identified, signaling (S7'; S7), by means of the boundary detecting and processing device (302), the presence of a location (200; 100) where branches (41', 42'; 41, 42) of the area (A) converge or diverge, if the number of boundaries is identified to exceed 2, and wherein the boundaries are road boundaries, wherein the step of detecting the boundaries (21'-24') includes obtaining lengths (L1, L2) of each of the first and second boundaries (21', 22') detected and calculating a difference between the lengths (L1, L2) of the first and second boundaries (21', 22'), and further includes initiating a detection of at least one further boundary (23') of the area (A) if the difference calculated exceeds a non-zero threshold value and wherein the further boundary (23') detected is a third boundary (23') of the area (A) and the detection of the third boundary (23') is initiated to proceed in a direction (15') towards an interior (I) of the occupancy grid (O), away from the gird boundary (8), in a manner so as to approach the vehicle (V), wherein the detection of the further boundary (24, 23) comprises detecting a further starting point (SP4, SP3) in a direction (11, 12) orthogonal, to the orientation of the first or second boundary (21, 22) in a distal end region (13, 14) thereof, and wherein the detection of the at least one further boundary (23') comprises detecting a further starting point (SP3') in a direction (11') orthogonal, to the orientation of a longer one (21') of the first and second boundaries (21', 22') in a distal end region (13') thereof.
  2. The method according to claim 1, wherein detecting the first boundary (21'; 21) and detecting the second boundary (22'; 22) each includes detecting a starting point (SP1, SP2) in the occupancy grid (O) in the direction (7) transverse to the direction of movement (2) of the object (V), assigning an initial region of interest window (W) in the occupancy grid (O) surrounding the starting point (SP1, SP2), and beginning with the initial region of interest window (W) forming a chain (C1, C2) of region of interest windows (W) by repeatedly assigning a further region of interest window (W).
  3. The method according to claim 1 or 2, wherein the step of detecting the boundaries (21'-24'; 21-24) includes terminating the detection of each of the first (21'; 21) and second boundary (22'; 22) if the current region of interest window (W) exceeds a grid boundary (8) of the occupancy grid (O) or if a sharp change of the orientation of the first or second boundary (22') is detected in the current region of interest window (W).
  4. The method according to claim 1, wherein the detection of the third boundary (23') is terminated in an end region (17') of the third boundary (23), and wherein the step of detecting the boundaries (21'-24') further includes initiating a detection of a fourth boundary (24') of the area (A) starting from the end region (17') of the detected third boundary (23').
  5. An apparatus (300) for detecting, in an environment of a vehicle (V), the presence of a location (200; 100) where branches (41', 42'; 41, 42) of an area (A) within the environment converge or diverge, the apparatus (300) comprising: an occupancy detector (301) adapted to provide an occupancy grid (O) comprising a plurality of grid cells, each grid cell comprising a value for a probability of an occupancy in the environment of the vehicle (V), wherein the occupancy grid is computed based on Dempster-Shafer Theory; a boundary detecting and processing device (302) adapted to detect boundaries (21'-24'; 21-24) of the area (A) based on values of grid cells of the occupancy grid (0), in a direction orthogonal to the vehicle (V), , wherein detecting the boundaries (21'-24'; 21-24) includes detecting a first boundary (21'; 21) on a first side (31) of the vehicle (V) in a direction (7) orthogonal, to a direction of movement (2) of the vehicle(V), and detecting a second boundary (22'; 22) on a second side (32) of the vehicle (V) opposite the first side (31) and wherein the step of detecting the boundaries (21-24) includes obtaining a distance (9), in the direction (7) orthogonal to the direction of movement (2) of the vehicle (V), between the first and second boundaries (21, 22) detected, at a plurality of points (10) spaced from each other along the direction of movement (2) of the vehicle (V), and further includes initiating a detection of at least one further boundary (23, 24) of the area (A) if the distance (9) between the first and second boundaries (21, 22) at the plurality of points (10) continuously increases as a distance of the point (10) from the object (V) increases and adapted to identify the number of boundaries (21'-24'; 21-24) detected, and further adapted to provide an output signaling the presence of a location (200; 100) where branches (41', 42'; 41, 42) of the area (A) converge or diverge, if the number of boundaries is identified to exceed 2, and wherein the boundaries are road boundaries, wherein the step of detecting the boundaries (21'-24') includes obtaining lengths (L1, L2) of each of the first and second boundaries (21', 22') detected and calculating a difference between the lengths (L1, L2) of the first and second boundaries (21', 22'), and further includes initiating a detection of at least one further boundary (23') of the area (A) if the difference calculated exceeds a non-zero threshold value and wherein the further boundary (23') detected is a third boundary (23') of the area (A) and the detection of the third boundary (23') is initiated to proceed in a direction (15') towards an interior (I) of the occupancy grid (O), away from the gird boundary (8), in a manner so as to approach the vehicle (V), wherein the detection of the further boundary (24, 23) comprises detecting a further starting point (SP4, SP3) in a direction (11, 12) orthogonal, to the orientation of the first or second boundary (21, 22) in a distal end region (13, 14) thereof, and wherein the detection of the at least one further boundary (23') comprises detecting a further starting point (SP3') in a direction (11') orthogonal, to the orientation of a longer one (21') of the first and second boundaries (21', 22') in a distal end region (13') thereof.
  6. An advanced driver assistance system comprising the apparatus according to claim 5.
  7. An autonomous driving vehicle comprising the apparatus according to claim 5.
  8. A robot system comprising the apparatus according to claim 7.

Description

The invention relates to a method and apparatus for detecting, in an environment of an object, the presence of a location where branches of an area within the environment converge or diverge. Moreover, the invention relates to an advanced driver assistance system, to an autonomous driving vehicle and to a robot system. Background Although applicable in principle to any system applying free space detection and/or environment modelling, the invention and its underlying problem will be hereinafter described in connection with an advanced driver assistance system. Modern vehicles like cars or trucks, for example, usually comprise driver assistance systems which may support the driver when performing driving maneuvers. For performing automatic driving maneuvers, a free space in the surrounding of a vehicle has to be detected. EP 3 029 602 A1 describes an approach to estimate free space surrounding a vehicle using multiple cameras and an occupancy grid map. A technique to form an occupancy grid and a histogram-based technique for determining a road boundary are described in "Efficient occupancy grid computation on the GPU with lidar and radar for road boundary detection" by F. Homm, N. Kaemp-chen, J. Ota and D. Burschka, 2010 IEEE Intelligent Vehicles Symposium, San Diego, CA, 2010, pp. 1006-1013. Driving maneuvers may involve, for instance, driving the vehicle along roads and/or highways following a selected route. For such maneuvers, there is in particular a need to reliably and efficiently detect locations where the area that can be used by the vehicle to run on, e.g. the road or highway, diverges into branches or where such branches converge. Typical examples of such locations are a highway exit, where an exit lane branches from the remaining lanes, or a highway entry, where an entry lane converges with other lanes. Other examples may be found at highway interchanges, with some lanes leading towards one destination and others towards another. Lane or road marks, for example, may often be applied on the surface of the road or highway using a suitable kind of paint or the like. Such lane or road marks, as well as traffic signs, are commonly used to provide information to a human driver in a conventional manner. The layout of roads, including the direction of traffic and the directions in which e.g. exits branch from the main road, as well as the presence, absence, sizes, colours and design of markings and traffic signs may be quite different from one country to another, and may also vary within one country from one road type to another. Also, depending on the specific situation, markings and signs may sometimes be present or absent in a manner different from the usual way. Moreover, markings and signs may be dirty, damaged and/or worn, depending on conditions such as the age thereof, the traffic intensity or the climate. It would be desirable to reliably detect locations where an area, such as a road or highway, diverges into branches, as e.g. in an exit situation, or where such branches converge, as e.g. in an entry situation. Summary Hence, it is an aspect of the invention to detect the presence of locations of the afore-mentioned kind, where branches of an area diverge or converge, without the need to rely on the presence and correctness of specific signs, e.g. traffic signs and/or road/lane markings on the ground. According to the invention, this can be accomplished by a method having the features of claim 1 and/or by an apparatus having the features of claim 13 and/or by an advanced driver assistance system having the features of claim 14 and/or by an autonomous driving vehicle having the features of claim 15 and/or by a robot system having the features of claim 16. The invention provides a method for detecting, in an environment of an object, the presence of a location where branches of an area within the environment converge or diverge, wherein the method comprises steps of: providing an occupancy grid comprising a plurality of grid cells, each grid cell comprising a value for a probability of an occupancy in the environment of the object;detecting boundaries of the area based on values of grid cells of the occupancy grid;identifying the number of boundaries detected;depending on the number of boundaries identified, signalling the presence of a location where branches of the area converge or diverge. The invention furthermore proposes an apparatus for detecting, in an environment of an object, the presence of a location where branches of an area within the environment converge or diverge. The apparatus may in particular be configured for performing the method of the present invention. The apparatus comprises: an occupancy detector adapted to provide an occupancy grid comprising a plurality of grid cells, each grid cell comprising a value for a probability of an occupancy in the environment of the object;a boundary detecting and processing device adapted to detect boundaries of the area based on values of