Method for detecting obstacles on board a machine adapted to move along at least one predetermined path

Abstract

The invention relates to an obstacle detection device for use on a train, containing geographic coordinates of the volume occupied by the train as it travels along a predetermined path and comprising a lidar (13) adapted to receive echoes of waves emitted from an object and to calculate, from said echoes, a direction and distance, relative to the train, of said object. The obstacle detection device determines the train's location; then, by means of the lidar (13), determines a direction and distance, relative to the train (1), of a detected object. Based on the train's location and the determined direction and distance of the detected object, the device determines, on the basis of said recorded geographic coordinates of the volume, whether the detected object is located within said volume and triggers an emergency action. Figure for the abstract: Fig. 1

Description

Title of the invention: Method for detecting obstacles on board a machine adapted to move along at least one predetermined path Technical field

[0001] The invention relates to the field of obstacle detection on board a machine adapted to move along at least one predetermined path, for example a train. Prior art

[0002] In an autonomous or driverless train control system for railway applications, if the signaling system fails, a driver must board the train to manually drive it, with a visual inspection, to a suitable location, for example the next station, for passenger disembarkation. This results in significant delays and compromises safety.

[0003] Existing solutions known for obstacle detection use machine learning or deep learning algorithms, which cannot be certified for safety or require cameras, whose performance may vary depending on lighting conditions.

[0004] A solution for obstacle detection that overcomes these drawbacks is therefore necessary. Summary of the invention: The solution proposed by the present invention comprises, according to a first aspect, an obstacle detection method implemented by an electronic obstacle detection device on board a machine adapted to move along at least one predetermined path; said obstacle detection device comprising: a database in which definition data are recorded, including 3D geographic coordinates of the overall volume occupied by the machine when it travels along the predetermined path; a remote sensing unit adapted to emit waves in the direction of the path extending in front of the machine, to receive echoes of the emitted waves from at least one object and to calculate, from said waves and echoes, a direction and a distance, relative to the machine, from said object; said process includes the following steps: a / determine the current 3D location of the machine; b / using the remote sensing unit, determine a direction and a distance, relative to the machine, of at least one currently detected object; c / based on the current 3D location determined, and the current direction and distance determined of the at least one detected object, determine, on the basis of said recorded 3D geographic coordinates of the volume, whether the at least one detected object is located inside said volume; d / if it is determined that at least one detected object is located inside said volume, trigger an emergency action.

[0005] This invention makes it possible to detect if an obstacle is present on the path, without it being necessary to identify the type of obstacle.

[0006] In certain embodiments, such a method will also include at least one of the following features:

[0007] - if it is determined that at least one detected object is located inside said volume, Emergency action is triggered only based on the outcome of at least one additional step among: estimate the size of the detected object based on the received echoes, and verify that said estimated size is greater than a threshold size; check the performance of the remote sensing block based on a number of points n in relation to the range r of the remote sensing and the threshold size 5 of the objects;

[0008] - if it is determined that at least one detected object is located inside said volume, The following steps are implemented before triggering any emergency action: - i / the object point inside the volume that is closest to the machine inside the considered bulk envelope is first selected, and called point 1; - ii / the number m of points inside the volume located at a maximum of a predefined distance s from the nearest selected point is determined; - iii / if m + 1 < np: pointl is not considered to belong to a relevant obstacle: pointl is eliminated, no emergency action is triggered, and the next nearest point is then examined by the algorithm repeating from step i; n being the predefined number of theoretically reflected points and p being the percentage of reflected points perceived with a given level of integrity; otherwise, if m + 1 > np, pointl is considered to belong to a relevant obstacle and an emergency action is triggered;

[0009] - if the path along which the machine moves subsequently divides into at least two alternative paths, the database containing definition data, including 3D geographic coordinates, of the overall volume occupied by the machine when it travels along each of the alternative paths: - The situation of alternative paths is detected based on the current determined 3D location and the 3D geographic coordinates of the two global volumes occupied by the machine when it travels along at least two alternative paths; and - steps c and d are then completed for at least two alternative paths;

[0010] - the machine being a train and the predetermined path being a railway track.

[0011] According to another aspect, the invention relates to a computer program adapted to be stored in the memory of an obstacle detection device comprising, furthermore, a microcomputer, said computer program comprising instructions which, when executed on the microcomputer, implement the steps of a process according to the previous aspect of the invention.

[0012] In another aspect, the invention relates to an obstacle detection device suitable for being placed on board a machine adapted to move along at least one predetermined path; said obstacle detection device comprising: a database in which definition data are recorded, including 3D geographical coordinates, of the overall volume occupied by the machine when it moves along the predetermined path; a remote sensing unit adapted to emit waves in the direction of the path extending in front of the machine, to receive echoes of the emitted waves from at least one object and to calculate, from said waves and echoes, a direction and a distance, relative to the machine, from said object; said method being adapted for the performance of the following operations: a / determine the current 3D location of the machine; b / using the remote sensing unit, determine a direction and a distance, relative to the machine, of at least one currently detected object; c / based on the current 3D location determined, and the current direction and distance determined of the at least one detected object, determine, on the basis of said recorded 3D geographic coordinates of the volume, whether the at least one detected object is located inside said volume; d / if it is determined that at least one detected object is located inside said volume, trigger an emergency action.

[0013] In certain embodiments, such a device will also include at least one of the following features:

[0014] - the obstacle detection device is, if it is determined that at least one object detected is located within said volume, adapted to trigger an emergency action based on the result of at least one additional operation performed by the device from among: estimate the size of the detected object based on the received echoes, and verify that said estimated size is greater than a threshold size; verify the performance of the remote sensing unit based on a number of points n compared to the range r of remote sensing and the threshold size s of objects;

[0015] - the obstacle detection device is, if it is determined that at least one object detected is located within said volume, adapted to perform the following operations before triggering any emergency action: - i / the object point inside the volume that is closest to the machine inside the considered bulk envelope is first selected, and called point 1; - ii / the number m of points inside the volume located at a maximum of a predefined distance s from the nearest selected point is determined; - iii / if m + 1 < np: pointl is not considered to belong to a relevant obstacle: pointl is eliminated, no emergency action is triggered, and the next nearest point is then examined by the algorithm repeating from operation i; n being the predefined number of theoretically reflected points and p being the percentage of reflected points perceived with a given level of integrity; otherwise, if m + 1 > np, pointl is considered to belong to a relevant obstacle and an emergency action is triggered;

[0016] - the database contains definition data, including coordinates 3D geographic coordinates of the global volume occupied by the machine when traveling along each of the two alternative paths, and the obstacle detection device is, if the path on which the machine travels subsequently divides into at least two alternative paths, adapted to detect the situation of alternative paths on the basis of the current determined 3D location and the 3D geographic coordinates of the two global volumes occupied by the machine when traveling along the at least two alternative paths; and to then perform operations c and d for the at least two alternative paths. Brief description of the drawings:

[0017] The invention will be better understood and other features, details and advantages will become more apparent from the following description, which is not exhaustive, and from the accompanying figures, which are provided by way of example.

[0018] [Fig-1] Fig. 1 is a schematic view of a processing device in a embodiment of the invention.

[0019] [Fig.2] Fig.2 represents the steps of an obstacle detection process in a embodiment of the invention.

[0020] [Fig. 3] Fig. 3 is an illustration of a train situation implementing a a processing device according to an embodiment of the invention traveling on a railway track.

[0021] [Fig.4] Fig.4 illustrates an embodiment in which two lidars having Different fields of vision are used.

[0022] [Fig.5] The [Fig.5] is a top view illustration of a switching situation in an embodiment of the invention.

[0023] [Fig.6] The [Fig.6] represents a view in a plane P of a train clearance gauge.

[0024] [Fig.7] Fig.7 illustrates the construction of the train's envelope in one embodiment of the invention.

[0025] Identical references may be used in different figures when they indicate identical or comparable elements. Description of the embodiments

[0026] The invention is described below with respect to an embodiment with a train 1. The train 1 is adapted to run on a railway network comprising a plurality of tracks. Figures 3 and 5 show schematic views of the train 1 running on sections of the railway network. The train 1 is driven manually or autonomously.

[0027] Train 1 includes, for example, a locomotive facing the front part of the railway track that will be travelled by the train and pulling wagons of the train.

[0028] Fig. 1 represents an electronic processing device 10 on board train 1. The processing device 10 includes a database 11, a localization block 12, a lidar block 13 and an obstacle detection block 14.

[0029] The localization block 12 is adapted to determine the location of the train 1 over an interval Tl (for example, Tl is in the range of 100 ms to 1 s depending on the obstacle detection block 14 and the requirement). The localization block 12 includes, for example, a satellite receiver adapted to determine the location of the satellite receiver based on localization signals including a known code from satellites, and / or includes a rangefinder and / or an inertial measurement unit. The localization block 12 provides, for example, the absolute 3D coordinates.

[0030] The lidar block (from the English "Laser Imaging Detection and Ranging") 13 is installed on the front of the locomotive, and faces the part of the railway track towards which the train 1 is heading, as illustrated in [Fig.3].

[0031] The lidar block 13, as is known, comprises at least one laser source adapted to emit laser pulses and includes sensors. Emitted laser pulses, when they encounter objects, are reflected by these objects: some of these echoes are received by the lidar block 13 and captured by the sensors. Based on the received echoes measured by the sensors, the lidar block 13 is adapted to measure the travel time of the laser echoes and to calculate the distance from the source to each reflecting object and also the direction of this object relative to a reference axis, for example, the longitudinal axis of the front of the train. A 3D image of the objects in the field The vision of lidar block 13 can thus be obtained according to an interval T2 (for example T2 is within the range from 50 ms to 3 s).

[0032] The laser source is installed - and the lidar emission is tuned - so that the median axis of the lidar field of view coincides with the median axis of the railway track when the railway track in front of the train is straight.

[0033] The scanning speed of the lidar block 13 influences the number of points and echoes that are measured. The choice of optics and scanner has a significant influence on the resolution and range of the lidar system. The lidar range is the length of the area in front of the train that can be monitored by means of the laser waves.

[0034] Database 11 contains data revealing the topology of the railway network, allowing the 3D coordinates of each point on the railway track to be known, or at least a precise approximation.

[0035] For example, each railway track being represented by the median axis between the right and left rails of the railway track: - the median axis having been segmented into successive portions, each portion was modeled by a segment; the succession of segments is recorded in the database, for example by chaining their identifier, the shape and 3D geographic coordinates of the endpoint are known and recorded in the database 11; and / or - the database 11 contains the 3D coordinates of a set of points along the median axis, the succession of these points being indicated in the database 11, for example by chaining their identifier, and the distance between each point and the next point being equal to or less than a predetermined distance d; for example, d is in the range from 0.5 m to 4 m; d is, for example, chosen so that it is equal to 1 meter (m); for example, this embodiment is considered below: [Fig.6] illustrates, in a given plane P perpendicular to a plane of the railway, the point 20 in the median axis between rail 21 and rail 22 of the railway.

[0036] Database 11 also contains data defining a clearance envelope for the train along each railway track. The determination of this data will now be described with reference to [Fig. 2], illustrating the steps performed in an embodiment of an obstacle detection method 100 according to the invention.

[0037] Let P be a given plane perpendicular to a segment of railway track. The train clearance gauge (plane clearance gauge) is the finite surface, indicated by the reference S_clsd, consisting of, or containing, any point in said given plane P which intersects with the train when the train, from one end to the other, crosses this plane on the railway track.

[0038] In one embodiment, the surface in question is modeled by a geometric shape a simple metric and / or is established as being slightly larger than the outline of the set of intersection points in said given plane. The margin between the surface in question and the outline of the intersection points is given by the infrastructure owner in order to guarantee the train clearance from the construction phase.

[0039] In the considered embodiment represented in [Fig.6], the contour of the gauge of train 30 illustrated in plane P, as modeled, is defined by the successive segments S3o_iS3o_2, S3o_2S3o_3, S3o_3S3o_4, S3o_4S3o_5, S3o_sS3o_6, S3o_eS3o_i between the 6 vertices S30_i, S30_2, S30 3, S30 4, S30_5 and S30_6- The modeled surface S_clsd corresponds to the hatched area delimited by these segments.

[0040] The margin mentioned above must be defined in such a way that, for example, when a train 1 enters a tunnel, the walls of the tunnel are outside the surface S_clsd or that known objects close to the envelope of the train gauge, for example a platform, are not detected as obstacles in step 100_3 described below (in order to limit irrelevant obstacle detection events).

[0041] Of course, it is possible to choose a number of vertices defining the plane gauge of the train's footprint other than 6.

[0042] In a preliminary step 100_l, as seen in [Fig.2], the definition data, including its 3D geographic coordinates, of the overall volume occupied by the train when it travels along the entire railway (from the starting point to the final destination point) are determined on the basis of the topographic data of the railway and the gauge plan clearance 30 and the definition data of the determined volume, including its 3D geographic coordinates, are recorded in the database 11.

[0043] In this regard, in one embodiment, the coordinates of the bulk envelope of the train 31 delimiting said volume are determined by deterministic algorithms and recorded.

[0044] For example, as illustrated in [Fig. 7], in order to limit the size of the recorded data, the 3D coordinates of the 6 vertices of the clearance gauge 30 are calculated when it is placed at each of the points 20 of the set, perpendicular to the segment defined by said point and the next point of the set of points along the direction of the train. This makes it possible to obtain a sample of the clearance envelope of the train. The envelope is obtained by connecting each point of the contour of the clearance gauge 30 to the corresponding point in the next clearance gauge (for example, by connecting point S3o_4 of one clearance gauge to point S30_4 of the next clearance gauge).

[0045] On [Fig.7], the overall volume, Vol, occupied by train 1 is shown in grey.

[0046] The absolute positions of the vertices of all these sets of 6 vertices (1 together all the d meters) are recorded in database 11.

[0047] The preliminary step 100_l is performed once for each train with respect to each railway track, before the obstacle detection solution defined by steps 100_2, 100_3 and 100_4 is used in operation. According to the embodiments, step 100_l is performed by obstacle detection 14 in train 1 or by an electronic block for determining the clearance envelopes in a central railway network system outside of train 1.

[0048] In one embodiment, in order to ensure compatibility with different train orientations, e.g. on a straight railway track or in a curve with a lateral profile, the sensor lever arms are taken into account when transforming the clearance contour 30 into lidar coordinates in order to determine whether the reflected points are located in the clearance contour (in step 100_3).

[0049] The obstacle detection block 14 is adapted to detect whether obstacles are present on the train's path and to trigger actions if it is indeed detected that obstacles are present, as described in detail below.

[0050] Reference is again made to [Fig. 2] illustrating the steps performed by the processing module 10 in an embodiment of an obstacle detection method 100 according to the invention. The steps 100_2 and 100_3 described below are repeated at an interval T (for example, T is in the range from 50 ms to 3 s, depending on the obstacle detection block 14 and the requirement).

[0051] During a step 100_2, when the train is moving along the railway track defined by the rails 21, 22 of the railway track, at a given obstacle detection time t (interval T;

[0052] During a substep 100_21, the obstacle detection block 14 obtains the most recent calculated 3D image of the currently scanned area from the lidar block 13. The lidar point cloud is then sliced ​​into images by the lidar block 13. The integration time depends on the required frame rate.

[0053] During a substep 100_22 parallel to substep 100_21, the obstacle detection block 14 obtains from the localization block 12 the last determined location of train 1.

[0054] During a step 100_3 performed by the obstacle detection block 14, the coordinates of the clearance envelope in front of the train are determined from the database 11 based on the determined location of the train (and the known direction of travel on the railway track) and transformed into lidar coordinates (so as to have a single reference system). The length of the clearance envelope considered (for example, corresponding to a fixed number 31 of successive train clearance gauges 30) depends on the detection range requirements (based on the necessary braking distance and the lidar range). The obstacle detection block 14 then determines whether the objects detected by lidar are located inside or outside the volume delimited by the considered bulk envelope.

[0055] If at least one object is detected inside, during a step 100_4, an emergency action is triggered by the obstacle detection block 14: an alarm signal is generated, and / or emergency / service braking is initiated (for example, if the obstacle is at a distance less than or equal to the braking distance corresponding to the current speed of the train). To this end, the obstacle detection block 14 establishes an interface with the signaling system, the rolling stock, or generates alerts for the driver.

[0056] By applying the method according to the invention, and with reference to [Fig.3], the detection of the object El, located outside the volume described by the succession 31 of planar templates of footprint 30, will not give rise to an emergency action, while the object E2, which is located inside the volume, will cause an emergency action.

[0057] In one embodiment, when considering the points of the generated 3D lidar image, one or more of the following operations are performed to determine whether some of these points belong to an object that is indeed an obstacle requiring the triggering of step 100_4: a minimum size (5) of the object (or the part of the object located inside the volume) is predefined: block detection estimates a size of the detected object based on the lidar image and, if the size is less than s, the detected object is not considered to be a relevant obstacle and no emergency action is triggered; lidar performance: number of points n in relation to the lidar range (r) and the required obstacle size 5: that is theoretically, with the range r, for an object of size s, n laser beams should be received, and therefore reflected, in the form of n reflected points (but, in reality, this is not guaranteed 100%; if the lidar supplier is able to give a percentage (sufficiently proven), e.g. at least p out of n can be reflected and appear on the lidar image, it is possible to know that with the range r, for an object of size s, we can have at least np points on the lidar image). lidar failure mode: percentage (p) of points (relative to a detected object of size s) reflected and perceived with a certain level of integrity (z).

[0058] Assuming that the objective is to detect obstacles that present a certain degree of intrusion into the train's clearance, for example, the following algorithm is used: The object point of the 3D lidar image, located inside the volume, which is closest to the train (point 1 of the 3D lidar image) inside the considered bulk envelope, is selected first; the number of points (m) inside the envelope located close together (i.e. located at a maximum distance of 5) from the nearest selected point is determined; if m + 1 < np: point 1 is not considered to belong to a relevant obstacle: point 1 is eliminated, and the next nearest point is then examined by the algorithm. Otherwise, it is considered to belong to a relevant obstacle.

[0059] It should be noted that: - p depends on r and i and the specific lidar used; - n depends on r.

[0060] This configuration helps to avoid "dust" or noise. To prevent machine learning, aggregation within the lidar point cloud is avoided. A threshold for the number of reflected points is defined to determine whether it is an obstacle or interference (e.g., dust).

[0061] Management of a nearby switch

[0062] In one embodiment, when a switch is nearby, both railway tracks will be scanned for obstacle detection according to the invention. As illustrated in [Fig. 5], train 1 moves from one railway track (rails 21, 22) to another section of track, which can be a first railway track (21_1, 22_1) or a second railway track (21_2, 22_2); the two alternative paths are known in database 11; the train's clearance profile for both railway tracks after a switch is thus taken into account by the obstacle detection method: both railway tracks are subject to obstacle detection. Because if the train locates itself autonomously (independently of trackside equipment), it is not known which track the train will take after the switch. Furthermore, when the train is at the switch, the location can still be ambiguous.Once the train has passed the switch, its location is known, and obstacle detection then only occurs on a single track. Therefore, the invention operates with autonomous train localization, independent of trackside information.

[0063] The determination of the train's location by the locating block 10 corresponds, in one embodiment, to a required level of safety.

[0064] In one embodiment, the information contained in database 11 corresponds to a required level of accuracy and integrity.

[0065] The demonstration of the security level can be accomplished through the following options: - if i is known, the integrity of the detection can be calculated; - if i is not known, the behavior of the lidar can be monitored by comparing the lidar detection and information from known reference points; lidar sensor redundancy.

[0066] In one embodiment, the processing device 10 comprises a microprocessor and a memory comprising instructions which, when executed by the microprocessor, lead to the completion of one or more of the steps 100_2 to 100_4. Alternatively, at least some of the steps can be accomplished by specialized hardware, commonly a digital integrated circuit, either specific (ASIC), or based on programmable logic (e.g. FPGA).

[0067] The obstacle detected according to the invention as described above can be a train, a human or any other object.

[0068] Due to lidar performance limitations, the wider the field of view (FoV), the lower the point density. Consequently, it may not be possible to find a lidar that covers both distance and width. Therefore, multiple lidars can be used. In one embodiment, at least two lidars are used, instead of just one, for long-range (FoV1) and short-range (FoV2) respectively, in order to cover the total distance, as illustrated in [Fig. 4]. The advantages include the fact that both width and distance are covered and that the FoV overlap space can be detected with both / all lidars so as to have independent detection chains to facilitate the demonstration of the safety level.

[0069] The invention offers a solution that can be certified in terms of safety for obstacle detection in a railway context, taking into account the speed of the train, the size of the obstacle and the performance of the lidar.

[0070] The solution is useful in manual driving mode and autonomous driving mode.

[0071] Lighting conditions have no effect on the detection results.

[0072] The solution enables the detection of objects of a certain size up to a defined distance, depending on the lidar's performance (range, point density, etc.). In one embodiment, the lidar's range covers the braking distance for the corresponding speed.

[0073] When a train is in manual driving mode, based on the speed limit, in the worst-case scenario, the emergency braking distance is defined as the obstacle detection range and therefore the range of the lidar unit. This is intended to ensure that the train will be able to stop before reaching the obstacle.

[0074] If the speed limit is 5 km / h, e.g. in the scenario of a train coupling, the braking distance can be 10 m.

[0075] If the speed limit is 15 km / h, e.g. the restricted manual mode speed limit imposed by some metro operators, the braking distance can be 40 m.

[0076] If the speed limit is 25 km / h, e.g. the Restricted Manual mode speed limit imposed by some metro operators, the braking distance can be 80 m.

[0077] If the speed limit is 40 km / h, e.g. the speed limit of Staff Res- mode responsible (responsible personnel) imposed by ETCS or other main line signalling systems, the braking distance may be 170 m.

[0078] Only deterministic algorithms are used according to the embodiments of the invention (no need for machine learning or deep learning), which makes the demonstration of the level of security possible, demonstration here meaning a method to demonstrate the level of security (the risk rate can be calculated).

[0079] Lidar failure modes are defined for the specific lidar by the lidar block supplier. Alternatively, lidar monitoring functions can be implemented in the processing device 10.

[0080] The invention has been disclosed above with the use of a lidar block. Other technologies may be used instead of lidar, for example radar or sonar technology, or any suitable technology using the detection of echoes of waves generated on board the train.

[0081] The invention has been disclosed above in relation to a train, but it can be used more generally with any machine adapted to move along any trajectory among a set of known trajectories, such a machine being, for example, a metro, a tram, a boat, an airplane, a drone, with or without automatic and autonomous driving.

Claims

Claims

1. A method of detecting obstacles (E1, E2) performed by an electronic obstacle detection device (10) on board a machine (1) adapted to move along at least one predetermined path (21, 22); said obstacle detection device (10) comprising: a database (11) in which are recorded definition data, including 3D geographical coordinates, of the overall volume (31) occupied by the machine (1) when it moves along the predetermined path; a remote sensing block (13) adapted to emit waves in the direction of the path extending in front of the machine, to receive echoes of the emitted waves coming from at least one object and to calculate, from said waves and said echoes, a direction and a distance, relative to the machine (1), of said object; said method comprising the following steps: - a / determining the current 3D location of the machine (1);- b / by means of the remote sensing block (13), determining a direction and a distance, relative to the machine (1), of at least one currently detected object; - c / based on the determined current 3D location, and the determined current direction and distance of the at least one detected object, determining, on the basis of said recorded 3D geographical coordinates of the volume, whether the at least one detected object is located inside said volume; - d / if it is determined that the at least one detected object is located inside said volume, triggering an emergency action.;

2. A method for detecting obstacles (E1, E2) according to claim 1, wherein, if it is determined that at least one detected object is located inside said volume (31), an emergency action is triggered based on the result of at least one additional step among: - estimating a size of the detected object on the basis of the received echoes, and verifying that said estimated size is greater than a threshold size; - verifying the performance of the remote sensing block (13) on the basis of a number of points n with respect to the range r of the tele- detection and threshold size of objects.

3. An obstacle detection method according to claim 1 or 2, wherein, if it is determined that at least one detected object is located inside said volume (31), the following steps are performed before triggering any emergency action: - i / the object point inside the volume which is closest to the machine (1) inside the considered space envelope is first selected, and called pointl; - ii / the number m of points inside the volume located at most a predefined distance s from the selected closest point is determined; - iii / if m + 1 < n.p: pointl is not considered to belong to a relevant obstacle: pointl is eliminated, no emergency action is triggered, and the next closest point is then examined by the algorithm repeating from step i; n being the predefined number of theoretically reflected points and p being the percentage of reflected points perceived with a given integrity level; otherwise, if m + 1 > n .p, pointl is considered to belong to a relevant obstacle and an emergency action is triggered.

4. A method for detecting obstacles according to the preceding claims, wherein, if the path on which the machine (1) is traveling will subsequently divide into at least two alternative paths (21_1, 22_1, 21_2, 22_2), the database (11) containing definition data, including 3D geographical coordinates, of the overall volume occupied by the machine (1) when traveling along each of the alternative paths: - the alternative path situation is detected on the basis of the determined current 3D location and the 3D geographical coordinates of the two overall volumes occupied by the machine when traveling along the at least two alternative paths; and - steps c and d are then performed for the at least two alternative paths.

5. A method of detecting obstacles according to the preceding claims, wherein the machine (1) is a train and the predetermined path (21, 22) is a railway track.

6. Computer program adapted to be stored in memory of an obstacle detection device (10) further comprising a microcomputer, said computer program comprising instructions which, when executed on the microcomputer, cause the steps of a method according to one of the preceding claims to be carried out.

7. Obstacle detection device (10) intended to be placed on board a machine (1) adapted to move on at least one predetermined path (21, 22); said obstacle detection device (10) comprising: a database (11) in which are recorded definition data, including 3D geographic coordinates, of the overall volume (31) occupied by the machine when it travels along the predetermined path; a remote sensing unit (13) adapted to emit waves towards the path extending in front of the machine (1), to receive echoes of the emitted waves coming from at least one object and to calculate, from said waves and said echoes, a direction and a distance, relative to the machine, of said object; said method being suitable for carrying out the following operations: a / determine the current 3D location of the machine (1); b / by means of the remote sensing block (13), determine a direction and a distance, relative to the train, of at least one object currently detected; c / based on the determined current 3D location, and the determined current direction and distance of the at least one detected object, determining, on the basis of said recorded 3D geographic coordinates of the volume (31), whether the at least one detected object is located inside said volume; d / if it is determined that the at least one detected object is located inside said volume, trigger an emergency action.

8. Obstacle detection device (10) according to claim 7, capable, if it is determined that at least one detected object is located inside said volume (31), of triggering an emergency action depending on the result of at least one additional operation performed by the device (10) among: estimating a size of the detected object based on the received echoes, and verifying that said estimated size is greater than a threshold size; check the performance of the remote sensing block based on a number of points n in relation to the remote sensing range r and the threshold size 5 of the objects.

9. Obstacle detection device (10) according to claim 7 or 8, capable, if it is determined that at least one detected object is located inside said volume (31), of performing the following operations before triggering any emergency action: - i / the object point inside the volume which is closest to the machine (1) inside the considered space envelope is first selected, and called pointl; - ii / the number m of points inside the volume located at most at a predefined distance s from the nearest selected point is determined; - iii / if m + 1 < np: pointl is not considered to belong to a relevant obstacle: pointl is eliminated, no emergency action is triggered, and the next closest point is then examined by the algorithm repeating itself from operation i; n being the predefined number of theoretically reflected points and p being the percentage of reflected points perceived with a given integrity level; otherwise, if m + 1 > np, pointl is considered to belong to a relevant obstacle and an emergency action is triggered.

10. An obstacle detection device (10) according to any one of the preceding claims 7 to 9, wherein the database (11) contains definition data, including 3D geographical coordinates, of the overall volume occupied by the machine when traveling along each of the alternative paths, the obstacle detection device (10) being, if the path (21, 22) on which the machine is traveling subsequently divides into at least two alternative paths (21_1, 22_1, 21_2, 22_2), adapted to detect the situation of alternative paths on the basis of the determined current 3D location and the 3D geographical coordinates of the two overall volumes (31) occupied by the machine when traveling along the at least two alternative paths; and to perform operations c and d for the at least two alternative paths.

Images