FR2912234A1 - Remote controlled or self guided traveling object's i.e. vehicle, navigation determining method, involves searching viable trajectories which prohibit obstacle, while reaching target, and selecting and following effective trajectory - Google Patents

Abstract

The method involves digitizing a predetermined non strengthening target (O) and an obstacle (xo), and searching viable trajectories (xa-xc) which prohibit the obstacle, while reaching the target. An effective trajectory is selected and followed, where the target and obstacle are in the form of map represented in a grid or matrix (61) of vectors of a space of remote controlled and self guided traveling object states. An independent claim is also included for a navigation system comprising a memory module.

Description

The invention relates to navigation systems, steering systems or

  guidance to a target or objective of mobile craft such as aircraft, drones, ships, vehicles, land or marine robots. The navigation of a mobile is the art of driving from one point to another, according to a predetermined route, either manually through a pilot, or automatically by autoguiding or remote control. The steering consists in keeping the mobile on a trajectory coinciding with the predetermined route, and the guidance consists more precisely in keeping it on a desired trajectory, often of a ballistic nature. For example, the navigation systems for determining the route of the aircraft, the automatic pilots of the aircrafts for the aid of driving without visibility or for the piloting of the drones, and finally the missile guidance systems 15 according to a predetermined trajectory towards a target of known geographic coordinates. These systems comprise autonomous, generally hybrid, navigation means with inertial navigation means coupled to position and / or speed resetting means, thanks to data provided by other sensors on the vehicle (odometers, Doppler radar). or transmitted by radio, for example by GPS satellites (global positioning system) or systems of terrestrial radio guidance beacons (LORAN, TACAN, ILS, ...). All these means are of course implemented on onboard digital computers constituting navigation / control systems and if the navigation and registration information are continuous analog information, they are, at one time or another, converted into data. before use by the navigation system software. The navigation software used generally fall within the domain of automatisms, responding to physical laws modeled as difference equations and continuously regulated by means of filtering.

2

  digital, and even in the field of auto-adaptive automation, that is to say, able to adapt during the piloting or the mission, since able to calibrate automatically in real time the modeling parameters of the physical laws used.

  The automatic readjustments and calibrations above are generally obtained thanks to the implementation of predictive stochastic filters (that is to say, responding to predictive statistical laws), these filters being integrated into the mathematical models of the physical laws below. evoked above. These systems, which could be described as deterministic systems with stochastic regulation, are not in themselves able to avoid an obstacle, not predetermined, whose occurrence does not obey any statistical regularity, in other words said tychastique, and thus to guarantee to reach a goal not necessarily predetermined. By not predetermined, when speaking of an obstacle or an objective, one understands the fact that their position or their dynamic parameters are tainted with an uncertainty without necessarily of statistical regularity, because subjected to unpredictable perturbations called tyches, as will be recalled later. If these systems detect such an obstacle, they can possibly change their trajectory but can not change their route to circumvent the obstacle and continue to fulfill their mission. In addition, we have recently conducted so-called viability mathematical studies, notably at the Laboratory of Applications of Regulated Tychases Systems, www.lastre.asso.fr on techniques for regulating the flow of vehicles in a road network, or aircraft in an air network, presenting risks of traffic congestion, or the regulation of driving robots to rally targets while avoiding impromptu obstacles, random presence.

  The problem is to control a robot that must evolve in a rugged environment and strewn with obstacles to reach a target or a goal. Since the target 30 is contained in an environment and an evolutionary system, it is sought: 1. the core of viability of the environment with respect to the evolutionary system, that is to say the subset of the states of the environment to from which at least one viable evolution develops, s 2. the capture basin of the target by the evolutionary system, that is to say the subset of the states of the environment from which at least one evolution starts viable in the environment and reaching the target in finite time unimpeded by unavoidable obstacles. Sustainability algorithms calculate viability nuclei and catch basins as well as viable evolutions capturing the target if needed.

  It is then in any case to regulate or control, by what are known as regulons, the evolutions of trajectories of mobile machines, whereas these trajectories are bumped up with fires. Briefly, the mathematical theory of viability relates to differential inclusions, the differential calculus of correspondences and, in metric spaces, the theory of control and differential games, the numerical and algorithmic analysis, the morphological equations governing evolution of forms and making it possible to analyze the co-evolution of variables and constraints, impulse differential inclusions, inertial optimization, algebra and tensor dynamics.

  Systems implementing viability theory may be referred to as tychastic systems.

  The applicant then had the idea to apply the viability theory to the navigation of mobiles equipped with automatic navigation systems and thus to combine the stochastic deterministic systems and the tychastic systems. To this end, the invention firstly relates to a method for determining the navigation of a controlled mobile from a starting point to a goal not necessarily predetermined and having to navigate in a space comprising at least one non-predetermined obstacle, the method being characterized by the fact that, to avoid the obstacle, it is digitized as well as the objective, viable paths are sought.

4

  avoiding it but always reaching the goal, we choose and follow the best trajectory.

  The controlled mobile can be manually by a pilot who then has directions indications in real time, or a remote controlled mobile by the operator of a remote control station that has the same information calculated on the spot or on the mobile to from the detection data transmitted by the mobile.

  The mobile can still be self-guided and calculate and then operate itself automatically these same route information.

  Preferably, the objective and the obstacle are digitized in the form of maps represented in the form of a grid of vectors of the space of the states of the mobile, and in each state of the grid, a feedback applicable to driving or guiding the mobile which guides the mobile to a next state.

  The feedback associates with each state of the mobile its acceleration, its course and its incidence and is calculated to optimize predetermined criteria and the mobile being able to start from a set of starting states, a single catch basin is computed. the objective.

  Advantageously, the grid of states is gradually refined by dichotomy, and the feedback on the refined grid is calculated.

  The invention also relates to a navigation system for a self-steering mobile arranged to achieve an objective, comprising: a module for detecting and capturing environmental data characteristic of an objective or an obstacle coupled to a module for alerting said system; a memory module comprising mission data; A scanning module for lens grids and obstacles; a module for managing said grids, and a memory module containing the mobile dynamic model, for a module for generating a tychastic calculation kernel; a tychastic calculation module; A module for selecting and tabulating a feedback to be applied to the navigation of the mobile, and a stochastic navigation calculation block comprising a module for generating a navigation command derived from this feedback.

  The invention will be better understood in the light of the following description of the navigation system 5 and the method for determining the navigation commands according to the invention and the drawing accompanying it, in which: FIG. 1 is a block diagram functional navigation system according to the invention; FIG. 2 is an operating flow diagram of the navigation method 10 according to the invention; FIG. 3 is a simplified diagram of an example of a mobile state grid for calculating a viable trajectory according to the method of the invention, and FIG. 4 is a simplified diagram of the viability computation automatism. moving a state of the mobile to a next state according to the method of the invention. With reference to FIG. 1, the navigation system 200 of a mobile, here a vehicle (not shown) with autopilot, remote control or self-guided, essentially comprises:

  A) an input block 1 formed of the following modules: 1) a module 11 for detecting and capturing environmental data characteristic of the presence of a target or an obstacle, comprising a set of sensors, for example type photographic or radar, and coupled to a module 9 for alerting the system 200 on receipt of new data of the same type, 2) a module 50 for manual entry and processing of control data and data of the same type as above, or simply a read-only memory including guidance and / or mission data provided in advance; 3) possibly a radio module 95 of data download of the same type 30 as above, if the mobile is remote controlled, or if it is piloted, an alphanumeric keyboard 95, or a digital interface 95 data input capable of connecting the system to a mass memory (not shown) transmitting or containing mission data to be insured;

  B) a module 2 for digitizing and coding the data transmitted by block 1, including those concerning a target and possibly obstacles, and put under

6

  form of data introduced into maps or grids of the space of dynamic states of the mobile which will be explained later;

  C) a grid management module 3 above, comprising grid memories 31 and 32 storing the coded data of these grids, respectively concerning the mission and the environment, especially concerning the target (in the grid 31) and the obstacles (in grid 32);

  D) a module 4 for storing and restoring the mobile dynamic model io expressed in the form of motion equations and error equations known to those skilled in the art;

  E) an input interface generator module on the one hand of the grid 31 and 32 of the target or objective and obstacles and on the other hand of the dynamic model, and provided for the generation of a core classical tychastic calculus for:

  F) a tychastic calculation module 6, in particular for calculating a catch basin and, by a specific submodule 60 included in this module 6, for calculating a tychastic feedback. The tychastic module 6 also includes a memory module 20 storing the provisional results of its calculations, essentially the possible trajectories of the mobile in the phase space (space positions, speeds) of the mobile. Phase spaces are commonly used in nonlinear systems analysis and perturbation theory (Poincaré method). The phase space is here limited to the trajectories determined by the set of initial states of the mobile;

  G) a module 7 for selection and extraction in the memory 20 of the above results and then of the memory recording, in the memory of the module 7, of the capture basin and a feedback to be applied, stored in form a tabulation;

  H) A stochastic navigation calculation block 8 comprising: 1) a module 81 of navigation sensors representative of the state of the mobile at any moment, for example odometers, GPS, inertial sensors, etc., relatively to its environment and transfer this data to module 6 above,

7

  2) a module 82 for extracting the feedback from the module 7 and generating a navigation command in position, speed and acceleration derived from this feedback, 3) a stochastic navigation module 83, taking into account the command above, comprising, for example, a hybrid inertial navigator 10, or any other type of mobile browser, and a command 40 of mechanical, electric or hydraulic control or control surfaces acting on the control of the mobile, l0 4) a memory 84 containing the theoretical instantaneous state of the mobile useful to the browser 10 for extracting error data from the sensor data 81 reflecting the instantaneous real state of the mobile. This error data allows predictive hybrid stock navigation. Incidentally, they allow, if the mission data so provide, to reach the target with zero speed.

  By control or control surfaces, it is understood here all engine control devices, directional or depth, hydraulic, electrical and mechanical, to ensure the driving or steering of the mobile from its means of guiding, driving and navigation systems, for example the VCT (Flight Controls) and / or FMS (Flight Management System) aircraft.

  The memories 31 and 32 contain the descriptive data of the mobile environment in the form of a digitized geographical map supplemented with digitized states of the mobile in the form of point grids xa, xb representative of the mission of the mobile, particularly in position and speed, and allowing to work in phase space. The module 20 stores in its memory the calculated trajectories such as xa-xb.

  The module 3 also comprises, in grid 31, the navigation instructions entered manually or automatically picked up during the mission (considered as tyches by the system 200) and control rules (dependent on the mobile) or calculation of feedback ( depending on the tyche), to observe according to each state of the mobile.

8

  Navigation and route designation instructions can be preprogrammed, remotely controlled, calculated, or manually entered via the channel of module 50 or module 95.

  In particular, the block 1 can receive additional data concerning the mission of the mobile, for example complementary geographical descriptions, by the action of a pilot navigating on the alphanumeric keyboard 95, or via the radio receiver 95 receiving them. a beacon, or a removable mass storage drive such as CD-ROM or ROM (read only memory) or a GPS receiver, etc.

  Finally, the block 1 and the module 2 develop navigation instructions in the memories or grids 31 and 32 of the module 3, for example from a RAM (random access memory). The instructions can be in descriptive language directly interpretable in real time by the generator module 5.

  The feedbacks contain corrections of parameters such as acceleration, heading, angle of attack, etc., depending on the speed, the altitude or the depth of the vehicle, and the points or states of passage. The feedbacks associate with each position and speed of the mobile its acceleration, course and bearing.

  The mission data defines, for example, operational actions (civil or military actions, etc.) to be carried out at these crossing points and on the final objective.

  The browser 10 may for example be a helper inertial navigator (or hybrid) with positional registration (GPS or other radio beacon) or speed (Doppler radar, sonar or odometer) with continuous calibration of certain modeling parameters by statistical filtering.

  The detection module 50 is arranged to detect and locate obstacles on the map, and more particularly new obstacles. It is connected to the feedback module 60 arranged, him, to cause, by the calculation of feedbacks, a change of trajectory of the mobile according to the new obstacles.

  For this, the detection module 50 must include sensors for detecting new obstacles or tyches. The location of the obstacles is reported by the 25

9

  digitizing computer 2 on the digitized grids 31 and 32 containing the environment: the geographical map, the obstacles and the constraints in the phase space and in particular the initial or mission constraints and the objective O, etc.

  As sensors, depending on the type of mobile, it is possible to use radars or sonars on board, mechanical sensors (feelers), laser sensors or other telemetry devices, and the above set can be completed by the interface 95 or be short-circuited by him by the module 5 transmitting directly to the computer 60 new obstacle coordinates or new constraints, introduced in the scanned cards or grids 31, 32 during the mission.

  The alarm or warning module 9 signals the system 200 and / or the mobile that new obstacles have been detected or that the target has been modified, blocks in the module 8 the use of the feedback 7 and possibly implement remedial strategies and then reinitialize the input block 1 and the modules 1 to 7 to calculate a new feedback and trigger it to be taken into account in the module 8, as will be explained later. The warning module 9 is activated by the other modules 50 and 95 of the block 1, and is more specifically arranged to cause the calculation, by the calculation unit 6 and via the module 5, of a new feedback to be substituted. to the one previously present in the memory of the module 7. Moreover, the computer 6 is connected to the browser block 8 by the bidirectional logic link 65 to serve to coordinate their actions.

  The grids 31 and 32 form a single matrix or grid 61 of state vectors xa, 30xb, xc, ... of the mobile in the phase space, the mission and the environment, more simply a matrix map. 61 or matrix 61 of points xa, xb, xc, ...

  The points xa, xb, xc,... Comprise, in particular, geographical coordinates of a geographical map which are distant from each other by a step d depending on the navigation data, such as the positional accuracy of the mobile, its velocity relative to the time step of calculating the difference equations (or

10

  numerical differential equations) modeling the mobile guidance laws, and requirements for geographic accuracy of the mission, etc.

  For example, in the matrix 61, the obstacles are materialized by sets {xo} of points, geographically located in (in the broad sense) obstacles. The set {xl} of the possible starting points of the mobile constitute initial states. Objective O must be part of the set of points xa, xb, xc ...

  The feedback module 60 essentially comprises software for calculating viability, and uses, as input data, the calculation kernel developed by the module 5, whose gate 61 and the dynamic model of the mobile stored in memory 4 are part of .

  Indeed, the viability calculations proposed here differ substantially from the viability calculations ordinarily known in the state of the art in that a non-predetermined and specific feedback of the unexpected new stresses occurring during a period of time is computed. mission, including frontier and obstacles, tychastique type, and the objective O not necessarily predetermined, as will be described below. 20 They put into practice the theory of viability, as explained in particular in the following documents:

  AUBIN J.-P. (1991) Viability theory, Birkhuser 25 http://lastre.asso.fr/aubin/WViabTheoryComplet.pdf

  SAINT-PIERRE Patrick Approximation of the Viability Kernel - Applied Mathematics & Optimization 29: 187-209 (1994)

  30 AUBIN J.-P. & CATTE F. (2002) Bilateral Fixed-Point and Algebraic Properties of Viability Kernels and Capture Basin of Sets, Set-Valued Analysis http://lastre.asso.fr/aubin/AUB-CATTE- SVA.pdf

  The purpose of the theory of viability is to describe the mathematical processes of control-command of non-linear servomechanisms, that is to say, subjected to tyches, here avoidance of obstacles. It eliminates orders that do not

11

  do not achieve the goal before choosing among those that remain those to apply to the servomechanism.

  Here, the viability calculations carried out consist of calculations applied step by step from any point of the matrix 61 likely to be an initial point of computation, and at the beginning of the moving of the mobile, points xl of the initial constraints, to determine trajectories corrected by regulons or feedbacks to pass states xa, xb ... of the matrix 61 to states xc, xd ... with the guarantee of achieving the objective O while avoiding the set {xo} of states to obstacle and in particular the states new obstacles.

  In the considered application of execution of a mission, the set {xl} of the initial states obtained contains, if it exists (if the mission is impossible), the initial catchment basin, which one can seek at expand as we will see below. More specifically, with reference to FIG. 4, the system 200 comprises two parts S and T, a stochastic S, which is essentially executed by the block 8, and a tychastic T, essentially executed by all the elements 3, 5. 6, modifying by feedbacks or regulating the execution of the stochastic part of the mission at the appearance of each task.

  These two parts S and T work according to the master-slave scheme, the navigation block 8 being the slave of the computer 6 via the link 65. Thus the navigation calculations of the block 8 and the tychastical calculations of the calculator 6 25 can be coordinated as specified below.

  The system 200 can be summarily represented by a servomechanism which comprises an action chain 14 modeling the behavior of the mobile by linear differential equations of the variables Xi representative of this behavior and the driving parameters, functions of the time ti discretized and indexed by the index i, and: 1) a first return chain 12 or counter-reaction 12 denoted U i, stochastic, guiding the mobile according to the points xa, xb, xc, ... towards the objective O, according to the parameters steering, but without taking into account the presence of obstacle, 35 according to the dynamic and statistical laws modeled by equations of the type: X'i = f (Xi, Ui) 12

  Xi + 1 = Xi + (ti + 1 ù ti) .X'i (ti + 1 ùfi) = 1 / Fo Fo being a predetermined base frequency. 2) a second return chain 13 or feedback 13 which modifies Ui by the regulators Vi, via the above control parameters. The Vi regulators are calculated according to the existence or the appearance of new obstacles or the designation of a new objective in the grid 61.

  At the appearance of a tyche, the regulons are calculated as indicated above, Io tabulated and stored for each state xa, xb, xc of the matrix 61, so as to be recoverable at the passage of the mobile at each point xa , xb, xc ... and to allow the system 200 to change at these points, when the mobile is there, the operation of its stochastic part.

  For simplicity, the modifications may consist in incrementing Ui successively with regularized vectors Vi, replacing Ui by Ui + Vi in the above differential equations which it applies at each point xa of the mobile state; the verifications of the accessibility by Vi of a point xb, xc, ... immediately adjacent to distance d from the position of the mobile (which excludes, in particular, the diagonal points in the matrix 61) and the viability the choice of this next point to reach objective O having already been realized and being guaranteed by the viability calculations made.

  These operations are performed at each incrementation of i, and a neighbor point is stored in memory as the next point.

  We proceed step by step until we reach the objective O (which is part of the grid) or on a point already included in a viable trajectory already obtained previously. In a first viability calculation, it was possible to determine a capture basin in the matrix 61, according to the theory of viability.

  According to this theory, if during the viability calculation for a portion of a route starting with an initial state x1, a state or point xo constituting an obstacle, or another possible initial state already analyzed, or a 30

13

  point belonging to a trajectory already recognized as unsustainable, the trajectory being computed is not viable and the initial state xl is not part of the catch basin.

  Before any mission, the non-viable trajectories are automatically eliminated and the mobile starts immediately from a point xl belonging to the catch basin. But the appearance of a new obstacle can reduce this catch basin. After any viability calculation, the feedback module 60 calculates a new catch basin and thus checks whether the mobile is still included therein. To widen the catch basin, matrix 61 is refined by dichotomy, and the above process is repeated.

  The dichotomy is a method of dividing all segments of the grid into two segments. The number of points of the matrix 61 is multiplied by two in its dimensions. One can also proceed by superimposing on the matrix 61 a matrix 62 equal to but offset from it by the distance d / 2 in its two horizontal and vertical directions, as in FIG. 3. In FIG. In fact, since the phase space is at least six-dimensional, this treatment must be multiplied by at least three.

  Since the overall matrix obtained is finer, there are additional points xa, xb, xc,... Offering additional possibilities of finding more and more smoothed paths to the objective O, and therefore of increasing the number of points belonging to the resulting catch basin, especially outside the initial catch basin.

  If for a feedback Vi + 1 we determine points xa, xb, xc,

  . of the trajectory 30 already included in a route obtained for Vi or an earlier feedback, we can retain Vi + 1 immediately, and we can stop the calculation of the current catch basin for the initial state considered, since we already know the following of the viable route. The calculation of trajectory is thus optimized in computation time ... DTD: 35 The method of navigation by the system 200 therefore consists, at the beginning of the mission, in trying all the initial states {xl} (the catch basin the most possible, or only when one is on a mission, the initial point xl where the mobile is at the moment of the appearance of a new obstacle {xo}, and to try all possible feedbacks Vi in each point xa, xb, xc, to retain only those leading the mobile on an objective O.

  The next set of feedback-state pairs (Vi, xa) resulting from the feedback Vi retained for all the points xa, xb, xc is then sought according to predetermined or determined criteria during the mission, for example the shortest distance or in terms of travel time, or criteria that can be transmitted in real time by a pilot via module 50 or 95.

  This determines the best feedback in real time and is instantly applied to the system 200, with the certainty of achieving the objective O.

  But one can also choose the feedback: 1) as being that which minimizes at every instant the speed of the controls to avoid their beat (chattering), 2) as being the most robust relative to the perturbations which can act on non-determined parameters of the dynamic model, targets or 20 obstacles.

  Furthermore, when a control mechanism is chosen independently of the target and the obstacles, the computer 6 calculates the catch basin of the target by a single feedback, the most robust, to provide a safety zone. Finally, if several feedbacks are chosen independently of the target and the obstacles, the computer 6 calculates their sequence and the catch basin corresponding to the sequence to provide the safety zone.

  In summary, the navigation of the controlled or guided mobile is determined by searching for a trajectory from an initial state x1 to the objective O, which is not necessarily predetermined and which must navigate in a space strewn with obstacles {xo} that are not predetermined, and for avoid obstacles, we monitor their appearance, and at each appearance of one of them, we locate it, we search the viable trajectories with the guarantee of reaching the objective O, then we choose and follow the best trajectory xa-xb, xa-xc,

15

  Finally, the module 7 chooses the best feedback from the memory 20, receives from the computer 6 its calculation results stored in the memory 20, including the feedback table and viable states (Vi, xa), and transmits the selected feedback to the module 82 for developing browser commands 83.

  If the link 65 is provided for this purpose, the radio module 95 can warn the base of the mobile (not shown) that the new obstacle is taken into account.

  The operation of the system 200 will now be explained.

  During the navigation, with reference to FIG. 2, at a step 100, the navigator 83 reads, at instant t, the instantaneous positions, speeds and accelerations from the sensors 81, comparing them to the theoretical state 84 and deduces estimated navigation errors.

  During this step, it also reads the navigation / control parameters of the trajectory xa-xb to be followed in module 7, defining in particular the heading, the cruising speed and the other navigation and mission parameters already mentioned, that they are designated either manually or remotely controlled, or in the form of regulons calculated by the computer 6 as we saw previously and in a step described later.

  In the following steps 110, 111, the navigator 83 performs a stochastic navigation, notably from the estimated errors, it carries out its conventional operations of hybrid navigation, including the recalibrations in the phase space and the calibrations of the dynamic models well known to the skilled person. The dynamic models in memory 4 are then updated.

  For this, on the trajectory xa-xb above, the navigator 10 performs a calculation or basic navigation cycle according to the predetermined base frequency Fo during which simultaneously it calibrates continuously (step 111) its modeling parameters and it calculates (step 110) the position on the map and the estimated speeds of the mobile, by hybrid navigation and Kalman filtering, using for example the mathematical methods explained in the optimal inertial navigation and statistical filtering under the direction of Pierre

16

  Faune, published by Dunod (Paris 1971), then deduces the counter reaction Ui. It then increments an index i of cycles executed at the base frequency Fo, i being zero at the beginning of the navigation.

  In step 120, the browser 10 loops on the basic cycle of steps 111, 110 otherwise, it performs a step 130.

  In step 130, it checks the result of the navigation carried out until now by searching for the correspondence of the coordinates of the mobile and its navigation parameters with those of the trajectory xa-xb to follow and deduces: the proximity L of point or state xb to reach, position E deviations from the trajectory xa-xb to follow.

  Above a certain threshold S1, E causes a control / navigation correction 15 tending to cancel E, by modifying the heading, the speed, ..., modification that can be performed in step 130 or in step 110 .

  Note that S1 substantially corresponds to the precision with which we want to reach the objective O. Below another threshold S2, which may be substantially equal to Si, the browser 10 considers that the crossing point xb is reached and modifies the navigation in progress by imposing on it the navigation parameters Ui and regulating Vi of the crossing point reached xb, that is to say the regulation-state torque (Vi, xb) stored in memory of the module 7, which torque must to reach the next point xc.

  At the end of this step, it executes the step 140 in which it tests the arrival of a possible message M1, issued by the calculator tychastique 6, and a message M2 to follow issued by the module 7, as it will be explained later. As long as there is no M2 message, it loops on hybrid navigation 111, 110.

  Otherwise the browser 10 goes to step 150 in which it replies with a message P to the message M1 and still continues the inertial navigation waiting message M2 after viability calculations are performed, as described below. 35

17

  During the stochastic navigation phase above (steps 100 to 150) of the vehicle, in step 201 in the drawing of Figure 2, the assembly 1 continuously monitors the environment described in the matrix 61.

  When at any given moment, a tychastic event occurs, the set 1 detecting for example a new obstacle, or simply if a new obstacle or new constraints occur via the modules 50 or 95, a tychastic phase is triggered, timed by the counting an index j incremented at the frequency Fo from zero, which comprises the steps 210 to 240 io explained now.

  During step 210, the tychastic event is formalized by the computer 2 in digitized data which are introduced into the memories of the module 3. The module 3 updates the matrix 61 and by the interface 5 transmits it to the tychastic calculator 6, simultaneously with the dynamic model stored in memory 4.

  In step 220, the tychastic calculator 6 warns the browser 10 by sending the message M1 via the link 65, warning that a new trajectory and new regulons must be sought. The navigator 10, following the message M1, while continuing the stochastic navigation phase, in a step 150, transmits by link 65, to the feedback module 60, the useful parameters P: position and current speeds of the vehicle, data that become , together with the data of matrix 61, the new initial conditions of the viability search.

  In step 230, the tychastic calculator 6 then has all the data necessary to perform the viability calculations above. In particular, the chases calculator 6 can calculate a new catch basin and, if the vehicle is not present, trigger a specific action concerning its mission.

  For example, since it can not guarantee to reach the objective O, it can cause the mission to be stopped or the generation of a radio link requesting alternative instructions. 20

18

  But it can also seek to expand the catchment area as seen above.

  Once the feedback module 60 has determined new trajectories bypassing the new obstacles or taking into account the new constraints, but still leading to the objective O imposed at the beginning of the mission, or the new objective O, it looks for other solutions of trajectory continuity meeting the same conditions, as far as it can find in a certain predetermined calculation time T corresponding to the allowable autonomy of the stochastic navigation phase above.

  In practice, he compares the index j with a maximum value J such that: J = Fo.T

  He can then look for finer solutions by matrix dichotomy 61 as seen above. It stores them in memory 20.

  During step 240, the tychastic calculator 6 then searches for the best solution according to a predetermined criterion, the time required or distance to travel to reach the objective O. It is during this last step 240 that it transmits this solution to the module 7 which substitutes the corresponding data to those present in the memory 20 and the transcode before transmitting it to the module 82.

  At the time of the substitution, the module 7 warns the browser 10 by the message M2 (via the link 65) that it will shortly be operational, and the module 95 that the new obstacle has been taken into account, the latter being then able to warn the base of the mobile.

  The navigator 10, in the step 150, reads M2, can calculate a connection trajectory between the position and the current speeds of the vehicle at the moment of the substitution and the position and the speed formalized in P corresponding to the initial conditions of the search. viability. This can be done using an interpolation technique between the corresponding parameters. 35

19

  In the next step 160, the browser 10 decrees the end of the viability calculations at the termination of the connection calculations above, and loops on the initial process of reading of the step 100 taking into account the new data in the memory 20 of the module 6 (or the memory of the module 7).

  Since at the time of reception of the message M1, the old route is destined to be abandoned, it is not necessary for the browser 10 to maintain the accuracy of its monitoring or to carry out statistical regulation (estimation of the error, calibration, resetting) on the navigation during the execution of the calculations of viability of viability. It can therefore continue in pure inertial navigation, and, if the computers 6 and 10 are one, allow to realize substantial savings in computing time.

Claims (16)

  A method for determining the navigation of a controlled mobile from a point (xl) of departure to a goal (0) not necessarily predetermined and having to navigate in a space with at least one obstacle (xo) not predetermined , the method being characterized by the fact that, to avoid the obstacle, it is digitized as well as the objective, viable paths are sought (xa, xb, xc, ...) avoiding it but always reaching the objective (0), we choose and follow the best trajectory (xa, xb, xc,   2. The method of claim 1, wherein the mobile is remote control.   3. The method of claim 1, wherein the mobile is self-guided.   4- Method according to one of claims 1 to 3, wherein the objective and the obstacle are digitized in the form of cards represented in the form of a grid (61) of vectors (xa, xb, xc, ... ) of the state space of the mobile.   5. The method of claim 4, wherein, in each state (xa, xb, xc, ...) of the grid (61), a feedback (Vi) applicable to driving or guiding the mobile which guides the mobile to a next state.   6. The method of claim 5, wherein the feedback (Vi) associates with each state of the mobile its acceleration, its course and its incidence.   7- Method according to one of claims 5 and 6, wherein the feedback is calculated to optimize predetermined criteria.   8- Method according to one of claims 1 to 7, wherein, the mobile being capable of starting from a set of start states (xl), calculating a single catch basin (xl) of the objective ( 0).   9- Method according to claim 8, wherein the grid (61) of the dichotomous states is progressively refined, and the feedback on the refined grid is calculated.   10- Method according to one of claims 1 to 9, wherein the best trajectory is chosen as being that which minimizes at any time the speed of the controls to prevent their beat.   11- System (200) for navigation of a self-steering mobile arranged to achieve an objective (0), comprising: a module (11) for detecting and capturing environmental data characteristic of an objective or of an obstacle coupled to a module (9) for alerting said system; a memory module (50) comprising mission data; 21 - a module (2) for scanning in grids (31, 32) of the objective and obstacles; - a module (3) for managing said grids (31, 32), and - a memory module (4) containing the mobile dynamic model, for - a module (5) for generating a tychastic calculation core; a module (6) for calculating tychastics; a module (7) for selecting and tabulating a feedback to be applied to the navigation of the mobile, and a stochastic navigation calculation block (8) comprising a module (82) for generating a navigation command. derived from this feedback. to   12- System according to claim 11, comprising a radio interface (95).   13- System according to one of claims 11 and 12, comprising an alphanumeric keyboard (95).   14-System according to one of claims 11 to 13, comprising a mass memory (95). 15   15-System according to claim 12, comprising a security module (90) arranged to transmit to the module (95) the feedbacks calculated by the tychastique module (6) and warn that the obstacles have been taken into account.   16-System according to one of claims 11 to 15, comprising a logic link (65) connecting the tychastique module (6) and the stochastic navigation calculation block (8), and arranged to coordinate their calculations.