Dynamic scene analysis method, and associated analysis module and computer programme

• US 10,290,116 B2
• Filed: 06/02/2015
• Issued: 05/14/2019
• Est. Priority Date: 06/06/2014
• Status: Active Grant

• Current Assignee: Commissariat a L'Energie Atomique (AES Corporation), Centre National De La Recherche Scientifique, Inria Institut National de Recherche EN Informatique et EN Automatique
• Original Assignee: Commissariat a L'Energie Atomique (AES Corporation), Centre National De La Recherche Scientifique, Inria Institut National de Recherche EN Informatique et EN Automatique
• Inventors: Laugier, Christian, Negre, Amaury, Perrollaz, Mathias, Rummelhard, Lukas
• Primary Examiner(s): Hsieh, Ping Y
• Assistant Examiner(s): Torres, Jose
• Application Number:
  US15/316,779
• Publication Number:
  US 20170256071A1
• Time in Patent Office: 1,442 Days
• CPC Class Codes: B60T 2201/022 Collision avoidance systems
• CPC Class Codes: B60T 7/22 initiated by contact of vehicle, e.g. bumper, with an external object, e.g. another vehicle , or by means of contactless obstacle detectors mounted on the vehicle
• CPC Class Codes: B60W 2420/403 Image sensing, e.g. optical camera
• CPC Class Codes: B60W 40/04 Traffic conditions
• CPC Class Codes: B60W 40/12 related to parameters of the vehicle itself , e.g. tyre models
• CPC Class Codes: G01C 21/3807 characterised by the type of data
• CPC Class Codes: G01C 21/3833 characterised by the source of data
• CPC Class Codes: G01C 21/387 Organisation of map data, e.g. version management or database structures
• CPC Class Codes: G01S 17/87 Combinations of systems using electromagnetic waves other than radio waves
• CPC Class Codes: G01S 17/931 of land vehicles
• CPC Class Codes: G06F 17/18 for evaluating statistical data , e.g. average values, frequency distributions, probability functions, regression analysis forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks
• CPC Class Codes: G06T 2207/10016 Video; Image sequence
• CPC Class Codes: G06T 2207/20076 Probabilistic image processing
• CPC Class Codes: G06T 7/20 Analysis of motion motion estimation for coding, decoding, compressing or decompressing digital video signals H04N19/43, H04N19/51
• CPC Class Codes: G06T 7/277 involving stochastic approaches, e.g. using Kalman filters
• CPC Class Codes: G06T 7/70 Determining position or orientation of objects or cameras camera calibration G06T7/80
• CPC Class Codes: G06V 20/58 Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
• CPC Class Codes: G08G 1/04 using optical or ultrasonic detectors
• CPC Class Codes: G08G 1/166 for active traffic, e.g. moving vehicles, pedestrians, bikes

Abstract

A method for analyzing a dynamic scene partitioned into cells which involves determining a probability of occupancy of a cell and a probability or probabilities of movement of the cell by solving the equation P(OV|ZC)=ΣA0^−1V-1P(CA00⁻¹VV⁻¹Z)/ΣA00⁻¹VV⁻¹P(CA00⁻VV⁻¹Z) comprising the determination of the speeds and positions of dummy particles in the grid depending on those determined at the (k−1)^th iteration and the probability P(V|V−1); the determination of the particles located in each cell depending on the determined positions, and the solving of the equation, for a cell, is split into the solving of a static part corresponding to P(0=empty, V=0|ZC) and P(0=occupied, V=0|ZC) and the solving of a dynamic part corresponding to P(0=occ, V=v^k_i,|ZC), i=32 1 to n^k, in which n^k is the number of particles determined in cell C for the k^th iteration.

Claims

1. A method for analyzing a dynamic scene observed with the aid of one or more sensors, the method comprising:

defining a grid that is partitioned into cells and corresponding to the observed scene;

collecting at least one new observation of the one or more sensors at a k^th iteration;

determining, as a function of the new collected observation, a first probability of occupancy of each cell of the grid modeling the operation of the one or more sensors;

determining, for each cell, at the k^th iteration, a second probability of occupancy of the cell and a set of probabilities of motion of the content of the cell as a function of the first probability of occupancy of the cell determined at the k^th iteration,

wherein the second probability of occupancy of the cell and of the set of probabilities of motion of the content of the cell as a function of the first probability of occupancy of the cell determined at the k^th iteration is determined based on;

C, an identifier of the cell considered;

A, an identifier of the cell which contained, at the (k−

1)^th iteration, what is contained in the cell considered at the k^th iteration;

O, an occupancy state of the cell considered, from among the empty and occupied states;

O<sup>−

1</sup>, an occupancy state of the cell at the (k−

1)^th iteration;

V, a velocity of the cell considered;

V−

1, a velocity of the cell at the (k−

1)^th iteration; and

Z, observations of the sensors from the first iteration up to the k^th iteration;

wherein respective velocity and the respective position of a set of dummy particles in the grid are determined at the k^th iteration as a function of the velocities, positions of the particles determined at the (k−

1)^th iteration and of a probability P(V|V−

1);

the method comprises a step of determining the particles located in each cell as a function of the positions determined and in that the solving of an equation, for a cell, is split into the solving of a static part corresponding to P(0=empty, V=0|ZC) and P(O=occupied, V=0|ZC) and into the solving of a dynamic part corresponding to the P(O=occ, V=v_i^k,|ZC), i=1 to n^k, where n^k is the number of particles determined in a cell C for the k^th iteration; and

wherein the static part of the cell C at the k^th iteration being determined as a function of the static part of the cell C determined at the (k−

1)^th iteration and of P(O|O<sup>−

1</sup>);

the probability of P(O=occupied, V=v_i^k,|ZC) of the dynamic part of the cell C being determined at the k^th iteration as a function of the probability P(O=occupied, V=v_i^k-1,|ZA) calculated at the (k−

1)^th iteration for the dynamic part of a cell A and of P(O|O<sup>−

1</sup>),

where the particle p_i determined in the cell C at the k^th iteration with a velocity v_i^k-1 was situated in the cell A at the (k−

1)^th iteration with a velocity v_i^k-1, and on completion of the k^th iteration, one or more pairs (p_i, (v_i^k, x_i^k)), where x_i^k the position of the particle p_i at the k^th iteration, are duplicated within a cell or deleted so that the number of pairs per cell is dependent on the dynamic part determined for the cell.

2. The method of claim 1, further comprising selecting the pair to be duplicated or deleted, wherein the selection of said pair being carried out as a function of the probability P(O=occupied, V=v_i^k|ZC) is determined at the k^th iteration, and where v_i^k is a velocity component of the pair.

3. The method of claim 1, wherein the total number of particles in the grid is constant during the iterations.

4. The method of claim 1, wherein P(O=empty, V=0|ZC) is determined at the k^th iteration, is denoted as P^k(O=empty, V=0|ZC), and is determined as a function of the product of a first term dependent on the first probability of occupancy and of a second term, said second term being dependent on:

P^k-1(O=occ,V=0|ZC).(1−

ε

)+P^k-1(O=empty,V=0|ZC).ε

,

where ε

=P(O=occ|O<sup>−

1</sup>=empty)=P(O=empty|O<sup>−

1</sup>=occ); and

/or

coeff (v_i^k).P^k-1(O=occupied, V=v_i^k-1,|ZA).(1−

ε

), where coeff (v_i^k) is a decreasing function of ∥

v_i^k∥

, and/or

a probability of appearance p_a of a new object in the observed scene.

9. A non-transitory computer accessible medium that includes computer-executable instructions stored thereon that are executable by a computing device to perform the method of claim 1.

5. A device effective to analyze a dynamic scene observed with the aid of one or more sensors, the device is configured to:

define a grid partitioned into cells and corresponding to the observed scene;

collect at least one new observation of the one or more sensors at a k^th iteration and determining, as a function of the new collected observation, a first probability of occupancy of each cell of the grid and that models the operation of the one or more sensors;

determine, for each cell, at the k^th iteration, a second probability of occupancy of the cell and a set of probabilities of motion of the content of the cell as a function of the first probability of occupancy of the cell determined at the k^th iteration,

wherein the device is adapted to determine said second probability of occupancy of the cell and the set of probabilities of motion of the content of the cell as a function of the first probability of occupancy of the cell determined at the k^th iteration based on;

C, an identifier of the cell considered;

A, an identifier of the cell which contained, at the (k−

1)^th iteration, what is contained in the cell considered at the k^th iteration;

O, an occupancy state of the cell considered, from among the empty and occupied states;

O<sup>−

1</sup>, an occupancy state of the cell at the (k−

1)^th iteration;

V, a velocity of the cell considered;

V−

1, a velocity of the cell at the (k−

1)^th iteration;

Z, an observations of the sensors from the first iteration up to the k^th iteration;

wherein the device is further configured to;

determine, at the k^th iteration, the respective velocity and the respective position of a set of dummy particles in the grid as a function of the velocities, of the positions of the particles determined at the (k−

1)^th iteration and of the probability P(V|V−

1);

determine particles located in each cell as a function of the positions determined and to split the solving of an equation, for a cell, into the solving of a static part corresponding to P(O=empty, V=0|ZC) and P(O=occupied, V=0|ZC) and into the solving of a dynamic part corresponding to the P(O=occ, V=v_i^k,|ZC), i=1 to n^k, where n^k is the number of particles determined in the cell C for the k^th iteration;

determine the static part of the cell C at the k^th iteration as a function of the static part of a cell C determined at the (k−

1)^th iteration and of P(O|O<sup>−

1</sup>); and

determine the probability of P(O=occupied, V=v_i^k,|ZC) of the dynamic part of the cell C at the k^th iteration as a function of the probability P(O=occupied, V=v_i^k-1,|ZA) calculated at the (k−

1)^th iteration for the dynamic part of a cell A and of P(O|O<sup>−

1</sup>),

where the particle p_i determined in the cell C at the k^th iteration with a velocity v_i^k was situated in the cell A at the (k−

1)^th iteration with a velocity v_i^k-1;

wherein on completion of the k^th iteration, duplicate or delete one or more pairs (p_i, (v_i^k, x_i^k)), where x_i^k is the position of the particle p_i at the k^th iteration, within a cell so that the number of pairs per cell is dependent on the dynamic part determined for the cell.

6. The device of claim 5, adapted to select said pair to be duplicated/deleted, as a function of the probability P(O=occupied, V=v_i^k|ZC) is determined at the k^th iteration, and where v_i^k is a velocity component of the pair.

7. The device of claim 5, wherein the total number of particles in the grid is constant during the iterations.

8. The device of claim 5, adapted to determine P(O=empty, V=0|ZC) at the k^th iteration, denoted P^k(O=empty, V=0|ZC), as a function of the product of a first term dependent on the first probability of occupancy and of a second term, said second term being dependent on:

P^k-1(O=occ,V=0|ZC).(1−

ε

)+P^k-1(O=empty,V=0|ZC).ε

,

where ε

=P(O=occ|O<sup>−

1</sup>=empty)=P(O=empty|O<sup>−

1</sup>=occ); and

/or

coeff (v_i^k).P^k-1 (O=occupied, V=v_i^k-1,|ZA).(1−

ε

), where coeff (v_i^k) is a decreasing function of |v_i^k|, and/or

a probability of appearance p_a of a new object in the observed scene.