Device for controlling the trajectory of a vehicle
A device for controlling, in real time, the trajectory of an autonomous vehicle includes a control module which produces, in real time, from a state vector at each point in time, a first steering command in order to stabilize the trajectory of the vehicle relative to a vehicle path. The device includes an anticipation module which generates a variable representative of a meta-vector of deviations for each predicted position of the vehicle at points in time resulting in a given quantity at the current point in time and of the state vector at the current point in time. The control module produces the first steering command by quadratic optimization of a relationship between the generated representative variable and a meta-vector of successive steering commands for each predicted position of the vehicle at points in time resulting in a given quantity at the current point in time.
1. A control device for real-time control of trajectory of an autonomous vehicle, the control device comprising:
processing circuitry configured to
produce in real time on the basis of a real or estimated vector of states (ξ k , {circumflex over (ξ)} k ) at each instant k of following of a lane by the vehicle moving at a current speed (v), a first turning command (u k ) to stabilize the trajectory of the vehicle with respect to said lane, and
generate a quantity (Ξ k+1 ,h) representative of a meta-vector (Γ k ) of disturbances for each predicted position of the vehicle at instants succeeding in a given amount (N) at a current instant (k) and of the vector of states (ξ k , {circumflex over (ξ)} k ) at the current instant (k), wherein
the processing circuitry is configured to produce the first turning command (u k ) by quadratic optimization of a relation between said representative quantity (Ξ k+1 ,h) generated and a meta-vector (U k ) of successive turning commands for each predicted position of the vehicle at said instants succeeding in the given amount (N) at the current instant (k), and
said representative quantity (Ξ k+1 ,h) comprises, for each predicted position of the vehicle at said instants succeeding in the given amount (N) at the current instant (k), at least one product of the vector of states (ξ k , {circumflex over (ξ)} k ) at the current instant (k) by a numerical matrix A(v) modeling a dynamic relation of the vehicle and raised to a power equal to a succession rank (i) of said predicted position.
2. The control device as claimed in claim 1 , wherein said disturbances are curvatures of the lane for each predicted position of the vehicle.
3. The control device as claimed in claim 1 , wherein said representative quantity (Ξ k+1 ,h) comprises, for each predicted position of the vehicle at said instants succeeding in the given amount (N) at the current instant (k), at least one product of factors comprising a disturbance (γ k+i ) on the lane at the predicted position and the numerical matrix A(v) modeling the dynamic relation of the vehicle and raised to the power equal to the succession rank (i) of said predicted position.
4. The control device as claimed in claim 2 , further comprising an apparatus combining properties of an optical camera and of a radar so as to provide at least one geometry of lateral deviation (y(x)) of director line of the lane in the form of a polynomial.
5. The control device as claimed in claim 4 , wherein the processing circuitry is configured to calculate curvature (γ k+i ) on the basis of the geometry (y(x)) of median director line of the lane by formula:
γ
k
+
i
=
y
″
(
x
)
(
1
+
[
y
′
(
x
)
]
2
)
3
/
2
.
6. The control device as claimed in claim 1 , wherein a relation between said representative quantity (Ξ k+1 ,h) generated and the meta-vector (U k ) of successive turning commands comprises, for each quadratic product of an anticipated state ξ k+i , an elementary states weighting (Q k+i ) and an elementary command weighting matrix (R k+i ).
7. The control device as claimed in claim 6 , wherein the elementary states weighting matrix (Q k+N ) for a last anticipated state (ξ k+N ) comprises coefficients which are associated with a decreasing positive function of a Lyapunov type, so as to favor a convergence of the optimization.
8. The control device as claimed in claim 1 , wherein the processing circuitry is configured to compel all or some of elementary commands (u k+i ) of the meta-vector (U k ) of successive turning commands to comply with constraints of a type comprising at least one upper bound on the states, on the commands and/or on measurements.
9. The control device as claimed in claim 8 , wherein the constraints comprise at least one lower bound on the states, on the commands and/or on the measurements.
10. The control device as claimed in claim 9 , wherein the processing circuitry is configured to impose constraints on the elementary commands (u k+i ) of the meta-vector (U k ) of successive turning commands, so that a last state vector predicted (ξ k+N ) is in polytope (Ω) which ensures asymptotic convergence beyond a prediction horizon (N).
11. The control device as claimed in claim 8 , wherein the processing circuitry is configured to impose constraints on the elementary commands (u k+i ) of the meta-vector (U k ) of successive turning commands, so that a last state vector predicted (ξ k+N ) is in a polytope (Ω) which ensures asymptotic convergence beyond a prediction horizon (N).
12. A method for real-time control of trajectory of an autonomous vehicle, the method comprising:
producing, by processing circuitry, in real time on the basis of a real or estimated vector of states (ξ k , {dot over (ξ)} k ) at each instant k of following of a lane by the vehicle moving at a current speed (v), a first turning command (u k ) to stabilize the trajectory of the vehicle with respect to said lane; and
generating, by the processing circuitry, a quantity (Ξ k+1 ,h) representative of a meta-vector (Γ k ) of disturbances for each predicted position of the vehicle at instants succeeding in a given amount (N) at a current instant (k) and of the vector of states (ξ k , {circumflex over (ξ)} k ) at the current instant (k), wherein
the producing produces the first turning command (u k ) by quadratic optimization of a relation between said representative quantity (Ξ k+1 ,h) generated and a meta-vector (U k ) of successive turning commands for each predicted position of the vehicle at said instants succeeding in the given amount (N) at the current instant (k), and
said representative quantity (Ξ k+1 ,h) comprises, for each predicted position of the vehicle at said instants succeeding in the given amount (N) at the current instant (k), at least one product of the vector of states (ξ k , {circumflex over (ξ)} k ) at the current instant (k) by a numerical matrix A(v) modeling a dynamic relation of the vehicle and raised to a power equal to a succession rank (i) of said predicted position.
13. A method or real-time control of trajectory of an autonomous vehicle, the method comprising:
producing, by processing circuitry, in real time on the basis of a real or estimated vector of states (ξ k , {circumflex over (ξ)} k ) at each instant k of following of a lane by the vehicle moving at a current speed (v), a first turning command (u k ) to stabilize the trajectory of the vehicle with respect to said lane; and
generating, lav the processing circuitry, a quantity (Ξ k+1 ,h) representative of a meta-vector (Γ k ) of disturbances for each predicted position of the vehicle at instants succeeding in a given amount (N) at a current instant (k) and of the vector of states (ξ k , {circumflex over (ξ)} k ) at the current instant (k), wherein
the producing produces the first fuming command (u k ) by quadratic optimization of a relation between said representative quantity (Ξ k+1 ,h) generated and a meta-vector (U k ) of successive turning commands for each predicted position of the vehicle at said instants succeeding in the given amount (N) at the current instant (k), and
said representative quantity (Ξ k+1 ,h) comprises, for each predicted position of the vehicle at said instants succeeding in the given amount (N) at the current instant (k), at least one product of factors comprising a disturbance (γ k+i ) on the lane at the predicted position and a numerical matrix A(v) modeling a dynamic relation of the vehicle and raised to a power equal to a succession rank (i) of said predicted position.