Non-linear MPC Motion Planner for Autonomous Vehicles Based on Accelerated Particle Swarm Optimization Algorithm

Advances in centerline estimation for autonomous lateral control

“…• relative orientation with respect to the ideal centerline tangent; • roadshape (of the centerline) in front of the vehicle. In [35] the roadshape is described through third order polynomials in a curvilinear abscissa framework (s − y) that is centered according to the current vehicle position. The most important advantage with respect to Cartesian coordinates (X − Y ) is that each road characteristic can be described as a function of one parameter (i.e., the abscissa s), thus each function that approximates the lane center is at least surjective.…”

“…The most important advantage with respect to Cartesian coordinates (X − Y ) is that each road characteristic can be described as a function of one parameter (i.e., the abscissa s), thus each function that approximates the lane center is at least surjective. This property is very important because it is retained along with the whole optimization horizon in model predictive control approaches [35]. Fig.…”

Vehicle state estimation based on Kalman filters

“…In particular, GPS receivers publish the respective measurements at 10 Hz, while kinematic quantities are given at 20 Hz and IMU provides accelerations and rotational rates at 100 Hz. Since the closed loop frequency of a trajectory planner can be considered equal to 20 Hz (as indicated in [20]), state estimation should be provided at least at the same frequency. For this reason, it has been developed a switching filter triggered by the receiving of the serial data timed by the hard real time actuation system.…”

An integrated algorithm for ego-vehicle and obstacles state estimation for autonomous driving

“…The most common road definition models are: poly-line model, lane-let model, and Hermite spline model with increasing complexity and computational need in given order [13]. According to the different motion planners presented in [14,15], the road map model of the track can be approximated through cubic Hermite spline interpolation [16]. The most important advantage of curvilinear coordinates (s − n) with respect to Cartesian coordinates (X − Y ) is that each road characteristic can be described as a function of only one parameter (i.e., the abscissa s); thus, each function that approximates the centerline is at least surjective.…”

“…To conclude, the measurement update of the state estimates can be per-364 formed accounting for the cross covariance matrix given by (14), that is required 365 to compute the Kalman gain matrix as indicated in (15). The updated state 366 vector (x + k ) and covariance (P + k ) are obtained from equations ( 16) and (17).…”