Multirate Obstacle Tracking and Path Planning for Intelligent Vehicles

Predictive and Multirate Sensor-Based Planning Under Uncertainty

“…The absence of measurements in future instants of time (i>0) is modeled with delta functions [17,19,20], a MR technique for time-varying models that modifies the Kalman gain indicating the presence ( i =I) or absence ( i =0) of measurements in every estimation instant. The choice of the filter to be used depends on the kinematic model associated to the object.…”

“…This feedback, along with the fact that a force derived from a potential field has the structure of a controller (see (17)), involves the extension of the classical philosophy for robot control to the field of planning. Indeed, the attractive force, on the structural aspect of its equations, can be considered as a P-control (proportional) on the vehicle position, in its simplest form, or as a PD-control (proportionalderivative), if the attractive force is added to a friction force, thus also considering velocities [18].…”

“…Typically, this maximum radius must be one order of magnitude lower than the dimension of the setting where the vehicle is moving. Applying this idea the estimation quality index defined in [17] can be computed as in (11), being β a user adjustable distribution scaling factor.…”

“…As a result, at each execution cycle a set of attractive F att,i and repulsive forces F rep,i are obtained in the present (i=0) and future (i>0) instants of time. Their expressions are given by (17), where u MTDi,k is a unit vector in the direction of MTD i,k and opposite to the obstacle. Every obstacle k generates a repulsive force in every instant i, whose magnitude decreases as i and MTD i,k grow.…”

“…In [17,18] we presented a multi-rate (MR) predictive PF method for local planning that takes all the previous aspects into account, unlike other methods [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16] that only consider some of them. The Potential Field Projection method (PFP) combines the PF method [3] with the multi-rate Kalman filter (MR-KF) estimation [19,20], thus considering present and future vehicle and obstacles locations within the planning strategy.…”