| CPC B60W 30/095 (2013.01) [B60W 60/0016 (2020.02); G01B 21/02 (2013.01); G01B 21/16 (2013.01); G01B 21/20 (2013.01); G01B 21/30 (2013.01); G01C 21/20 (2013.01); G01C 21/3605 (2013.01); B60W 2554/4049 (2020.02)] | 7 Claims |
|
1. A processor-implemented method comprising:
receiving a plurality of inputs associated with a plurality of obstacles within a narrow passage of a real time environment, via a one or more hardware processors, the plurality of inputs comprise a plurality of dimensions of a configuration space of each of the plurality of obstacles (d), a plurality of obstacle geometry (o) of each of the plurality of obstacles, wherein the plurality of dimension of the configuration space comprise of a plurality of obstacle space (Cobs) and a plurality of free spaces (Cfree), wherein the narrow passage is a constrained space between the plurality of obstacle space (Cobs) and the plurality of free spaces (Cfree);
identifying a step size (α), via the one or more hardware processors, based on the plurality of dimensions of the configuration space of each of the plurality of obstacles (d) and the plurality of obstacle geometry (o) using a probability distribution function technique;
iteratively determining a plurality narrow passage sampling points in the narrow passage for the narrow passage path sampling, via the one or more hardware processors, wherein the step of the determining the narrow passage sample among the plurality narrow passages sampling point, comprises:
generating a random configuration (q,θ) in the obstacle based on a random uniform sampling technique, wherein the random configuration comprises a position (q) defined by x, y and z coordinates and an orientation (θ) defined by Euler angles (Ψ, Θ,Φ);
identifying a sampling point (q′,θ′) using the random configuration (q,θ) based on a levy flight function and a collision detection technique, wherein the levy flight function is a category of random walk in which the step size (α) has a levy distribution, wherein the sampling point (q′, θ′) lies in the plurality of free spaces (Cfree) and outside the plurality of obstacle space (Cobs), wherein when the dimension of the configuration space increase as in a manipulator sampling or a 3-dimensional (3D) sampling, then a number of levy distribution values increase, wherein the number of levy distribution values is equal to the dimension of the configuration space; and
identifying, based on a levy flight bridge sampler technique, the narrow passage sample for narrow passage path sampling from the sampling point (q′,θ′), wherein the levy flight bridge sampler technique comprises of:
joining the random configuration (q,θ) and the sampling point (q′,θ′) using a straight line;
extending the straight line along q′ till a point q″, wherein q′ is a midpoint of the extended line between qq″, wherein extension of the straight line is expressed as: (q,θ). q′=(q+q″)/2; and
identifying the sampling point (q′,θ′) as a narrow passage the sampling point based on determining if the point q″ lies in the plurality of obstacle space (Cobs) based on the collision detection technique;
simulating narrow passage scenarios in 2-dimensional (2D) and 3-dimensional (3D) environments, wherein simulations are conducted for a 3-link, 5-link and 7-link manipulator, and results are simulated considering a bar shape obstacle scenario, a joint shape obstacle scenario and a teeth shape obstacle scenario, wherein for the joint shape obstacle scenario, a total execution time and a number of collision checking calls for levy flight bridge sampler are less, and wherein for the teeth shape obstacle scenario, simulation results obtained indicate superiority of the levy flight bridge sampler in terms of sampling quality; and
displaying in real time the identified narrow passage path sampling on an output module.
|