Multiresolution on-line path planning for small unmanned aerial vehicles

Abstract

In this article we propose a new online multiresolution path planning algorithm for a small unmanned air vehicle (UAV) with limited on-board computational resources. The proposed approach assumes that the UAV has detailed information of the environment only in the vicinity of its current position. Information about far away obstacles is also available, albeit with less accuracy. The proposed algorithm uses an integer arithmetic implementation of the fast lifting wavelet transform (FLWT) to get a multiresolution cell decomposition of the environment, whose dimension is commensurate to the onboard computational resources. A topological graph representation of the multiresolution cell decomposition is constructed efficiently, directly from the approximation and detail wavelet coefficients. Hardware-in-the-loop simulation (HILS) results validate the applicability of the algorithm on a small UAV autopilot. Comparisons with the standard D * -lite algorithm are also presented. I. INTRODUCTIONAutonomous operation of UAVs requires both trajectory design (planning) and trajectory tracking (control) tasks to be completely automated. Given the short response time scales of modern aerial vehicles, on-board, real-time path planning is particulary challenging for small UAVs, which may not have the on-board computational capabilities (e.g. CPU and memory) to implement some of the sophisticated path planning algorithms developed in the literature.In a typical mission of a UAV, various sensors (e.g., cameras, radars, laser scanners, satellite imagery) having different range and resolution characteristics are employed to collect information about the environment the vehicle operates in. A computationally efficient path planning algorithm, specifically adopted for on-line implementation, should therefore choose the expedient information from all these sensors, and utilize the on-board computational resources on the part of the path (spatial and temporal) that needs it most. In a nutshell, a computationally efficient algorithm suitable for on-line implementation should be characterized by a combination of short term tactics (reaction to unforeseen threats) with long-term strategy (planning towards the ultimate goal).Multiresolution analysis is widely used in practice to mitigate the computational overhead in numerically costly applications, for example, computer graphics, using progressive, view-dependent, meshes [4]. The application of multiresolution methods to path planning problems is relatively recent. In [1], [13] multiresolution, hierarchical algorithms were used to alleviate the computational burden associated

Efficient Path Re-planning for AUVs Operating in Spatiotemporal Currents

“…Previous research, such as [11] and [12], looked at solving the dynamic path planning problem by using a hierarchical approach. In such an approach, the path planning system decomposes the searching space with several levels of resolution (higher resolution around the immediate current location of the vehicle, with decreasing resolution further away) and constructs an optimal path from the current location of the vehicle to the target location.…”

Multiresolution Hierarchical Path-Planning for Small UAVs Using Wavelet Decompositions

“…The key observation of the approach is the fact that since the scaling function j,k, and the wavelet functions i j,k, (i = 1, 2, 3) are uniquely associated with the square cell c j k, = I j,k × I j, , the corresponding nonzero approximation and detail coefficients encode the necessary information regarding the cell geometry (size and location) [16,34]. To this end, consider a cell c j 0 k, at level j 0 , whose dimension is 1/2 j 0 × 1/2 j 0 with its center located at (k, ).…”

On-Line Path Generation for Unmanned Aerial Vehicles Using B-Spline Path Templates

“…In other words, by taking into account the linear independency of the B-spline basis functions and the number of equations in Eq. (22), one perturbs (p 1) control points of each curve in order to impose the precise merging condition.…”

“…These templates contain a set of planar B-spline curves, which will be regarded as local path segments to smooth a discrete path sequence. It is assumed that the obstacle-free discrete path sequence is provided by a high-level path planner, such as the multiresolution path-planning algorithm proposed in [22,23], or a similar discrete search algorithm, such as A or D . The algorithm constructs an obstacle-free channel such that the discrete path sequence is represented by a series of square cells.…”