Path planning using a Multiclass Support Vector Machine
Graphical abstract
Introduction
From its inception, research on mobile robots has generated great expectations. To this day the field remains wide open and is constantly evolving. Mobile robots are used in applications such as industrial processes, oceanic and planetary exploration, military projects, rescue operations, and many more. One of the most interesting areas related to mobile robotics, one that has seen considerable research since the 1960s, is path planning. This area remains largely untested, however, and many methods are still under development, with some improving on previous methods or complementing existing techniques.
The problem we wish to address with the method presented in this paper is the generation of a trajectory which allows reaching any point on a map starting from the current position of a vehicle. The challenge is to be able to do this even if the map provided is noisy and unstructured. To this end, an innovative algorithm that combines techniques typically related to Artificial Intelligence and Machine Learning is used together with classical methods from graph theory. The main idea is to employ an SVM classifier in order to generate all the possible safe paths that a robot can follow between obstacles, creating a simple, weighted graph that allows the robot to easily compute the best paths towards a given destination.
The method described in this paper uses a map as its input. Obstacles in the map are represented and transformed into point features, which are tagged using a different label for each obstacle (or class). By doing so, every feature belonging to the same obstacle will have the same label, which will be unique to each obstacle. These generated features are used to train an SVM, yielding a hyperplane that divides all these classes, ensuring that the distance from each hyperplane to the nearest feature of each class is maximized. The intersections of this hyperplane with the plane of the map are extracted and joined together to form a graph (which is then cleaned, for optimization purposes). Edges in the graph will include the value that is desired to be minimized (in our tests, length of the sub-path). Once this graph is complete, it is quite easy and fast to travel along the map.
This document is divided as follows: in Section 2 a review of previous path planning related algorithms is presented. In Section 3 a Multiclass Support Vector Machine (MSVM) is briefly explained. The different steps of the algorithm are described in Section 4 and in Section 5 we show the results of several real application experiments. At the end of the document, in Section 6, we present some conclusions and discuss the advantages of our algorithm compared with similar methods.
Section snippets
Previous work
Path planning is a field that has been under development since the mid-1960s, but it was not until the publication of the work by Lozano-Perez in 1979 [20] that interest in this topic started to grow. The problem can be divided in two stages: global and local path planning. Global path planning tries to find the best possible path from a robot's current position to its target. The physical characteristics of the robot are not usually included in this step. Local path planning is designed for a
Support Vector Machine (SVM)
Support Vector Machines (SVMs) are a set of linear classifiers usually employed both for classification and regression based on supervised learning. This technique relies on the Vapnik Chervonenkis (VC) dimension from statistical learning theory and Structural Risk Minimization. SVMs are maximum margin classifiers that yield an optimal separation hyperplane between diverse data classes. In other words, SVMs transform a linear classification of vectors into a higher dimensional space.
Method
Given a map in which every obstacle in the robot's environment is represented, every feasible path over an obstacle-free area which maximizes the distance between objects is calculated. Other SVM-based path-planning methods [24], [33] yield a single path that connects two points in the map. In this paper, all possible paths are obtained, turning an unstructured map into a graph where edges represent different path segments and nodes their intersections. Nodes are the points at which the robot
Results
In this section, some path planning applications are described, showing the performance obtained by the method. This section is divided into simulated and real-world results. In the first set of experiments, the method is applied to a map computed through the imperfect segmentation of an aerial image. The second set of experiments present the real-time results obtained from our testing platform, Verdino.
The LIBSVM [4] library was used to carry out these experiments and other testing stages,
Conclusions
This paper describes a method that applies Multiclass SVM to robot path planning. Different tests demonstrate that the method is able to find a smooth path, if it exists, while maximizing the distance to obstacles. By generating a non-linear continuous surface whose distance to the support vectors is maximized, the paths can be calculated simply and directly. Also, the effect of the presence of outliers or noise in sensor readings can be reduced.
The method is ready to be applied to real-world
Acknowledgements
This work was funded by the STIRPE DPI2013-46897-C2-1-R project and by the Agencia Canaria de Investigación, Innovación y Sociedad de la Información (ACIISI), cofinanced by FEDER funds (EU).
References (45)
- et al.
Path planning for autonomous mobile robot navigation with ant colony optimization and fuzzy cost function evaluation
Appl. Soft Comput.
(2009) - et al.
Generating automatic road network definition files for unstructured areas using a multiclass support vector machine
Inf. Sci.
(2016) - et al.
A modified ant optimization algorithm for path planning of UCAV
Appl. Soft Comput.
(2008) - et al.
Real-time adaptive obstacle detection based on an image database
Comput. Vis. Image Underst.
(2011) - et al.
Clear and smooth path planning
Appl. Soft Comput.
(2015) - et al.
Concurrent dynamic programming for grid-based problems and its application for real-time path planning
Robot. Auton. Syst.
(2014) - et al.
Curvature continuous path generation for autonomous vehicle using B-spline curves
Comput.-Aid. Des.
(2010) - et al.
Mobile robot path planning using artificial bee colony and evolutionary programming
Appl. Soft Comput.
(2015) Support Vector Machines for Pattern Classification
(2005)A tutorial on support vector machines for pattern recognition
Data Min. Knowl. Discov.
(1998)
LIBSVM: a library for support vector machines
ACM Trans. Intell. Syst. Technol. (TIST)
Local path planning for off-road autonomous driving with avoidance of static obstacles
IEEE Trans. Intell. Transp. Syst.
Support-vector networks
Mach. Learn.
Differentially constrained mobile robot motion planning in state lattices
J. Field Robot.
On the learnability and design of output codes for multiclass problems
Mach. Learn.
Solving multiclass learning problems via error-correcting output codes
J. Artif. Intell. Res.
Which is the Best Multiclass SVM Method? An Empirical Study
RL-ART2 neural network based mobile robot path planning
Classification by pairwise coupling
Ann. Stat.
A comparison of methods for multiclass support vector machines
IEEE Trans. Neural Netw.
A simple decomposition method for support vector machines
Mach. Learn.
Probabilistic roadmaps for path planning in high-dimensional configuration spaces
IEEE Trans. Robot. Autom.
Cited by (28)
Cooperative motion planning and control for aerial-ground autonomous systems: Methods and applications
2024, Progress in Aerospace SciencesRGSO-UAV: Reverse Glowworm Swarm Optimization inspired UAV path-planning in a 3D dynamic environment
2023, Ad Hoc NetworksCitation Excerpt :Here cost function is formulated in two parts: the first one is for optimization of the path and the second one is for preventing a collision. Based on the Multiclass Support Vector Machine (MSVM) mechanism, the researchers presented a path-planning method in [3] in unstructured environments. The authors have proposed a path-planning method for evaluating high-quality waypoints throughout the path of the UAV to deal with obstacles [18].
Online rescue method based on offline learning of dynamics knowledge for launch vehicles under thrust-drop fault
2022, Applied Soft ComputingCitation Excerpt :Pourtakdoust et al. [24] developed a new approach of optimal trajectory generation for aerospace with uncertainties in system and environment utilizing neural networks. In addition, the databased trajectory planning also has been applied in trajectory design of orbit transfer [25–28], the prediction of autonomous real-time landing site selection [29], robot path planning [30]. However, there are few databased trajectory planning methods designed for the OROTO problem.
2D multi-area coverage path planning using L-SHADE in simulated ocean survey
2021, Applied Soft ComputingCitation Excerpt :Complex areas are used to verify the simulation results of the two methods. With L-SHADE optimization method and ECDM, path planning is simulated by the area delimitation according to Figure 17 in [33]. The results are shown in Fig. 11.
Mobile robot path planning using membrane evolutionary artificial potential field
2019, Applied Soft Computing JournalTrajectory planning and tracking control for towed carrier aircraft system
2019, Aerospace Science and TechnologyCitation Excerpt :At present a lot of key achievements have been made for path planning of robot. Generally, there are four popular types of algorithms that are widely used, namely the search algorithm based on nodes, the potential field method, the intelligent algorithm and the optimal control method [1–8]. However, there are few research focused on trajectory-planning for taxi of aircraft, and the methods mainly focus on search algorithm and optimal control method.