SimultaneousLocalization and Mapping of GPS Denied Indoor Environment with Laser Scanner forUnmanned Aerial Vehicles1. IntroductionAutonomous unmannedaerial vehicles have gained popularity in the field of civil applications overthe years, especially in the tasks such as disaster response and rescue whichmay be dangerous or difficult for human to perform.
With compared to groundrobots, UAVs are not limited by the rough terrain and hence they are more agilethat can move faster. Solving the computational problem of modelling a map foran unknown environment while simultaneously keeping track of the aerial robotitself in the map is the key ability required for robots to operateautonomously in real life environment. In outdoor environment,robots can take advantage of the rapid development of Global NavigationSatellite System (GNSS) and Inertial Navigation System (INS) to locate itself.However, localization based on GNSS is not very practical in indoor environmentwhere GNSS signal is significantly weakened and even lost. This problem of indoorSLAM for ground robot is normally solved with different types of range findersuch as laser scanner, cameras, etc.
Unfortunately, unmanned aerial vehiclesare also constrained by limited payload capacity and frame size. Therefore,light and small sensors with low computational costs are necessary for UAVs tofly autonomously in indoor environment. Moreover, the extra challenge faced isthat flying robots have one additional degree of freedom which makes theenvironment more complicated. The robot is also more difficult to control withcomplex body dynamics.A navigation system for aquadcopter and modification of an open-sourced software, Hector SLAM, arediscussed in this report.
The approach requires relatively low computationalresources and hence light onboard processors can be used. Robot OperatingSystem (ROS) with APIs of the ROS navigation stack are used. However, theapproach is only suitable for SLAM in small indoor areas where loop closure isnot required. It is not suitable in scenarios such as SAFMC Category D2 whereflying robots need to fly in a spacious indoor environment with limitedfeatures to find victims in a simulated category five hurricane scenario.Therefore, the open-sourced software need to be modified so that large loopsare closed. Moreover, while the 2D laser scanner can only create a scanningplane, the altitude information can be obtained with a mirror mounted on thelaser scanner to direct some laser beams 90 degree to the ground.
Therefore, a3D map can be generated and together with the inertial measurement units inPixhawk, becomes part of the flying vehicle’s control system.2. Literature ReviewThe integration of laserscanner with other sensors have been studied intensively by research community.Scan matching is the most common approach being implemented to estimate posechange from laser scanner measurements.
There is some open-sourced softwarelike Gmapping, which uses Rao-Blackwellized particle ?lters 1. However, thesolution works good only in planar environment and rely heavily on sufficientlyaccurate odometry. In addition, the method does not leverage the high updaterate of modern laser scanners. Therefore, for an entirely unknown environment, theflying platform may need to perform some extreme rolling and pitchingmanoeuvres. Iterative Closest Point(ICP) is an iterative algorithm that minimizes difference between twopointclouds from consecutive laser scans without extracting the line or plane features.Therefore, it is useful to use ICP for unstructured environments as it isindependent of environment features. Paul 2 proposed that for everypoint of a current scan, the smallest Euclidean distance of the point in thereference scan is selected.
However, the method is relatively slower because theentire poincloud is examined to ?nd the most appropriate transformation. Furthermore,the iterative process in the algorithm could lead to unstable performance asthe matching loop continues until convergence. There are cases where ICPs donot converge to a correct solution if the initial guess of the transformationis not good.
The probabilistic scanmatching method is proposed by Montesano 3. This method is likeICP but has two stages: correspondence location and displacement estimation. Firstly,the correspondence between each two consecutive scans is calculated. Then the transformationis estimated. This probabilistic model considers all uncertainties involvedwith drift of the sensor and the noise in measurement. Moreover, it allowscorrespondence to be found through probability integration over all possibleassociations between the range measurements of the two scans. The ICP and PSMmethods are most popular in solving point-based scan matching problem due toits effectiveness and simplicity.The feature-based scanmatching approach is proposed by Jiayuan 4, which derives thepose change of the robot by transforming the raw laser scan points into a moreefficient representation via tracking the feature parameters change.
Somecommon features used includes line segments, edges, and corners, etc.Split-merge algorithm can be used to extract line features and extended 1D SIFTis used to detect point features. Orientation information is extracted from 2Dline correspondence and initial guess for transformation estimation can also beimproved.
While this method is more accurate and robust than ICP approach, itrequires even higher computational resources to process laser scans in real-time.It is hard to achieve accuracy with the limited resources of the UAV onboardcomputer using this method.Normal Distribution Transform (NDT) proposed by Peter 5 is a mathematicalproperty-based scan matching method. This transformation can be used to deriveanalytic expressions for matching another scan.
It aligns scans to a mixture ofnormal distributions representing preceding scans and no explicitcorrespondences between points or features must established. Therefore, themethod is more robust without correspondences. All derivatives can becalculated analytically. Therefore, it is fast enough to run in real time.However, the method still requires certain features and featureless corridorsmay pose a problem on the algorithm.
A flying vehiclelocalization approach for indoor environments based on a laser scanner is proposedby Bachrach 6. It uses a single laserscanner with globally consistent range image alignment method for localizationand map construction. All new scans are registered against the union ofpreviously acquired and registered scans. This method does not spread out theerrors and is order independent. However, the approach has limited ?eld of viewwhen applied to environment mapping rather than to navigation. 3D map isgenerated when UAV moves in a vertical direction with a simple 2D lidar.