Integration ofVision based SLAMandNonlinear Filter forSimple Mobile RobotNavigation

Simultaneous Localization andMapping (SLAM)isa enhance itsperformance (7), because measurement modelof widely usedtechnique whichcanbuild upamapatunknownvision sensor isnotlinear. Incomparison withprevious works, environment andfigure outitsposition without prior map thispapertakes distributed particle filter approach without information. Byapplying SLAM method, this paperpresents landmark information andgeometric constraints. Distributed simple integrated navigation system forplanar mobile robotparticle filter isknowntohavepowerful performance under equipped withvision sensor andencoders. Because ofthe nonlinear, multi-modal modelandfeature points varying nonlinearity ofthemeasurement model, itisusedaparticle conditions (13). filter whichhaspowerful performance undernonlinear and multi-modal conditions. Finally, theDR (deadreckoning)Inaddition tovision sensor, range sensor isgenerally used performance isdemonstrated byanalyzing thevariance of tocompute the3-Dposition offeature point, because itis estimated position error whenthenumberofparticles and difficult toestimate theposition incaseofsingle vision sensor feature points arechanged. system. Thispapertakesdelayed initialization methodto compute thefeature point position using single vision sensor I. INTRODUCTION