Target objectSensorRobot/PlatformSimulation environmentSpatial representationView search spaceShape knowledgeView determinationPath planningTerminationPerformance evaluationSmall objectFurniture sizeIndoorBuilding/Structure2D layoutTriangulation-basedlaserStructured lightStereo cameraMonocular cameraTime-of-Flight (ToF)Range sensor(unspecified)No experimentManual operation1D translator2D translatorCartesian robotRotation machineRobot armGround mobile robotGround mobile robotwith robot armMicro aerial vehicle(MAV)Not consideredIn-houseGazeboOpenRAVEAirSimOpenAI GymV-REPOpenHRPNomad simulatorBlenSorNo simulation2D occupancy map3D voxel grid3D occupancy mapwith label3D occupancy mapwith probability2D/3D vector field3D TSDF2D polygonal layout3D point cloud3D surfaceCylinder view spaceCircular view spacePlanar view space3D view spaceSpherical view spaceIcosahedron viewspaceNot consideredInformation-theoretic(entropy reduction)Seen free spacePotential fieldsFixed # of operationCompletion based onground truthNegligibleimprovement/changeCompleteness(mesh/voxel orfeasible views)Maximum # of scanNot describedVisual completenessVisual trajectory path# of scansTimeDistance traveledVolume (explored)Surface (scancoverage)Entropy reducedQualityReconstruction errorSelf measurementonlyNoneMachine learning3D object recognitionSurfacetrend/Surface patchfittingPrior 3Dpre-scannedSurfacereconstructionPrimitive fittingAngle visibility(incident angleconstraint)# of unknown voxelOcclusion boundarysurface gainTargeting surfaceedgeTargeting occlusionsurfaceMVC/Global surfacenormalFrontier-basedexplorationVisual informationMachine learningSurface predictionReinforcementlearningSampling density onsurfaceUnknown surfacegainVector field guidedRRT approachInformation gain inedges/actionsNot consideredReal-time obstacledetectionGenetic approachConsider motionuncertaintyA* approachDijkstra approachEnvironmentmodelingOutside thebounding box