Combiningvisualandinertialmeasurementshasbecomepopularinmobilerobotics,sincethetwosensingmodalitiesoffercomplementarycharacteristicsthatmakethemtheidealchoiceforaccurateVisual-InertialOdometryorSimultaneousLocalizationandMapping(SLAM).Whilehistoricallytheproblemhasbeenaddressedwithfiltering,advancementsinvisualestimationsuggestthatnon-linearoptimizationofferssuperioraccuracy,whilestilltractableincomplexitythankstothesparsityoftheunderlyingproblem.Takinginspirat
2025/9/2 18:16:15
5.22MB
VSLAM