GTSAM creator Frank Dellaert: prefer fixed-lag smoothers over EKF when CPU allows
fdellaert · x · 2026-11-19
GTSAM creator and Georgia Tech professor Frank Dellaert shared practical advice on robot state estimation: while GTSAM offers EKF, IEKF, and EqKF for heavily embedded devices, he'd always prefer fixed-lag smoothers whenever some moderate CPU headroom is available. He also notes factor graphs trivially fuse other sensors (radar, lidar, vision), even with delayed data.
More from Embodied
- GTSAM author Frank Dellaert: prefer fixed-lag smoothers over EKF when CPU allows — fdellaert · 2026-11-19
- Kevin Weil-co-founded Warp Sciences hires roboticists to run real scientific experiments — kevinweil · 2026-11-19
- Voxel51 adds natural-language search for robot grasp failures across MCAP and LeRobot data — heyshrutimishra · 2026-11-19
- Tesla Cybercab output rising as Giga Texas spotter flags revised Starlink hatch — whurley · 2026-11-19
- Flow Engineering Brings Agentic AI to Hardware Design; Rivian Engineers Grew 40 to 1,500 in 4 Months — PTrubey · 2026-11-19
- How Robots Juggle Cameras, Lidar, Motor Control and AI: A ROS2 Starter Series — chrismatthieu · 2026-11-19