EPIC Lab

EPIC Lab builds estimation and perception systems for robots that localize, map, and act under uncertainty.

State EstimationVisual-inertial navigation, aided inertial navigation, calibration, and robust multi-sensor fusion.View workHide work

Fast, reliable motion estimation.

Related papers and open-source systems from our VIO and multisensor navigation work.

Open source / TRO 2025

sqrtVINS

Square-root filtering for ultrafast, numerically stable visual-inertial estimation.

PaperCode

Open source / JFR 2025

MINS

A modular multisensor-aided inertial navigation system for camera, wheel, GNSS, and LiDAR fusion.

PaperCode

IROS 2026

miniVIO

Minimal visual-inertial odometry with compact motion constraints for efficient estimation.

Paper

Spatial PerceptionSLAM, semantic mapping, active sensing, and scene understanding for physical systems.View workHide work

Perception that carries geometry.

Selected work on dense maps and planar structure for spatially aware robot perception.

ICRA 2024

NeRF-VINS

Real-time NeRF map-based navigation connecting localization with dense scene representation.

Paper

Open source / ICRA 2023

Plane-Regularized VIO

Monocular visual-inertial odometry that uses planar regularities in indoor structure.

PaperCode

Physical AIPlanning under uncertainty, target tracking, perceptive locomotion, and deployment-aware robot intelligence.View workHide work

Robots that work beyond demos.

Current workshop activity around active perception, interaction, and long-horizon deployment.

IROS 2026 Workshop

Full-Shift Robot Workers

Active perception and interaction for robots that need to operate across human environments.

WorkshopCall

Explore papers, code, and current work: Publications, GitHub, and Google Scholar.