Visual-Inertial State Estimation

We develop probabilistic methods for high accuracy state estimation in unknown complex environments using measurements from cameras and IMUs. We focus on mathematical formulation for multi-sensor fusion, estimator initialization, loop closure, sensor calibration, and robust motion estimation under aggressive motions.
We integrate these features together to build our open source software: VINS-Mono and VINS-Mobile, the extremely easy-to-use visual-inertial state estimator and AR demonstration on Linux and iOS.
read more

Dense Mapping for Autonomous Navigation

We develop real-time methods for generating dense maps for large-scale autonomous navigation of aerial robots. We investigate into monocular and multi-camera dense mapping methods with special attention on the tight integration between maps and motion planning modules.
Without any prior knowledge of the environment, our dense mapping module utilizes a inverse depth labeling method to extract a 3D cost volume through temporal aggregation on synchronized camera poses. After semi-global optimization and post-processing, a dense depth image is calculated and fed into our uncertainty-aware truncated signed distance function (TSDF) fusion approach, from which a live dense 3D map is produced.
read more

Trajectory Generation for Aerial Robots

We develop online methods to generate safe and smooth trajectories for aerial navigation through unknown, complex, and possibly dynamic environments. We use convex optimization tools to ensure both collision avoidance and dynamic feasibility.
read more