Research


Research is the act of going up alleys to see if they are blind — Plutarch

TSAM 2


A novel bundle adjustment algorithm for Structure from Motion and SLAM problems that not only speeds up the computation but also distributes the workload in a hierarchical way. The evaluations show that our algorithm is exact and faster than alternative approaches.
Read more...

GroupSAC


A novel variant of the RANSAC algorithm that is much more efficient, in particular when dealing with problems with low inlier ratios. Our algorithm assumes that there exists some grouping in the data, based on which we introduce a new binomial mixture model rather than the simple binomial model as used in RANSAC.
Read more...

Epitomic Location Recognition


A generative model based on epitomic image analysis captures the appearance and geometric structure of an environment while allowing for variations due to motion, occlusions and non-Lambertian effects.
Read more...

Tectonic SAM


Large-scale 3D reconstruction has recently received much attention from the computer vision community. Bundle adjustment is a key component of 3D reconstruction problems. However, traditional bundle adjustment algorithms require a considerable amount of memory and computational resources.
Read more...

Visual Odometry


Based on the matched features, a two-point algorithm for the resulting quadrifocal setting is carried out in a RANSAC framework to recover the unknown odometry. In addition, if the change in rotation can be accurately derived from other sensors, e.g. inertial sensors, a one-point algorithm can be used to obtain the remaining translational unknowns.
Read more...

Slip Detector


We use the wheel odometry data to decide the robot status by searching the unusual frequency in the spectrum. Also monocular images are used to registered across frames to get the transformantion information, which is also a candidate to decide whether the robot is moving or not.
Read more...

THMR-V Mobile Robot


THMR-V (Tsinghua Mobile Robot V) is a system that performs well with a speed up to 150km/h. Multi-region and bi-threshold edge extraction is used for image processing, and an ATN-based approach is used for lane-based localization.
Read more...