SAT-Tan: Solving Tangram Puzzles with SAT

Tangram Solving

jfLOAM: Jordan's Fast Lidar Odometry and Mapping

LOAM is the top-ranking algorithm on the KITTI autonomous driving benchmark. It was developed by Ji Zhang, who sits across the hall from me at CMU. He has now spun out a company to commercialize his work, but the rest of us are left with the 10 year old ROS implementation he has shared on github. With jfLOAM, I've rebuilt LOAM from the ground up with a focus on end-to-end latency.

Lidar Odometry and Mapping

LEGO Voxel Sculpting

I've always loved the interplay of computing and art (and LEGO!). Recently, I wondered whether I could generate a valid LEGO sculpture from a triangle mesh. Spoiler: I can!

Lego Block Bunny

Why FAST is Fast

Features from Accelerated Segment Test (FAST) is the first corner detection algorithm fast enough for realtime video processing. The original paper provides a good explanation of the algorithm but neglects key implementation details that are critical to the speed of the algorithm. This article is my attempt to independently explain the algorithm with emphasis on implementation. For reference, I provide my single-file C99/C++ implementation in the style of the stb libraries.

Fast KeyPoint Detection