06-07-10

So here are some comments on my research since I've been here. Overall impressions of the lab: its huge, 2 floors in one building and a floor in another, plus another field lab in the outback. Aparently around 200 people though I wouldn't say I've seen that many, more like 40 or 50 I see around regularly. They have a staff of technicians as well as dedicated software developers. The centre has some big robots like amphibious atvs, a backhoe, 4 wheel segway systems, big unmanned aircraft and submersibles. There is more even larger stuff offsite. Also on the first full week there Sanjiv Singh stopped by to give a talk about doing mapping from aircraft with 3D LADAR, cool stuff.

My work is related to segmentation with Riegl data. I have developed some utilities to improve evaulation of the algorithms, as well as working on variations of the algorithms for object and ground segmentation. I haven't directly worked with Velodyne data, but I have seen some work about Velodyne sensors and calibration, which is interesting. They require constant calibration and are generally much less precise than Riegl systems, but thats the compromise for lower cost and much higher frame rate. Some people in the perception research group put a robot with a Velodyne on it on the back of a pickup truck and drove it around the city. The data from that looks amazing, all the people and vehicles moving around in 3D.