1T-5
Time-of-flight camera based Probabilistic Polygonal Mesh mapping
○Louis-Kenzo Cahier，高橋 徹，尾形哲也，奥乃 博（京大）
Time-of-flight cameras, a type of optical sensor that produces 2D range images, have recently become available; they offer high frame rate 3D ranging at no processing cost but with high noise. We explore the use of this source of information to build and maintain a new type of 3D environment map called Probabilistic Polygonal Mesh. Every new scan is used to update a probabilistic displacement texture wrapped on the mesh; this texture is then used to simplify, refine, or adjust the base mesh. The result is a sensor noise resistant, dynamic 3D model of the environment destined to be used by mobile robots.