An Automatic 3D Reconstruction System For Indoor Environment Using A 2D Laser Radar

In recent years people have seen an increasing interests on theresearch of mobile robots, an intelligent machine which is capable ofperforming some disgusting, risky or monotonous tasks historicallyassigned to human beings. A simple classification can be considered withthe different working environments: indoor mobile robots and outdoormobile robots. The focus of this dissertation is an automatic 3Dreconstruction system for indoor environment using a 2D laser radar.First, this paper introduces the hardware of the system. The systemmainly consists of three parts: a). Two-dimensional laser scanning radar;b). Scanning device and driving; c). control and data acquisition module.The scanning mechanical device, including a bracket and the rotation axis,driven by a stepper motor.MCS-51 is a control unit at the core of thecontrol system. SCM system is used to connect the interface betweenRS-232 and the computer. It's major tasks are: a). to receive the control commands from PC, such as the initial position of scanning, startscanning and stop scanning, set of scanning range and speed etc., toexplain these orders into the stepper motor's control signals; b). to sendthe current crew pitch angle upward.. SCM system only needs to knowthe angle stepper motor rotate and the angle of the initial position, we canknow that pitch angle. Pitch angle is sent every 4 ms. And the laserradar's scan cycle is 26ms. The data acquisition module is responsible forgetting the data from the high-speed data acquisition interface card,receiving the pitch angle sent by MCS-51. Then we get the radar data andthe position of the scanning plane of every frame. The space coordinatesare realized by the coordinate transformation.Second, it introduces the 3D reconstruction of the system. Wheneverthe system gets a 3D laser scanning radar data, conducting coordinatetransformation, then the data is handled separately by two different ways.First, matching with the point clouds of the last frame by ICP and gettingthe change of the position between them; Another way is to extract datafor linear and planar in order to get the current plane model. Finally, thechanges sought by the location of the plane model and the current framewill have to expand the integration of the whole scene plane model. Theproceeding is repeated until getting the 3D plane model of the wholescene.Finally, the experimental results show the system is low cost, high accuracy, stability and reliability can be achieved automaticthree-dimensional reconstruction of the indoor scene.