扫地机器人L型墙测试数据

扫地机器人在工作时时常会遇到一些特殊场景，比如L型墙场景，如何能在此场景中准确定位与避障是个难点也是重点，在此场景，扫地机深度视觉模组获取的深度数据的精度就显得十分重要。搭建扫地机真值系统，用真值相机和目标深度模组进行不同材质、光线、距离下L型墙数据采集，获取L形墙场景的数据，并对数据进行处理得到校正后的数据用于目标深度模组的校正，提升深度模组在扫地机L型墙场景下的测量精度，以解决扫地机定位与避障问题。采扫地机器人真值系统采集不同材质、光线、距离下的L型墙数据，包括原始深度相机数据、真值相机采集的点云数据；对原始深度相机数据进行系统误差去除；对真值相机采集的L型墙点云数据进行修复，通过平面拟合算法进行L型墙的墙面校正、再通过滤波算法对点云进行平滑去噪，得到修复后的真值相机点云；将修复后的真值相机点云与深度相机进行点云配准，得到校正后的L型墙的点云数据——校正后数据laser-x、laser-y、laser-z和校正后数据tof-x、tof-y、tof-z。Robotic vacuums frequently encounter special scenarios during operation, such as the L-shaped wall scenario. Achieving accurate positioning and obstacle avoidance in such scenarios is both a critical challenge and a research priority, where the precision of depth data collected by the robot's depth vision module plays a vital role. To resolve the positioning and obstacle avoidance problems of robotic vacuums, a robotic vacuum ground truth system was constructed. This system collects L-shaped wall data under different materials, lighting conditions and distances using a ground truth camera and the target depth module, acquires raw data from L-shaped wall scenarios, processes the collected data to obtain calibrated data for calibrating the target depth module, thereby improving the measurement accuracy of the depth vision module in the L-shaped wall scenarios of robotic vacuums. Specifically, the ground truth system collects L-shaped wall data under varying materials, lighting and distances, including raw depth camera data and point cloud data captured by the ground truth camera; systematic errors are eliminated from the raw depth camera data, then the L-shaped wall point cloud data collected by the ground truth camera is repaired by first calibrating the wall surfaces of the L-shaped wall via plane fitting algorithms and then smoothing and denoising the point cloud using filtering algorithms to obtain the repaired ground truth camera point cloud; finally, point cloud registration is conducted between the repaired ground truth camera point cloud and the depth camera data to obtain corrected L-shaped wall point cloud data, namely the corrected datasets laser-x, laser-y, laser-z and tof-x, tof-y, tof-z.