Galea, M & Vidal Calleja, T 2019, 'Point Cloud Edge Detection and Template Matching with 1D Gradient Descent for Wall Pose Estimation', https://ssl.linklings.net/conferences/acra/acra2019_proceedings/views/b…, Australasian Conference on Robotics and Automation, ARAA, Adelaide, Australia, pp. 1-10.
Mobile manipulation in unstructured construction environments involves a range of complex robotic problems. We address a perception requirement for autonomous brick placement; estimating the pose of a partially built wall to
facilitate the placement of the subsequent brick. Our method uses RGB-D data to extract the surface edge points of the wall and classify them as horizontally or vertically aligned. The contribution of this paper encompasses a wall template that encapsulates its surface edge features and a novel 1D gradient descent template matching algorithm for pose estimation. We apply our method in mobile manipulator brick placement, demonstrating its robotic applications. Evaluation methods prove the efficacy of the proposed framework, both quantitatively and qualitatively and using both simulated and real data.