Sequential clustering by triangle-cascaded robot deployment

A virtual robot deploys its joints and linkages step by step in a 2-D region with presented obstacles. Each step of deployment constructs a piece of virtual robot trajectory based on only a few obstacles in front. The virtual robot trajectory serves as an envelope for obstacle clusters. Sequential clustering is thus called to solve this issue. The innovation of triangle cascading, composed of joint discrimination and apex least-square deployment, reflects the idea of sequential clustering. Simulation covers triangle cascading, gap comparison and common rim or common apex for further deployment, link and reduction of joint trajectory. An alternative test pattern using random-distributed obstacles validates algorithms developed in this paper. A hybrid clustering combining fuzzy c-means and hierarchical clustering shows a qualified approach for the validation eventually.