Design Of Optimal Mass Transport Based Multi-Robot Collaborative Scanning System

Reconstructing and mapping indoor environments is critical to a large variety of applications,ranging from 3D content creation for augmented and virtual real ity to localization for domestic robot navigation.On the hardware side,we have witnessed the emergence and proliferation of commodity range sensors(e.g.,Micr osoft Kinect,Intel RealSense,etc.)that capture depth data in real-time.On the so ftware side,researcher have made incredible progress developing online RGB-D r econstruction methods.The most representative one is KinectFusion,proposed by Microsoft Research in 2012,which uses multi-frame depth images to reconstruct three-dimensional scenes.We present an autonomous scanning system which allows multiple robots to perform collaborative scanning for dense 3D reconstruction of unknown indoor scenes.Our method plans scanning paths for several robots,allowing them to eff iciently coordinate with each other such that the collective scanning coverage and reconstruction quality is maximized while the overall scanning effort is minimiz ed.To this end,we define the problem as a dynamic task assignment and introdu ce a novel formulation based on Optimal Mass Transport(OMT).Given the currently scanned scene,a set of task views are extracted to cover scene regions which are either unknown or uncertain.These task views are assi gned to the robots based on the OMT optimization.We then compute for each r obot a smooth path over its assigned tasks by solving an approximate traveling s alesman problem.In order to showcase our algorithm,we implement a multi-robot auto-scanni ng system.Since our method is computationally efficient,we can easily run it in real time on commodity hardware,and combine it with online RGB-D reconstru ction approaches.In our results,we show several real-world examples of large in door environments;in addition,we build a benchmark with a series of carefully designed metrics for quantitatively evaluating multi-robot auto scanning.Overall,we are able to demonstrate high-quality scanning results with respect to reconstru ction quality and scanning efficiency,which significantly outperforms existing mu lti-robot exploration systems.