LoRaPCR: Long Range Point Cloud Registration through Multi-hop Relays in VANETs

Abstract

Point cloud registration (PCR) can significantlyextend the visual field and enhance the point density on distantobjects, thereby improving driving safety. However, it is verychallenging for vehicles to perform online registration betweenlong-range point clouds. In this paper, we propose an onlinelong-range PCR scheme in VANETs, called LoRaPCR, wherevehicles achieve long-range registration through multi-hop short-range highly-accurate registrations. Given the NP-hardness of theproblem, a heuristic algorithm is developed to determine best reg-istration paths while leveraging the reuse of registration resultsto reduce computation costs. Moreover, we utilize an optimizeddynamic programming algorithm to determine the transmissionroutes while minimizing the communication overhead. Resultsof extensive simulations demonstrate that LoRaPCR can achievehigh PCR accuracy with low relative translation and rotationerrors of 0.55 meters and 1.43◦, respectively, at a distance ofover 100 meters, and reduce the computation overhead by morethan 50% compared to the state-of-the-art method.