A Low-cost Simultaneous Localization And Mapping Algorithm For Last-mile Indoor Delivery

Abstract

With the development of e-commerce, more and more express packages need to be delivered. The Last-mile indoor task always takes most time of a whole delivery due to the complex and unfamiliar indoor environment. Generally, there aren't enough existing indoor localization algorithms that are able to meet the business needs. To benefit the public, in this paper, an advanced low-cost and accurate intelligent localization and mapping algorithm is proposed. Three strengths, according to the experiment results, are concluded. First, the algorithm could run on Android devices, and it is able to save the cost of infrastructure as well as battery resources. Second, the algorithm can achieve an accuracy of less than 5cm, which is enough for general commercial purposes. Last, the system could intelligently shift the sensors between the Inertial Measurement Unit (IMU) sensors and the camera. To test our algorithm, we used the robot to execute the delivery of an indoor mailbox, obtaining a result of high accuracy (>95%) and low battery cost (saving more than 56%). Our algorithm is possible to be deployed in autonomous delivery vehicles or drones to provide last-mile delivery service.