Real-Time Autonomous Miniature Car Perception and Control for Package Delivery

Abstract

Autonomous vehicles increasingly play an important role in daily life. Low-stakes use cases can enable people’s familiarity and confidence in these systems, helping to establish public trust. The Neutronomous project aims to enable food and package delivery via a low-cost, easy-to-fabricate one-tenth scale autonomous car. Three Raspberry Pis execute real-time parallel computations for sensing and localization via an environmentally robust LiDAR, and state estimation and high-level planning via an IMU-enabled GPS. This build is ideal for a self-driving robot to navigate college and corporate campuses, as well as small towns. The Neutronomous car’s practical and accessible implementation paves the way for continued integration of autonomous vehicles in public spaces.