I-Corps: A Self-driving Autonomous Electric Scooter

University Of Maryland, College Park, College Park MD

Investigators

Abstract

The broader impact/commercial potential of this I-Corps project is the development of a self-driving electric scooter. Electric scooters (e-scooters) are a vital part of a new urban mobility model, improving sustainability, accessibility, and equity over cars and buses for short distances. The EPA estimates that eliminating half of US car trips less than one mile would save $575M/year in fuel costs. In 2018, 38.5 million trips were taken on e-scooters in the US, which is twice as many as in 2017. The US micromobility market is predicted to be worth between $200–300B by 2030. Worldwide, investors already have increased investments in micro-mobility start-ups by $5.7B into since 2015. This said, large scooter fleets are expensive to maintain and do not guarantee easy access for riders. E-scooters that intelligently re-position themselves without a rider may provide value to scooter operators by improving the rider experience, increasing e-scooter usage, decreasing servicing costs, and aiding the regulatory process. This I-Corps project is based on the development of artificial intelligence (AI) logistics, operations, and management solutions easily retrofitted to electric-scooters and other micromobility platforms to enable self-driving. The proposed solution enables deployment, recharging, and unit utilization optimization of an electric scooter. The technology will enable the e-scooter to safely and intelligently maneuver without a rider for the purposes of ride summoning, automatic parking, and adaptive re-positioning for increased usage. The onboard components of the AI system consist of a sensor module for perception and localization, a planning module for intelligent decision-making, and a motor module for powered actuation via sensor feedback. Offboard components include machine-learning algorithms for predicting the spatial distribution of ride demand. To date, an autonomous driving solution based on novel models of e-scooter dynamics has been developed, including a perception-based steering algorithm. In addition, an e-scooter geospatial ride-demand analysis (with data provided by the Maryland Transportation Institute) has been performed as well as designed human-subject experiments to understand e-scooter safety and ergonomics. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.