Situation-Aware Intelligent Cognitive Assistant With Deep Visual Learning and Ambient Data Analytics for Urban Rideshare Safety

Electric rideshare scooters have revolutionized urban transit in the recent years. However, adverse riding conditions as well as rider negligence have created compromises in the safety of these urban transit facilities. This creates an opportunity to introduce an intelligent decision-making methodology that generates suggestive actions for riders based on ambient riding conditions. Efficient decision-making procedures utilizing improved technology could be incorporated to enhance the safety of rideshare scooters. The improvements would enable rider aware deployment of safety measures when riding. Developments in the fields of sensor and deep learning technology can be combined to optimize object detection. Additionally, machine vision techniques and vibrational data can be utilized to identify obstacles in the surroundings. Modern decision-making tools like Case Based Reasoning and Fuzzy logic can perform efficient decision-making to identify suggestive actions for rideshare scooter riders. This thesis performs requirement analysis for creating the intelligent cognitive assistant, creates the system architecture to achieve efficient data processing and reasoning, and showcases an Engineering design strategy for improving the performance of the system to improve rideshare safety. The project incorporates ambient data collected using cell phone sensors and performs analysis using deep visual learning, machine vision, and vibration analysis to create situational awareness. The processed data is fed into reasoning models like CBR to generate safety scores which can used to form suggestive outputs for the user. Results of real-world implementation of the decision model and parameter tuning for the model are presented as the proof of concept.M.S