Automated Taxi Booking Operations for Autonomous Vehicles

TL;DR

The paper addresses automating taxi booking operations for autonomous vehicles by designing an end-to-end system that links AVs, a customer mobile app, and an AWS-based server. The approach leverages ROS middleware, a ROS bridge, WebSocket communications, and a Firebase real-time database to track GPS locations, assign the best AV for each request, and continuously stream position updates between the AV and customer apps. It demonstrates a practical prototype using a KIA Soul EV equipped with Lidar, IMU, GNSS, and an Android client, with real-time visualization on Google Maps. The results show linear growth in registration latency with load and higher booking latency due to the matching computation, illustrating the system’s real-time capabilities and an avenue for ML-driven optimization. The work highlights practical deployment considerations and security concerns for autonomous-taxi operations and outlines future research directions in ML-based optimization and distributed-system robustness.

Abstract

In a conventional taxi booking system, all taxi operations are mostly done by a decision made by drivers which is hard to implement in unmanned vehicles. To address this challenge, we introduce a taxi booking system which assists autonomous vehicles to pick up customers. The system can allocate an autonomous vehicle (AV) as well as plan service trips for a customer request. We use our own AV to serve a customer who uses a mobile application to make his taxi request. Apart from customer and AV, we build a server to monitor customers and AVs. It also supports inter-communication between a customer and an AV once AV decided to pick up a customer.

Figures (8)

• Figure 1: Booking system framework from autonomous vehicle perspective.
• Figure 2: Communication schema between ROS and non-ROS system on autonomous vehicles.
• Figure 3: ROS environment and self-driving architecture of our autonomous car. We install Lidar, Odometry, inertial measurement unit (IMU) and GNSS in our KIA Soul EV car. ROS localization nodes are used to localize our car and search for customer location. Subsequently, ROS Route Planning node plans a travel route before sending control commands to our car using ROS Control Node.
• Figure 4: Booking system framework from network communication perspective. A pattern of our server WebSocket address is ws://IP_address:port/chat. The figure indicates that three have three drivers and three customers. The blue arrows are messages sent from customers and AVs in every second. It allows our server to track the current location of customers and AVs. The red arrows are messages dispatched from our server.
• Figure 5: This figure decribes the workflow of our server application. Our application collects data from customers and AVs using JSON strings {(1.1), (1.2)}. It parses data from received strings and updates to concurrent dictionaries considered as tables on real-time database {(2.1), (2.2)}. If a customer sends a taxi request {(2)}, a new thread is created to get real-time data from database and sends back to our car {(5.1)} and the customer Android application {(5.2)}.