The current mode of trolley handling operations is manpower-intensive, as it employs basic methods in the trolley collection process, such as the visual search for stray trolleys, and the trolley distribution process, as in the manual pushing of trolley trains. Further, the lack of information on the current status of the trolley system limits the potential of improvements that can be made to the system. This is unsustainable, given increasing passenger demand, airport development, and the growing manpower shortage.

The robot serves to automate the process of transporting a chain of 5 trolleys from one place to another. While the destination goal is determined by the trolley handler via their UI, the robot will self-determine the fastest path to be taken, and navigates itself to the destination. The LIDAR and ultrasonic sensors provides all the necessary information for the robot operating system (ROS) to execute the self-navigation and obstacle avoidance operations.

ROS

The robot runs on ROS (Robot Operating System) to integrate the hardware components for mapping, obstacle avoidance and navigation. It also enables the robot to communicate efficiently with the trolley handler through the backend database. The robot uses open-sourced packages such as AMCL, move_base, and the map server to carry out the necessary functions. The Robot to Database Node receives goals from the database and converts them to the map’s global frame coordinates. It also subscribes to the robot’s pose and updates the robot’s coordinates into another database table which will be displayed on the user interface.

Front-end User Interface

Mobile based application that is designed to be used on an iPad.

It’s goals are to display locations of the stray trolleys to the users and to the users the ability to move the robot.

The process of moving a robot can be found in the video (demo video).

Asset Tagging

Asset tagging was done through ultra-wide band (UWB) and infrared sensors (IR) to collect the needed information for analysis and visualisation: location data is provided by the ultra-wide band sensor while the presence of luggage is detected by the infrared sensor. To collect sufficient trolley usage data for analysis, a model was built to simulate the data generated by asset tagging through the UWB and IR sensors.

This information is then sent to the backend where processing of the data, such as the determination of the trolley statuses, is done and stored on the database. This allows us to provide relevant information to the handlers such as the location of stray trolleys, as well as an overview of the trolley system through the dashboard.

Dashboard

The dashboard was designed to achieve three main objectives: providing an overview of the system, getting a better understanding of the trolley handlers’ workflows, and gaining insight into visitors behaviours. 

It analyses and visualises the data collected through asset tagging to identify patterns in behaviour, which can then be used to support decision making.

The analysis is done by considering the amount of time spent by and the number of trolleys in various defined trolley states, as well as the locations and movement of trolleys.

The dashboard has three tabs, each targeting a different objective. For each panel, the user is able to choose the amount of data to be used e.g. yesterday’s data, by selecting from the dropdown. There are also help buttons in each panel the user can hover over to get more information.