ABB UAV Inspection Project

Project Introduction

This is a collaborative project originating from the ABB Corporate Research Center’s ‘Drone-based automated indoor equipment inspection system’. Working with my lab teammates and NoopLoop , we were responsible for developing an ultra-wideband (UWB)-enabled autonomous drone navigation system.

The core objectives included implementing UWB-based indoor positioning to achieve drone functionalities such as point-to-point navigation, hovering, and predefined waypoint navigation, while prioritizing commercially available, easy-to-assemble, and highly integrated components to minimize debugging efforts and enhance deployment efficiency.

Leveraging our laboratory's existing equipment and technological expertise accumulated from previous uav competitions, we successfully designed, constructed, and developed the system.

System Overview

System Architecture

The hardware platform primarily consists of the following components:

• Positioning System: NoopLoop UWB modules to provide real-time drone localization within the operational area.
• Flight Platform: DJI Mavic 2 drone.
• Communication Devices: Wi-Fi modules, DJI 2.4/5.8 GHz remote controllers, and LinkTrack modules.
• Core Processors: A mobile computing unit (Android smartphone) and a base station (laptop).

The system's hardware and data architecture are as follows.

Software Architecture

The data flow and operational workflow are as follows:

• Position Acquisition:The UWB base stations provided by NoopLoop capture the global position of the UWB tag (LTS Tag) mounted on the drone.
• Data Parsing & Forwarding (Base Station): The base station (laptop) parses and relays data by running ROS nodes to read serial port data packets.Using the LinkTrack protocol, it extracts the real-time position of the tag in the operational area and publishes this information to a designated ROS topic.
• Mobile Processing (Android Device): The Android mobile device registers as a client via the Wi-Fi local network and subscribes to the relevant ROS topics.Through ROSbridge, it retrieves the drone’s real-time position data from the published topic.
• Control Execution: The Android device calculates positional deviation between the drone’s current location and the target waypoint.Based on this deviation, it computes the desired velocity and transmits the target velocity commands via a USB-connected DJI remote controller to the Mavic 2 drone, enabling autonomous motion control.

The Android-based software architecture is as follows.

Demonstration

The videos below showcase the system in operation, include the mobile computing unit (Android device), ground computing terminal (base station laptop), and a third-person perspective of the drone's autonomous flight.

Achievements

With the support from NoopLoop, the project was successfully completed and delivered on-site, ultimately earning recognition for its smooth handover and being well received by the project acceptance committee.The photo below is a group picture with my teammates taken during the project handover to ABB.

Acknowledgement

We would like to express our sincere gratitude to ABB and NoopLoop for their unwavering support throughout the collaboration. Special thanks to the HITCSC Team for cultivation and mentorship. Finally, thanks to my teammate LIN Zhaochen for his contributions to the project.