Project Details

[Return to Previous Page]

Real-Time Distributed Processing for Drone Swarms (GLOBAL project with Chalmers University)

Company: AstaZero AB

Major(s):
Primary: CMPSC
Secondary: CMPEN

Non-Disclosure Agreement: NO

Intellectual Property: NO

AstaZero is a proving ground and Tech Infrastructure in the deep forests outside Gothenburg, Sweden, where companies test self-driving technologies. AstaZero provides independent validation that these systems work safely before they reach public roads. ? Previous student projects have used drones with cameras to help monitor the test track. One project used AI to automatically detect fence breaks around the perimeter. Another project coordinated two drones to create a wider surveillance view by combining multiple camera feeds in real-time. The current system processes video from the drones on a separate personal computer located at the test track. However, we've discovered that this setup doesn't have enough processing power to handle the AI image analysis in real-time. This creates delays in processing, causes video frames to be dropped, and limits the quality of surveillance we can provide. This year, however, more powerful server computers located at the test track will enable us to offload this computation and explore additional challenges with coordinating multiple drones in a swarm-like scenario. The Goal. Therefore, this project's goal is to develop an edge computing architecture for a scalable drone-based surveillance system, which will be showcased at the AstaZero proving ground at the project's end. The showcase will demonstrate a multi-drone system that efficiently and reliably offloads image processing workloads to edge computers, enabling real-time processing of video streams from multiple drones while maintaining low-latency communication. Additional research questions regarding system design for interacting with the drones, and then letting them solve mission goals autonomously, are part of this year’s project. The Task. Pennsylvania State University students will collaborate with Chalmers University of Technology students in this international project to reach the goal. Your task will be to design, implement, and test an edge computing system for scalable drone surveillance, building upon previous students' work. + Analyze existing dual-drone video processing code and identify computational bottlenecks and scalability limitations, including image stitching techniques for more than two drones. + Design a cloud-based edge computing architecture that enables workload offloading, taking into account the safety-critical aspect of the application. + Deploy and validate the system's real-time performance, measuring latency, throughput, and reliability. + Design general interfaces and APIs that enable the use of drones and camera feeds from multiple vendors and sources.