SafeDeep: Dependable Deep Learning for Safety-Critical Airborne Embedded Systems

Status:

finished

Start date:

2019-09-01

End date:

2023-12-31

Deep neural networks (DNNs) have shown to be very successful in several areas, e.g. for object detection in autonomous cars. DNNs may also be successful in airborne systems. One such possible application is guided landing. The enabling of safe landing in adverse weather conditions without full ground support from the instrument landing system, decreases aerospace greenhouse gas emissions as multiple landing attempts and aerospace congestion are mitigated. To land autonomously without support from ground infrastructure requires advanced airborne systems including algorithms for detecting the runway. These systems are safety-critical. 

This project addresses design methods for the use of DNNs in airborne safety-critical systems. DNNs cannot rely on traditional design assurance techniques described in documents from certification authorities or standardization bodies. In this project, the research focus is on mitigation techniques for design errors in both hardware and software and for adversarial effects which can lead to system failures. The expected results are design methodologies and fault tolerant architectures for airborne safety-critical applications using neural networks.

[Show all publications]

A Systematic Literature Review on Hardware Reliability Assessment Methods for Deep Neural Networks (Jan 2024)
Mohammad Ahmadilivani , Mahdi Taheri , Jaan Raik , Masoud Daneshtalab, Maksim Jenihhin
ACM Computing Surveys (CSUR)

Comparing Ext4 and ZFS for Onboard Data Processing (Oct 2023)
Liza Johansson, Hassan Said, Håkan Forsberg, Nandinbaatar Tsog, Oskar Flordal
The first European Data Handling & Data Processing Conference (EDHPC 2023)

Efficient On-device Transfer Learning using Activation Memory Reduction (Sep 2023)
Amin Yoosefi , Seyedhamidreza Mousavi, Masoud Daneshtalab, Mehdi Kargahi
International Conference on Fog and Mobile Edge Computing (FMEC)

Evaluating the robustness of ML models to out-of-distribution data through similarity analysis (Sep 2023)
Joakim Lindén, Håkan Forsberg, Ingemar Söderquist , Masoud Daneshtalab
1st Workshop on Advanced AI Techniques for Data Management and Analytics (AIDMA)

A novel method for detecting UAVs using parallel neural networks with re-inference (Sep 2022)
Hubert Stepien , Martin Bilger , Håkan Forsberg, Billy Lindgren , Johan Hjorth
33rd Congress of the International Council of the Aeronautical Sciences (ICAS 2022)

Curating Datasets for Visual Runway Detection (Oct 2021)
Joakim Lindén, Håkan Forsberg, Josef Haddad , Emil Tagebrand , Erasmus Cedernaes , Emil Gustafsson Ek , Masoud Daneshtalab
The 40th Digital Avionics Systems Conference (DASC'2021)

PartnerType
Saab AB, Avionics Systems Industrial

Masoud Daneshtalab, Professor

Room:
Phone: +4621103111


Håkan Forsberg, Senior Lecturer

Room: U1-081
Phone: +46-21-101381