About the project
Objective
This research (3Dfire) aims to enhance forest fire detection and characterization through advanced remote sensing and AI techniques, facilitating improved management and mitigation strategies within forest ecoregions. Our AI algorithms will streamline the processing of extensive freely available remote sensing data, like Sentinel-1 and -2, to drastically reduce the time and costs associated with fire monitoring and prediction. Consequently, it will provide a comprehensive understanding of fire dynamics within the region. Additionally, “3Dfire” will identify and analyze the risk factors and drivers of fires in the Miombo woodlands, including the impact of extreme climate events and anthropogenic factors on fire severity, frequency, and duration.
The project’s outcomes will offer valuable insights to researchers and policymakers responsible for forest fire prevention, minimizing societal and ecological impacts, and promoting sustainable forest management practices. Ultimately, “3DFire” is pivotal in establishing a comprehensive global-scale perspective on AI for forest applications and transitioning towards climate-smart forest management.
Background
Miombo woodlands cover 270 million ha across southern Africa and are increasingly threatened by natural and anthropogenic forces. Despite their importance for biodiversity and 100 million forest-dependent people, this ecoregion receives little attention from the scientific community. The annual forest loss in just a part of the Miombo ecoregion was recently estimated to be more than half a million hectares. Fires are a primary cause of vegetation loss in Miombo, occurring approximately 50% more frequently than in other global ecoregions. They also trigger massive CO2 emissions in sub-Saharan Africa and threaten many ongoing sustainable forest management projects, such as Reducing Emissions from Deforestation and Degradation (REDD). The lack of accurate fire records challenges the estimation of CO2 emissions and evaluation of fire management activities in forests.
Although some dimensions of Miombo’s wildfires, such as frequency and severity, have been explored through coarse- and medium-resolution satellite time-series data from MODIS and Landsat, knowledge of the exact spatial extent, duration, and timing of the fires and their relations with effective driving forces is still lacking. However, fire regimes in miombo are dominated by small burn patches, particularly in the drier regions. Their limited spatial resolution makes it difficult to detect and characterize these small fires using the aforementioned satellites. Extreme climate events significantly elevate the probability of forest fires, while anthropogenic drivers affect all dimensions of fires. Meanwhile, extreme climate events and human pressures on forests are expected to increase as global temperatures rise and the population of southern Africa doubles by 2050. To address these limitations, this research will focus on characterizing and mapping various dimensions of fires, including the severity, frequency, timing, and duration of forest fires in miombo. We will leverage the time series data of fires derived from the Sentinel data and employ deep learning-based algorithms, particularly our developed residual attention UNet5 (RAUNet5).
About the Digital Futures Postdoc Fellow
Zeinab Shirvani is a postdoctoral researcher affiliated with the Division of Geoinformatics at KTH. She obtained a PhD in Remote Sensing/Cartography focusing on forest disturbances from TUD, Germany, in 2020. Before her current position, Zeinab was a postdoctoral researcher at the Swedish University of Agricultural Sciences (SLU) for two years. Zeinab has made significant contributions to remote sensing and machine learning throughout her research career, particularly in mapping woodland fires in tropical dry forests using RAUNet. Her expertise lies in applying geospatial artificial intelligence and remote sensing techniques to study forest disturbances.
Main supervisor
Yifang Ban, Professor and Director of the Division of Geoinformatics at the Department of Urban Planning and Environment at KTH.
Co-supervisor
Ulla Mörtberg, Professor and associate professor (docent), Department of Sustainable Development, Environmental Science and Engineering (SEED) at KTH.
Watch the recorded presentation at the Digitalize in Stockholm 2023 event.
About the project
Objective
This project aims to solve theoretical and practical challenges in distributed optimization and learning in smart networked systems. We wish to design fast and practical algorithms that have theoretical convergence guarantees. We are focusing on two concrete topics: 1) safe resource allocation in power networks to avoid systems breakdown and 2) efficient asynchronous parallel and distributed optimization (better step sizes and delay-tolerant algorithm design).
Background
Networked systems such as power networks and IoT systems are important in our life. To make these systems “smart” (e.g., saving cost or improving utility), we need to learn models from data, equivalent to solving optimization problems. Consequently, designing efficient algorithms to solve optimization problems in these systems is of strong practical significance. Moreover, theoretical convergence analysis is also dispensable to these algorithms to guarantee their reliability.
About the Digital Futures Postdoc Fellow
Xuyang Wu is a Postdoctoral researcher at KTH Digital Futures, co-supervised by Prof. Mikael Johansson at KTH (DCS, EECS) and Prof. Sindri Magnússon at SU (DSV). He received a B.S. degree in Applied Mathematics from Northwestern Polytechnical University, China, in 2015 and a PhD in Communication and Information Systems at the University of Chinese Academy of Sciences, China, in 2020. He was a finalist for the best student paper award at IEEE ICCA 2019. His current research interests include distributed optimization and federated learning. In particular, he focuses on algorithmic foundations, convergence analysis, and resource efficiency in emerging systems such as IoT and cyber-physical systems. More information can be found on his homepage: http://xuyangwu.github.io/
Main supervisor
Mikael Johansson, Professor, Division of Decision and Control Systems, School of EECS, KTH.
Co-supervisor
Sindri Magnússon, Associate Professor, Department of Computer and Systems Science, Stockholm University.
Watch the recorded presentation at Digitalize in Stockholm 2022 event.
About the project
Objective
The project’s main goal is to analyze the sociotechnical challenges that arise from the design and use of data-intensive methods for advocacy purposes. The project will advance a core understanding of how data-driven techniques support or hinder community engagement and political organizing in disabled communities. It contributes to long-standing concerns of promoting data literacy and making technologies accessible to historically marginalized populations. In addition to an empirical understanding of the local needs of disabled communities, this work also advances technical knowledge of designing for and with people with disabilities as applied to data science tools.
Background
The current data ecosystem consists of techniques for collecting and analyzing data influencing wide-ranging aspects of our social lives, from healthcare decisions to housing allocations. In these data systems, people with disability are often further marginalized in the increasingly datafied society. In recent years, marginalized communities have been leveraging the same data-intensive methods to combat the adverse effects of big data systems, advocating for under-recognized issues and driving social change. This project examines what happens when traditional data-intensive methods are employed for advocacy efforts, such as when activists mobilize statistics to shed light on issues or create data visualizations to engage big audiences. The goal is to interrogate the limits and opportunities of data science as we know it when applied to disabled communities. I aim to understand data needs specific to these communities and to design and deploy a suite of tools to support their efforts.
About the Digital Futures Postdoc Fellow
Stacy Hsueh is a postdoc in the Interaction Design team at KTH Royal Institute of Technology. She’s interested in exploring how crip wisdom offers critical frameworks for interrogating inequity in data-driven technologies and imagining more just alternatives.
Main supervisor
Marianela Ciolfi Felice, Assistant Professor in Interaction Design at KTH.
Co-supervisor
Karin Hansson, Researcher at Stockholm University.
Watch the recorded presentation at Digitalize in Stockholm 2022 event.
About the project
Objective
This research aims to extract Parkinson’s disease (PD) related temporal features from brain activity and develop brain activity-based biomarkers for diagnosis and prognosis by leveraging advances in machine learning that can form the basis for classifying PD patients.
Background
PD is the second most common neurodegenerative disorder with debilitating consequences. PD diagnosis is based on behavioural symptoms and chemical/genetic markers. Such a diagnosis can be erroneous and often does not capture the disease severity. Despite being a brain disease, the brain’s electrical activity (e.g. fMRI, EEG/MEG) is completely ignored in PD diagnosis. This oversight is intriguing, considering experimental evidence indicating that PD-related changes profoundly impact the temporal dynamics of brain activity.
While biomarkers derived from brain activity usually focus on frequency domain parameters like oscillations and coherence, temporal features have been surprisingly neglected. In fact, except for certain forms of epilepsy, usually, brain activity is not considered in the diagnosis of brain diseases. However, currently, brain activity-based biomarkers are not specific enough. This may be because these are defined in the frequency domain (e.g. oscillations and coherence). Unlike in epilepsy diagnosis, temporal features of brain activity are ignored. Therefore, in my project I would like to change this. I hypothesize that brain region-specific temporal features will provide highly specific information about the disease severity and can provide a more precise diagnosis and prognosis of PD.
About the Digital Futures Postdoc Fellow
Satarupa Chakrabarti, after completing her postgraduation in Computer Science engineering, undertook a PhD degree in Computer Science from KIIT Deemed to be University, Bhubaneswar, India, where she defended her thesis related to designing a generalized epileptic seizure detection method with different feature-extracting techniques from biosensor data that showed the presence of temporal changes. She simultaneously worked as a Junior Research Fellow (JRF) with DST-SERB, India, in an interdisciplinary project (2018–2020).
She has a strong and diverse background in research, with primary research interests in biomedical engineering, signal processing, image processing, space physics, machine learning and deep learning. Satarupa is a Digital Futures Postdoctoral Research Fellow at KTH, based in the Computational Brain Science group at the Division of Computational Science and Technology. Her research experiences involved collaboration with multidisciplinary teams and conducting research on various electrical engineering and space physics projects that led to innovative, beneficial, and cutting-edge endeavours.
Main supervisor
Arvind Kumar, Associate Professor, Division of Computational Science and Technology at KTH.
Co-supervisor
Saikat Chatterjee, Associate Professor, Division of Information Science and Engineering at KTH.
About the project
Objective
Wildfire monitoring involves two main problems, i.e., active fire detection and burnt area mapping. Active fire detection aims to find the ongoing wildfire hotspots, while burnt area mapping is expected to detect the areas affected by wildfire.
This project mainly focuses on large-scale wildfire burnt area mapping and near real-time wildfire monitoring. In view of the limited transfering performance of the existing wildfire monitoring algorithms on a larger scale and various climate zones, this project aims to develop large-scale or even globally applicable models by exploiting global coverage, multi-source satellite remote sensing data and advanced machine learning/deep learning techniques.
Background
Wildfire has coexisted with human societies for more than 350 million years, always playing an important role in shaping the Earth’s surface and climate. Across the globe, wildfires are becoming larger, more frequent, longer-duration, and tend to be more destructive in terms of lives lost and economic costs because of climate change and human activities.
To reduce the damages from such destructive wildfires, it is critical to track wildfire progressions in near real-time, or even in real-time, to support fire-fighting and keep everything under control. Satellite remote sensing enables cost-effective, accurate, and timely monitoring of the wildfire progressions over vast geographic areas. The free availability of global coverage Landsat-8, and Sentinel-1/2 satellite data opens a new era for global land surface monitoring, providing an opportunity to analyse wildfire impacts around the globe.
About the Digital Futures Postdoc Fellow
Puzhao Zhang is a research fellow at the Division of Geoinformatics, Royal Institute of Technology. He received his PhD in Pattern Recognition and Intelligent Systems from Xidian University, China, at the end of 2019. His PhD research was focused on remote sensing and deep learning for change detection.
Since fall of 2017, he has worked on wildfire monitoring with radar and optical remote sensing and deep learning as a joint PhD student at KTH. His research interests include satellite imagery analysis, change detection, machine learning, deep learning, and spatio-temporal modelling for monitoring environmental change and biomass carbon dynamics.
Main supervisor
Yifang Ban, Professor, Division of Geoinformatics, KTH.
Co-supervisor
Josephine Sullivan, Associate professor, Division of Robotics, Perception and Learning, KTH.
Watch the recorded presentation at Digitalize in Stockholm 2022 event.
About the project
Objective
Our research project aims to propose a complete, fully distributed scheme in intelligent task allocation and security-based controller synthesis with relative displacement or bearing-only information for networked multi-agent systems like UAVs or UGVs. The designed methods are expected to be used directly in multiple unmanned autonomous systems. The research plays a significant role in improving the autonomy and swarming of the pursuit system and has great application potential in search and rescue, intelligent transportation, and public protection.
Background
Multiple autonomous unmanned systems, such as unmanned aerial vehicles, unmanned underwater vehicles, and unmanned ground vehicles, are projected to play important roles in many industrial and societal applications, such as search and rescue, cooperative payload carrying, and warehouse management. In the community of multi-agent control, the pursuit-evasion game has been a hot topic whose objectives are to achieve optimal sensor placement intelligent and distributed task allocation, following which the control problem is formulated as a distributed optimization problem with a defined performance index. The objectives of this project thus include the exploration of cooperative and distributed optimal control strategies with relative position measurements or bearing-only information.
About the Digital Futures Postdoc Fellow
Panpan Zhou is a postdoctoral researcher at the Department of Mathematics at KTH Royal Institute of Technology. Before joining KTH, she was a Postdoctoral Fellow in the Department of Mechanical and Automotive Engineering at the Chinese University of Hong Kong. She received her Bachelor’s degree in the School of Automation from Northwestern Polytechnical University, Xi’an, China, in 2017 and her PhD from CUHK in 2021. Her research interests include control theory and applications of multi-agent systems and motion planning of micro aerial vehicles.
Main supervisor
Xiaoming Hu, professor, Department of Mathematics, KTH.
Co-supervisor
Bo Wahlberg, professor, Division of Decision and Control Systems, KTH.
Watch the recorded presentation at the Digitalize in Stockholm 2023 event.
About the project
Objective
The objective of this project is to learn distributed control policies for multi-robot systems that scale on-demand, in multi-laterally evolving complex dynamic environments.
Background
Multi-robot systems such as drone swarms offer unparalleled advantages in assisting ground-based human-robot teams in humanitarian, and disaster-response missions thanks to their ability to operate in remote, communication-denied environments. However, in practical deployments such systems must learn how to balance various auxiliary objectives on-demand for maximizing the collective utility in highly dynamic environments. For example, a robust drone swarm control policy must prioritize the most severely affected area in a disaster-response event, and direct more robots to tackle it, on-the-fly.
In this project, we aim to design a framework that can help us learn such control policies while taking into account the robot’s varying capabilities. Upon designing such a framework, we further understand that we can minimize the human bias in hand-engineered task prioritizing.
About the Digital Futures Postdoc Fellow
Malintha Fernando received his Ph.D in 2023 from Indiana University, Bloomington (USA). His doctoral research focused on designing cooperative scalable control policies for multi-drone systems that are robust to communication failures. Such policies lend themselves to many Urban Air Mobility (UAM) applications such as autonomous parcel delivery, where decentralized decision-making under local information is required to achieve the necessary scalability over the geographical span and in the number of vehicles.
Prior to joining KTH, Malintha worked as a visiting lecturer in Machine Learning at Indiana University. He has completed an internship at Open Robotics, Mountain View, and completed his undergraduate education from University of Moratuwa, Sri Lanka.
Main supervisor
Silun Zhang, Assistant Professor, Division of Optimization and System Theory, Department of Mathematics, KTH.
Co-supervisor
Petter Ögren, Professor, Division of Robotics, Perception and Learning (RPL) at KTH.