1. Pedro Sá Costa, Architecture and communication systems for autonomous vehicles,
Recent traffic accidents involving autonomous vehicles cause great concern regarding safety and reliability in autonomous driving. These facts encourage research to improve the efficiency of the perception systems of these vehicles. Perception refers to the ability of an autonomous vehicle to collect information and extract relevant knowledge from the environment. In turn, cooperative perception is the sharing of local perception information with other vehicles or infrastructure through wireless communications. This cooperation impacts decision making and driving planning for increased reliability, safety, and reduced road congestion. In this proposal, we intend to elaborate a computational perception system for autonomous vehicles based on the cooperation capable of complementing the current systems to increase the confidence and safety for autonomous driving. This novel perception system will allow to detect moving and stationary objects to generate an obstacle map of a given geographical area. The combination of the local and remote information will give vehicles a more efficient perception system. This system cooperation will result from the data exchange between vehicles, and mobile edge computing infrastructure.
2.Milos Stankovic, Decentralized Cooperative Learning and Control for Networked Dynamical Systems
The subject of the proposal belongs to the emerging area of Intelligent Networked Cyber-Physical Systems (INCPS) dealing with complex, spatially distributed and networked heterogeneous multi-agent dynamical systems, representing one of the greatest challenges in modern science and technology. The impacts of INCPS will be ground-breaking, revolutionary and pervasive; this is evident today in emerging applications such as swarms of autonomous vehicles, smart buildings, cities and power grids, intelligent agriculture, transportation and manufacturing systems. Dimensionality, uncertainty, potential vulnerability, and information structure constraints, as fundamental characteristics of these systems, have led to the development of the decentralized decision making theory, providing scalability and robustness to structural uncertainties. However, high level of decentralization may increase vulnerability and decrease performance of the overall system. The general objective of the project is development of new methods, algorithms and practical tools for decentralized learning and intelligent control for resilient INCPS, based on applications of consensus techniques, robust statistics theory, game theory and reinforcement learning. More specifically the objectives are: 1) Development of decentralized and distributed estimation algorithms, for resilient cooperative learning in INCPS; and 2) Development of novel decentralized algorithms for cooperative multi-agent reinforcement learning control. Besides the short-term (1 year) expected results, the formulated objectives should also be considered in a long-term sense, having in mind the complexity, interdisciplinarity and generality of the approach.
3. Slavisa Tomic, Distributed Secure Localization
The fifth generation (5G) of networks is expected to provide significantly higher bandwidth and faster data rates with potential for interconnecting myriads of heterogeneous devices (sensors, agents, users, machines, and vehicles) into a single network (of nodes), under the notion of Internet of Things. The ability to accurately determine the physical location of each node (stationary or moving) will permit rapid development of new services and enhancement of the entire system. In many outdoor environments, this could be achieved by employing global navigation satellite system (GNSS) which offers worldwide service coverage with good accuracy. However, installing a GNSS receiver on each device in a network with thousands of nodes would be very expensive in addition to energy constraints. Besides, in indoor or obstructed environments (e.g., dense urban areas, forests, and canyons) the functionality of GNSS is limited to non-existing, and alternative methods have to be adopted. Many of the existing alternative solutions are centralized, meaning that there is a sink in the network that gathers all information and executes all required computations. This approach quickly becomes cumbersome as the number of nodes in the network grows, creating bottle-necks near the sink and high computational burden. Therefore, more effective approaches are needed. As such, this project aims at developing novel distributed solutions for target localization in large-scale networks, in which nodes have restricted energy resources. Besides guaranteeing good localization performance (both in terms of localization accuracy and computational complexity), the main goal of the project will be to provide secure solutions (localization in malicious environments, i.e., in the presence of one or more internal/external attackers whose objective is to impede our fundamental desire to achieving high accuracy). This malicious setting raises the bar even higher in terms of difficulty of the problem, but is of paramount importance in many practical applications.
In the 2018 FCT Individual Call to Scientific Employment Stimulus Copelabs has been awarded 1 Assistant Researcher position.
Some of the new research topics in 2020 are: distributed reinforcement learning, massive MIMO, variational inference and big data, compressive sensing, grant-free random access protocols for 5G and beyond.
Prof. Sanja Vranjes, Prof. Joel Rodrigues and Prof. Marco di Renzo accepted the invitation to serve on the External Advisory Board for Copelabs.
In 2018 Copelabs members produced 71 publications in peer-reviewed international journals and/or conference proceedings, out of which 16 are Q1 (Scimago quartile) journals and 10 are Q2 journals.