Goals and Objectives of PROSDS
Given the fundational research nature of the project, the expected results are fundamental and non-applicative in nature.
We expect to contribute to RESTART and SUPER objectives and in particular to the study of the “… Definition of innovative techniques for accelerated, fully automated and flexible mobile networks and services” as explicitly stated in SUPER objectives.
In particular our efforts will be devote to some of the following goals:
- Go beyond the current definition of collaborative perception essentially based on vision or the exchange of raw or partially processed data from ranging instruments (Radar, LiDAR) to include a broader vision of the concept of perception (trajectories, intentions, predictions, …);
- Include ethical considerations in the concept of perception as a constraint on the problem, in order to prevent and/or mitigate damage resulting from situations in which it is not possible to completely avoid the damage itself;
- Offer general and/or specific models of the required information and the methodologies necessary to integrate it to obtain specific perceptual objectives, for example to avoid collisions it may not be necessary to transmit images, but a spatial model of the colliding objects or actors may suffice;
- Propose techniques to alocate computing and communication resources to achieve one or more of the objectives described above; • Evaluate the performance of some of the proposals made both in terms of communication and information processing system and in terms of impact on traffic, safety and optimization of mobility space.
Structure
PROSDS is a small project with limited resources. It organized in two Work Packages running mostly in parallel.
WP1: Models and Algorithms.
Objectives:
- To study cooperative perception systems for SDS from different points of view: of distributed systems, of the telecommunications network necessary for the exchange of information, the optimization of the use of resources (roads, fuel, etc.), the minimization of computing resources; of automotive electronics;
- To propose models and algorithms (optimization, convergence and consensus, decision-making, etc.) that correctly represent the systems studied also with the use of Artificial Intelligence (AI) or Machine Learning (AA) techniques.
WP2: Performance Analysis
Objectives:
- To define the basic criteria (metrics) for evaluating the performance of a collaborative perception system for SDS in light of the models defined in WP1, render in engineering and quantitative terms the ethical criteria available in the literature on the topic;
- To evaluate some of the algorithms and methodologies identified in WP1 in relation to the different metrics identified, with particular attention to the aspects of safety (including any ethical considerations when it is impossible to avoid damage), efficiency of the use of infrastructure, energy saving both for vehicles and for the global communication and computing system;
- To inform WP1 on the achievable objectives and on the trade-offs imposed on the different metrics by the complexity of the system studied (e.g., increase in computational complexity vs. precision and coherence of distributed decisions).
Participants
Prof. Renato Lo Cigno PROSDS PI
Prof. Renata Mansini
Prof. Anna Richelli
Prof. Marina Zanella