Distributed Perception in the TROCA Environment / Proposal linked to the Center for Engineering Research (CPE) Proc. 2021/00199-8: SMART Networks and ServiceS for 2030 (SMARTNESS)

• Grant number: 25/03499-3
• Support Opportunities: Scholarships in Brazil - Doctorate
• Start date: August 01, 2025
• End date: July 31, 2029
• Field of knowledge: Engineering - Electrical Engineering
• Principal Investigator: Ricardo Ribeiro Gudwin
• Grantee: José Renato Borelli
• Host Institution: Faculdade de Engenharia Elétrica e de Computação (FEEC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Company: Universidade Estadual de Campinas (UNICAMP). Faculdade de Engenharia Elétrica e de Computação (FEEC)
• Associated research grant: 21/00199-8 - SMART NEtworks and ServiceS for 2030 (SMARTNESS), AP.PCPE

Abstract

The TROCA Project (Gudwin et al., 2020a; Gudwin et al., 2020b) is a research project previously developed at UNICAMP, under the guidance of Prof. Gudwin in collaboration with Ericsson during the period from 2018 to 2020. In this project, a cognitive architecture was developed for the control of transport robots in a smart factory environment.A Cognitive Architecture is a software architecture that integrates multiple modules or components into a system capable of performing cognitive tasks, using specific models from cognitive psychology to establish the internal functioning of each of these modules and integrate them, combining different strategies and technologies in an architectural way to build the system as a whole. Federated Learning (Aledhari et al., 2020; Zhang et al., 2021) is a research field within the Machine Learning area, in which the learning task is distributed among different agents, with one of the agents assuming the role of coordinator, sending learning subtasks to the other participants in a learning team, distributing the processing burden and integrating the multiple contributions of each team member to consolidate integrated knowledge.In this research project, we intend to use the TROCA environment to investigate a problem that we are calling: "Distributed Perception", in a direct analogy to what occurs in federated learning. The idea is to have multiple cognitive agents, each with limited access to sensors and restricted processing capacity (as, for example, in an IoT device), with one of these agents usually operating on more powerful hardware and assuming the role of federation coordinator. These cognitive agents would be able to communicate with each other, exchanging their limited perceptions, so that the coordinator can integrate these multiple contributions, creating a summarized and integrated perception, in order to derive a common understanding about the events occurring in the environment. This includes a description of multiple robots navigating the factory environment, as well as people moving in the same environment, providing a unified view of the movements of packages, robots and human users in the factory, making this information available to the cognitive architecture, in order to subsidize its decision-making process. (AU)