A Dataset Schema for Cooperative Learning from Demonstration in Multi-robot Systems

Abstract

Multi-Agent Systems (MASs) have been used to solve complex problems which demand intelligent agents working together to reach the desired goals. These Agents should effectively synchronize their individual behaviors so that they can act as a team in a coordinated manner to achieve the common goal of the whole system. One of the main issues in MASs is the agents' coordination, being common domain experts observing MASs execution disapprove agents' decisions. Even if the MAS was designed using the best methods and tools for agents' coordination, this difference of decisions between experts and MAS is confirmed. Therefore, this paper proposes a new dataset schema to support learning the coordinated behavior in MASs from demonstration. The results of the proposed solution are validated in a Multi-Robot System (MRS) organizing a collection of new cooperative plans recommendations from the demonstration by domain experts. Keywords Multi-Agent System · Learning from Demonstration · Dataset · Coordination · Multi-robot plan · clustering PACS 07.05.Mh Dataset Schema for Cooperative Learning in a MRS 3Learning from Demonstration (LfD) has been used to learn robots basic skills or even very simple setplays in a MRS [11]. Although, there is no register of using LfD to learn complex setplays. Therefore, this work proposes using LfD to offer domain experts a chance to watch robotic soccer matches and suggest new setplays for each situation for which they think the MRS has made a bad decision.Section 3 presents the state-of-the-art for learning coordinated plans in MAS. One of the main issues in LfD for setplays is the nature of the dataset generated from the domain experts recommendation. Some features in this dataset are not of primitive types as scalars or strings, but some complex types, such as objects, structures, trees, etc. Thus, we also define a strategy presented in Section 2.4, to handle this kind of complex data.The proposed solution has a two-level dataset, detailed in Section 4. To assess the feasibility of using this dataset to support setplays learning, the Fuzzy C-Means (FCM) algorithm is used to organize setplays recommendations into clusters. The choice of the FCM algorithm is due to the imprecision inherent in the friendly interface proposed for use by experts to generate the recommendations of setplays. The suggestions from experts are organized in clusters to solve the semantic equivalence issue presented in Section 2. Section 5 describes the assessment process and its results. Section 6 has conclusion and future work descriptions.

BahiaRT Setplays Collecting Toolkit and BahiaRT Gym

“…The BahiaRT Setplays Collecting Toolkit integrantes RoboViz and SPlanner to allow users to watch several game logs and generate setplays demonstrations that can be submitted to compose a setplays dataset to be used in reinforcement learning approaches [5,4].…”

Multi-Robot Coordination Analysis, Taxonomy, Challenges and Future Scope

“…In some cases, edge devices can also be used for running ML algorithms. Other solutions are to use cooperative learning [305] and distributed machine learning [306]. & Autonomous and Transfer Learning: Autonomous learning is less explored for MRS as compared to MAS.…”

Quantifying reputation risk using a fuzzy cognitive map: a case of a pharmaceutical supply chain

“…FCMs are dynamic supervised learning fuzzy-neural systems that are increasingly gaining momentum in business (Zare Ravasan and Mansouri, 2016;Jamshidi et al, 2018), medicine (Subramanian et al, 2015;Salmeron et al, 2019), robotics (Simões et al, 2019), environment (Papageorgiou et al, 2011) and information technology (Das et al, 2018) domains. The FCM approach consists of three major steps: (1) identification of the input concepts that influence the output concepts (2) establishment of the causal weight influence between different concepts (3) validation of the developed FCM model through "what-if" analyses.…”