Abstract
We consider the problem of multi agents cooperating in a partially-observable environment. Agents must learn to coordinate and share relevant information to solve the tasks successfully. This article describes Asynchronous Advantage Actor-Critic with Communication (A3C2), an end-to-end differentiable approach where agents learn policies and communication protocols simultaneously. A3C2 uses a centralized learning, distributed execution paradigm, supports independent agents, dynamic team sizes, partially-observable environments, and noisy communications. We compare and show that A3C2 outperforms other state-of-the-art proposals in multiple environments.

Abstract
Intelligent wheelchairs allow individuals to move more freely, safely, and facilitate users’ interaction with the wheelchair. This paper presents the results focused on the study and analysis of the state of the art related to topics as interaction, interfaces, intelligent wheelchairs and the analysis of the Intellweelsproject. The main goal is to create and implement a multimodal adaptive interface to be used as the control and interaction module of an intelligent wheelchair. Moreover, it will be important to have in mind the usability, by facilitating the control of a complex system, interactivity, by allowing the control using diverse kinds of input devices, and expansibility, by integrating easily with several intelligent external systems. This project features a complex input/output system with linked parameters simplified by a node system to create the input/output actions with automatic input recording and intuitive output association as well as a powerful, device-agnostic design, providing an easy way to extend the inputs, outputs and event the user interface. Results reveal a positive users’ feedback and a responsive way when using the multimodal interface in simulated environment. © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2020.

Abstract
Pokemon is one of the most popular video games in the world, and recent interest has appeared in Pokemon battling as a testbed for AI challenges. This is due to Pokemon battling showing interesting properties which contrast with current AI challenges over other video games. To this end, we implement a Pokemon Battle Environment, which preserves many of the core elements of Pokemon battling, and allows researchers to test isolated learning objectives. Our approach focuses on type advantage in Pokemon battles and on the advantages of delayed rewards through switching, which is considered core strategies for any Pokemon battle. As a competitive multi-agent environment, it has a partially-observable, high-dimensional, and continuous state-space, adheres to the Gym de facto standard reinforcement learning interface, and is performance-oriented, achieving thousands of interactions per second in commodity hardware. We determine whether deep competitive reinforcement learning algorithms, WPL theta and GIGA theta, can learn successful policies in this environment. Both converge to rational and effective strategies, and GIGA theta shows faster convergence, obtaining a 100% win-rate in a disadvantageous test scenario.

Abstract
Tackling multi-agent environments where each agent has a local limited observation of the global state is a non-trivial task that often requires hand-tuned solutions. A team of agents coordinating in such scenarios must handle the complex underlying environment, while each agent only has partial knowledge about the environment. Deep reinforcement learning has been shown to achieve super-human performance in single-agent environments, and has since been adapted to the multi-agent paradigm. This paper proposes A3C3, a multi-agent deep learning algorithm, where agents are evaluated by a centralized referee during the learning phase, but remain independent from each other in actual execution. This referee's neural network is augmented with a permutation invariance architecture to increase its scalability to large teams. A3C3 also allows agents to learn communication protocols with which agents share relevant information to their team members, allowing them to overcome their limited knowledge, and achieve coordination. A3C3 and its permutation invariant augmentation is evaluated in multiple multi-agent test-beds, which include partially-observable scenarios, swarm environments, and complex 3D soccer simulations.

Abstract
Digital tools with an emphasis on the so-called apps are a current topic whose potential, in school context, is still little explored. On the other hand, the formative evaluation is also very little explored and is of particular importance in the context of inclusive education, as an extended vision, in which all students have specificities. The review of the literature in the context of the cross-referencing of the two topics - formative evaluation and the use of apps - presents good indicators of how technology can complement the formative evaluation, guarantee a greater rooting of the same and guarantee performance more aligned with education inclusive, taking into account the profile of the student after leaving compulsory schooling. In this context, a descriptive and analytical study was carried out, using a survey on the use of digital tools and formative evaluation. The results obtained allowed us to conclude that the school environment will have to evolve, above all, to the level of material resources. This evolution is less pressing at the level of human resources and attitudes that promote the use of formative assessment techniques (FATs) and apps. Thus, there is an opportunity to improve existing applications in order to allow greater aid of formative evaluation, by attenuating its greater limitations. Due to the evolution of the applications, the prospects of gaining more use of the apps and FATs are widened. © 2020, The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG.

Abstract
This paper describes an application aimed at improving the current state of enforcement in the areas of food safety and economic activities in Portugal. More specifically, the application focuses on a flexible and interactive approach to generate inspection routes, to be followed by surveillance brigades with the aim of verifying Economic Operators’ compliance to national and European legislation on economic and food safety. The problem is modeled as a Multi-Depot Periodic Vehicle Routing Problem with Time Windows, and the algorithmic approaches employed seek to maximize either the number of inspected Economic Operators or a utility function that takes into account the utility gained from inspecting each Economic Operator. The generated solutions are shown in an intuitive platform, where human operators can visualize the solutions details (including georeferenced locations in a map) and fully customize them on time by manually removing or adding Economic Operators to be targeted. © 2020, The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG.