Abstract
Driving simulators are essential tools for training, education, research, and scientific experimentation. However, the diversity and quality of virtual environments in simulations is limited by the specialized human resources availability for authoring the content, leading to repetitive scenarios and low complexity of real-world scenes. This work introduces a pipeline that can process text-based narratives outlining driving experiments to procedurally generate dynamic traffic simulation scenarios. The solution uses Retrieval-Augmented Generation alongside local open-source Large Language Models to analyse unstructured textual information and produce a knowledge graph that encapsulates the world scene described in the experiment. Additionally, a context-based formal grammar is generated through inverse procedural modelling, reflecting the game mechanics related to the interactions among the world entities in the virtual environment supported by CARLA driving simulator. The proposed pipeline aims to simplify the generation of virtual environments for traffic simulation based on descriptions from scientific experiment, even for users without expertise in computer graphics. © 2025 by SCITEPRESS–Science and Technology Publications, Lda.

Abstract
Creating content for digital video game is an expensive segment of the development process, and many techniques have been explored to automate it. Much of the generated content is graphical, ranging from textures and sprites to typographical elements and user interfaces. Numerous techniques have been explored to automate the generation of these assets, with recent advancements incorporating artificial intelligence methodologies such as deep learning generative models. This study comprehensively surveys the literature from 2016 onwards, focusing on using machine learning to generate image-based assets for video game development, reviewing the deep learning approaches employed, and analyzing the specific challenges found. Specifically, the deep learning approaches employed, the problems addressed within the domain, and the metrics used for evaluating the results. The study demonstrates a knowledge gap in generative methods for some types of video game assets. Additionally, applicability and effectiveness of the most used evaluation metrics in the literature are studied. As future research prospects, with the increase in popularity of generative AI, the adoption of such techniques will be seen in automation processes. © 2018 IEEE.

Abstract
Intelligent wheelchairs aim to improve mobility limitations by providing ingenious mechanisms to control and move the chair. This paper aims to enhance the autonomy level of intelligent wheelchair navigation by applying reinforcement learning algorithms to move the chair to the desired location. Also, as a second objective, add one more chair and move both chairs in pairs to promote group social activities. The experimental setup is based on a simulated environment using gazebo and ROS where a leader chair moves towards a goal, and the follower chair should navigate near the leader chair. The collected metrics (time to complete the task and the trajectories of the chairs) demonstrated that Deep Q-Network (DQN) achieved better results than the Q-Learning algorithm by being the unique algorithm to accomplish the pair navigation behaviour between two chairs.

Abstract
Content generation is a heavy task in virtual worlds design. Procedural content generation techniques aim to agile this process by automating the 3D modelling with some degree of parametrisation. The novelty of this work is the procedural generation of the marine alga (Asparagopsis armata), taking into consideration the underwater environmental factors. The depth and the occlusion were the two parameters in this study to simulate how the alga growth is influenced by the environment where the alga grows. Starting by building a prototype to explore different L-systems categories to model the alga, the stochastic L-systems with parametric features were selected to generate different alga plasticities. Qualitative methods were used to evaluate the designed grammar and alga's animation results by comparing videos and images of the Asparagopsis armata with the computer-generated versions. © 2022 University of West Bohemia. All rights reserved.

Abstract
The preservation and sustainability of the marine ecosystem could benefit from the surge of new technologies to design autonomous vehicles. These underwater robots operate in a complex environment where the loss of human lives is highly probable. Consequently, a considerable percentage of the ocean remains unexplored due to the complexities of the underwater environment. Robotics can be a solution to overcome these limitations. However, training these complex systems is challenging and resource expensive. Human-in-the-loop input is essential in accelerating the training process by teaching the robots how to perform in specific scenarios and validate the simulated environment. This work presents a case study that simulates the dynamics of a Remotely Operated Vehicle in an underwater environment and uses imitation learning to train the vehicle to navigate autonomously toward a target. It was possible to measure and observe the similarity between the expert and the autonomous trajectories generated by the ROV. However, the imitation learning performance cannot surpass the expert, considering the time and the number of successes in finding the target. © ESM 2022. All rights reserved.