Luan Silveira

Underwater SLAM: Challenges, state of the art, algorithms and a new biologically-inspired approach

The unstructured scenario, the extraction of significant features, the imprecision of sensors along with the impossibility of using GPS signals are some of the challenges encountered in underwater environments. Given this adverse context, the Simultaneous Localization and Mapping techniques (SLAM) attempt to localize the robot in an efficient way in an unknown underwater environment while, at the same time, generate a representative model of the environment. In this paper, we focus on key topics related to SLAM applications in underwater environments. Moreover, a review of major studies in the literature and proposed solutions for addressing the problem are presented. Given the limitations of probabilistic approaches, a new alternative based on a bio-inspired model is highlighted.

Underwater Visual 3D SLAM Using a Bio-inspired System

Considering the various challenges in robotics, one of the most important concerns to a mobile robot build a map and estimate its location in an environment. The SLAM techniques build a map and simultaneously maintain the current location of a robot in the same time. Nowadays, probabilistic approaches dominate the field, anyway the last decade have witnessed important bio-inspired studies based on biological structures related to spatial navigation. Past and recent studies found place, head direction and grid cells, among others, related to the tasks of mapping and location in mammals. Continuous Attractor Neural Networks (CANN) are being proposed to simulate the SLAM performed by this brain structures. This work present a bio-inspired SLAM system to map a sub-aquatic 3D environment, through a simulation of an underwater robot equipped with a camera, using visual references to mimic the behavior of location tasks performed by mammals.

Autonomous navigation for underground energy line inspection robot

This work proposes architecture of an inspection robots navigation system, aiming at monitoring underground energy lines. This architecture is composed of two modules: i. feature extraction from environment; ii navigation approach. The feature extraction module is based on the use of the edge detector by Canny algorithm and Hough transform for identification of lines from images of environment to monitoring. The lines identified correspond to cable conformation inside the duct. This information will serve to help the navigation system. For the implementation of the navigation system two approaches were proposed: navigation based on artificial neural network and navigation based on PID control. The navigation architecture can be used in real or simulated scenarios, and it was tested in a simulated environment.

3D robotic mapping: A biologic approach

Mapping a 3D environment is a big challenge for roboticists, expecially in underwater environments. Nowadays, the most applied solution to this problem relies in Probabilistic Filters, but with the discovery of neurons in the mammalian brain associated with navigation tasks, biological approaches has been take place. This paper presents a system inspired in mammalian brain to solve the problem of mapping and localization of robots. Preliminaries results in simulated environments shows the relevance of the proposed method, which is highly parallelizable and capable of running in real time applications.

Challenges and State-of-the-Art Solutions to Underwater Slam

The unstructured scenario, the extraction of significant features, imprecision of sensors along with the impossibility of using GPS signals are some of the challenges encountered in underwater environments. Given this adverse context, the Simultaneous Localization and Mapping techniques (SLAM) attempt to localize the robot in an efficient way in an unknown underwater environment while at the same time, generate a representative model of the environment. In this paper, we focus on key topics related to SLAM applications in underwater environments. Moreover, a review of major studies in the literature and proposed solutions for address the problem are presented. As the main contribution of this work, a comprehensive review of the applications and topics in underwater SLAM were produced. Future trends and topics for research are also related.

Navigation System for an Underground Distribution Inspection Platform Using Simulation

This work proposes an architecture of an inspection robots navigation system, aiming at monitoring of power cables in underground distribution lines. This architecture is composed of two modules: i. feature extraction from environment, ii navigation approach. The feature extraction module is based on the use of the edge detector by Canny algorithm and Hough transform for identification of lines from images of environment to monitoring. The lines identified correspond to cable conformation inside the duct. This information will serve to help the navigation system. For the implementation of the navigation system two approaches were proposed: navigation based on artificial neural network and navigation based on PID control. The navigation architecture can be used in real or simulated scenarios, and it was tested in a simulated environment.

Using Space D* for Crowded Real Robots Environments

This paper describes a new method of path planning for multiple robots in unknown environments and its validation in real robots context. The method, called Space D*, is based on two algorithms: the D*, which is an incremental graph search algorithm, and the Space Colonization algorithm, previously used to simulate crowd behaviours. The path planning is achieved through the exchange of information between the robots. So decentralized, each robot performs its path planning, which provide a obstacle-free path with the least number of robots around. The major contribution of the proposed method is that it generates paths in spacious environments facilitating the control of robots, and thus presenting itself in a viable way for using in areas populated with multiple robots. The results obtained validate the approach and show the advantages in comparison to use only D* method.