Rodrigo Calvo

Multiclass SVM Model Selection Using Particle Swarm Optimization

Tuning SVM hyperparameters is an important step for achieving good classification performance. In the binary case, the model selection issue is well studied. For multiclass problems, it is harder to choose appropriate values for the base binary models of a decomposition scheme. In this paper, the authors employ Particle Swarm Optimization to perform a multiclass model selection, which optimizes the hyperparameters considering both local and globalmodels. Experiments conducted over 4 benchmark problems show promising results.

Bio-inspired coordination of multiple robots systems and stigmergy mechanims to cooperative exploration and surveillance tasks

Cooperative and distributed strategy is considered for a team of mobile robots to explore and patrol environments. The coordination strategy is based on modified version of the artificial ant system. The covered area is marked with amount of pheromone. The kind of pheromone causes repulsion of robots. They are directed to unexplored regions and the regions that were not recently explored for accomplishment cooperative tasks as exploration and surveillance. Previously, application of the strategy confirmed that exploration and surveillance general behaviors emerge from the individual agent behavior. The strategy is able to adapt the current system dynamics if the number of robots or the environment structure or both change. Three approaches relied on variation of pheromone releasing are presented for comparison as well as two mechanisms for steering adjusting. As performance criterion for patrolling, it is considered the time gap between two visits of the same region. Experiments results demonstrate that different configurations of strategies affect exploration and surveillance behaviors. The results show the performance of proposed approaches.

Intelligent autonomous navigation for mobile robots: spatial concept acquisition and object discrimination

An autonomous system able to construct its own navigation strategy for mobile robots is proposed. The navigation strategy is molded from navigation experiences (succeeding as the robot navigates) according to a classical reinforcement learning procedure. The autonomous system is based on modular hierarchical neural networks. Initially, the navigation performance is poor (many collisions occur). Computer simulations show that after a period of learning, the autonomous system generates efficient obstacle avoidance and target seeking behaviors. Experiments also offer support for concluding that the autonomous system develops a variety of object discrimination capability and of spatial concepts.

Reinforcement learning for hierarchical and modular neural network in autonomous robot navigation

This work describes an autonomous navigation system based on a modular neural network. The environment is unknown and initially the system does not have ability to balance two innate behaviors: target seeking and obstacle avoidance. As the robot experiences some collisions, the system improves its navigation strategy and efficiently guides the robot to targets. A reinforcement learning mechanism adjusts parameters of the neural networks at target capture and collision moments. Simulation experiments show performance comparisons. Only the proposed system reaches targets if the environment presents a high risk (dangerous) configuration (targets are very close to obstacles).

A distributed, bio-inspired coordination strategy for multiple agent systems applied to surveillance tasks in unknown environments

Multiple agent systems are applied to exploration and surveillance tasks. A new distributed coordination strategy, designed according to a modified version of the artificial ant system, is described. The strategy is able to adapt the current system dynamics if the number of robots or the environment structure or both change. Experiment simulations are executed to evaluate two versions of the strategy considering different multiple robot systems and environment structures. Results confirm that exploration and surveillance general behaviors emerge from the individual agent behavior. Different compiled data sets are considered to assess the strategies, namely: needed time to conclude the task; and time between two consecutive sensory on a specific region. The results show that the strategy is effective and relatively efficient to execute the exploration and surveillance tasks.

Parametric investigation of a distributed strategy for multiple agents systems applied to cooperative tasks

Distributed coordination strategy based on modified version of the artificial ant system directs mobile robots to unexplored regions and regions that were not recently explored for accomplishing cooperative tasks as exploration and surveillance. Previously, application of the strategy confirmed that exploration and surveillance general behaviors emerge from the individual agent behavior. The strategy is able to adapt the current system dynamics if the number of robots or the environment structure or both change. In this paper, parametric variation of strategy is executed according to pheromone evaporation and releasing phenomena. Experiment results demonstrate that different configurations of phenomenon affect exploration and surveillance behaviors. Different compiled data sets are considered to assess the strategies, namely: needed time to conclude the task; and time between two consecutive sensory on a specific region. The results show that there is a set of configuration of the phenomena to become the strategy more efficient to execute the exploration and surveillance tasks.

Integration of virtual pheromones for mapping/exploration of environments by using multiple robots

The multiple robot coordination strategies have several advantages when compared to strategies based on a single robot, in terms of flexibility, gain of information and reduction of map building time. In this paper, a local pheromone map integration method is proposed based on the inter-robot observations, considering a method for the environment exploration named the Inverse Ant System-Based Surveillance System strategy (IAS-SS). Simulation results show that the map integration method is efficient, the trials are performed considering a variable number of robots in an indoor environment. Results obtained from several experiments confirm that the integration process is effective and suitable to execute the control of the access to pheromones in a virtual way.

A robot on-line area coverage approach based on the probabilistic Lloyd method

Area Coverage is a standard problem in which Robotics techniques can be applied. An approach to solve this problem is through techniques based on Centroidal Voronoi Tesselations (CVT), considering that each robot is a generator used to build Voronoi polygons. In this work, a new approach named by Sample Lloyd Area Coverage System (SLACS), is proposed that does not need of the explicit building of the diagram based in the Probabilistic Lloyd method to estimate a Voronoi polygons centroid. In addition, it is proposed a method to close Voronoi diagrams to apply in a classic Lloyd CVT procedure. Both approaches are compared in empty and roomlike environments done in simulated tests using both Player interface and Stage simulator. Results obtained show that the proposed approach is well suited to solve the area coverage problem via mobile sensor deployment and it is a simple and effective substitute to a Lloyd CVT method.

A Hierarchical Self-Organizing Controller for Navigation of Mobile Robots

In this work an autonomous navigation system based in a modular neuro-fuzzy network for controlling mobile robots is proposed. Based on this system the robot is able to reach goals avoiding collisions against obstacles in an unknown environment. The system architecture belongs to the reactive paradigm. A reinforcement learning mechanism balanced with two innate behaviors, which are to avoid obstacles and seek to goals, guides the robot from an initial point to the goal. The validation of the proposal system has been done by using the Saphira simulator. The results obtained in the tests performed on Saphira simulator and on the Pioneer robot show the efficiency and learning capabilities of this system.

Autonomous cognition and reinforcement learning for mobile robots

This paper describes a new class of autonomous intelligent systems for robot navigation application focusing on the synthesis, analysis and discussion of the learning process. Systems in this class are able to learn independently of supervision. In fact, they learn interacting with the environment while exploring it. A reinforcement learning strategy (inspired on the classical animal conditioning) and Hebb-like rule learning mechanisms support the knowledge acquistion process. The intelligent system must learn navigate the robot in an unknown environment, guiding it to targets according to a safe trajectory (without collisions). Their modular and hierarchical architecture is based on fuzzy systems and neural network techniques. The proposed approach has been validated by using a simulator and a mobile robot. In both cases, the experiments show that the autonomous intelligent system has a clear evidence of independent learning capability and exhibits a good performance during the navigation. Furthermore, this approach is compared with other system where there is no intelligent mechanisms to guide the robot.