Intelligent Autonomous Navigation for Mobile Robots: Spatial Concept Acquisition and Object Discrimination

Abstract

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. Index Terms -intelligent autonomous navigation, neural networks, reinforcement learning, mobile robots, biologically inspired models.

Modular Neural Network and Classical Reinforcement Learning for Autonomous Robot Navigation: Inhibiting Undesirable Behaviors

“…If this is not the case, the design is more complex. In general, it is observed that systems generate undesirable behaviors, e.g., cyclic trajectories, if attractive objects are eclipsed [7]. This is the case considered in this paper: a reactive autonomous navigation system for mobile robots without a target sensory field.…”

A Hierarchical Self-Organizing Controller for Navigation of Mobile Robots

“…In [6], innovate neuron models provide the navigation system the capacity for learning spatial concepts. Target seeking and collision avoidance behaviors are incrementally associated with specific features of objects (e.g.…”

Event Detection and Localization in Mobile Robot Navigation Using Reservoir Computing

“…Each object has a particular color, denoting its respective class. Obstacles are considered repulsive objects while targets are attractive objects [2]. The robot model is shown in Fig.…”

“…The robot interacts with the environment by distance, color and 3ŷ fish (t) refers to the desired output after applying the fisher labeling given by (4) • to +90 • ). Each position holds three sensors (for distance, color and contact perception) [2]. In this work, the robot model has 17 sensor positions, differing from [3].…”

“…Autonomous robot navigation systems have been extensively developed in the literature [1][2][3][4]. Early navigation strategies are either deliberative (generation of robot trajectories based on path planning) or reactive (robot control based on a direct mapping of sensory input to actions).…”