No abstract

No abstract

Abstract:For navigation on outdoor surfaces, usually having different kind of roughness and soft irregularities, this paper proposal is that a wheeled robot combines the gradient method for path planning, alongside it adjusts velocity based on a multi-layer fuzzy neural network; the network integrates information about the roughness and the soft slopes of the terrain to compute the navigation velocity. The implementation is simple and computationally low-cost. The experimental tests show the advantage in the performance of the robot by varying the velocity depending on the terrain features.

This paper addresses the velocity control of wheeled vehicles regarding the terrain features, beyond detection and avoidance of the obstacles as most current works do. Terrain appearance average is used to enable the wheeled vehicle to adapt velocity such that, as speedy as possible, it safely navigates. The vehicle velocity adaptation imitates the human beings' driving behavior regarding the terrain features: humans use a quick and imprecise estimation of the terrain features but enough to drivenavigate without sliding or falling. A fuzzy neural network sets the vehicle velocity according to average estimations of terrain roughness. The terrain textures are modeled by the principal components that are enough to use pattern recognition for navigation purpose. One set of tests is executed using a small, wheeled robot, which adjusts velocity while navigating on surfaces such as ground, ground with grass, and stones paving. The other tests are done using images of roads of ground, concrete, asphalt, and loose stones, which are video filmed from a real car driven at less than 60 km/h of velocity; by applying the present approach, the required time/distance ratio to smoothly velocity change is granted.

In this paper is shown that the Appearance-Based modeling is the best pattern recognition method for supporting the velocity updating of wheeled-robots navigation. Although Appearance-Based recognition algorithms have lower accuracy than the ones for detailed pattern recognition, they successfully classify terrain textures by regarding the average of the appearance. Actually, the detailed recognition algorithms success in recognizing patterns depicted with lines, dots or borders, but they fail for recognizing patterns where the average appearance is required. As human driving experience shows, the assessment of the average appearance is needed for velocity updating during navigation on outdoor terrains. Human drivers make the velocity adjusting based on an estimation of the terrain average appearance. Hence, as the experimental result illustrate, the algorithms for average appearance recognition are the best option for training wheeled-robot for velocity updating while navigating over outdoor terrains.