Autonomous exploration using rapid perception of low-resolution image information

Abstract

We present a technique for mobile robot exploration in unknown indoor environments using only a single forward-facing camera. Rather than processing all the data, the method intermittently examines only small 32 × 24 downsampled grayscale images. We show that for the task of indoor exploration the visual information is highly redundant, allowing successful navigation even using only a small fraction of the available data. The method keeps the robot centered in the corridor by estimating two state parameters: the orientation within the corridor, and the distance to the end of the corridor. The orientation is determined by combining the results of five complementary measures, while the estimated distance to the end combines the results of three complementary measures. These measures, which are predominantly information-theoretic, are analyzed independently, and the combined system is tested in several unknown corridor buildings exhibiting a wide variety of appearances, showing the sufficiency of low-resolution visual information for mobile robot exploration. Because the algorithm discards such a large percentage of the pixels both spatially and temporally, processing occurs at an average of 1000 frames per second, thus freeing the processor for other concurrent tasks.

Extracting Minimalistic Corridor Geometry from Low-Resolution Images

“…11 shows normalized error in continuous frames with resolutions of 320 × 240 and 32 × 24. We also found that entropy decreases sharply when the robot reaches the end of the corridor [10]. Therefore, these could be used to detect the end of a corridor.…”

“…To detect this representation, we combine and extend two previous approaches. In previous work, Murali and Birchfield presented the use of ceiling lights for determining the center of the corridor using lowresolution techniques [10]. We extend this work by using other metrics like maximum entropy and maximum symmetry to estimate the center of the corridor when ceiling lights are not visible.…”

“…Empirically, entropy is maximum when the camera is pointing down the corridor [10]. The reason for this perhaps surprising result is that such an orientation causes scene surfaces from a variety of depths to be visible, yielding an increase of image information at this orientation.…”

“…Our approach is adapted from the work done by Murali and Birchfield [10], which uses the median of bright pixels (ceiling lights), maximum entropy, and maximum symmetry measures in the corridor image to determine its center and therefore the orientation. This approach has several advantages over existing techniques: It is simple, computationally efficient, and yields good results even for low-resolution images.…”

A Taxonomy of Vision Systems for Ground Mobile Robots

“…Another paper presents a combined machine learning and computer vision approach for robots to localize objects. It allows a humanoid robot to quickly learn to provide accurate 3D position estimates of objects seen [106]. In addition, the approach localizes objects robustly when utilized in a robot's workspace at arbitrary positions, even while the robot is moving its torso, head and eyes.…”