Evaluation of Field of View Width in Stereo-vision-Based Visual Homing

SummaryVisual homing is a local navigation technique used to direct a robot to a previously seen location by comparing the image of the original location with the current visual image. Prior work has shown that exploiting depth cues such as image scale or stereo-depth in homing leads to improved homing performance. While it is not unusual to use a panoramic field of view (FOV) camera in visual homing, it is unusual to have a panoramic FOV stereo-camera. So, while the availability of stereo-depth information may improve performance, the concomitant-restricted FOV may be a detriment to performance, unless specialized stereo hardware is used. In this paper, we present an investigation of the effect on homing performance of varying the FOV widths in a stereo-vision-based visual homing algorithm using a common stereo-camera. We have collected six stereo-vision homing databases – three indoor and three outdoor. Based on over 350,000 homing trials, we show that while a larger FOV yields performance improvements for larger homing offset angles, the relative improvement falls off with increasing FOVs, and in fact decreases for the widest FOV tested. We conduct additional experiments to identify the cause of this fall-off in performance, which we term the ‘blinder’ effect, and which we predict should affect other correspondence-based visual homing algorithms.

Three Landmark Optimization Strategies for Mobile Robot Visual Homing

Visual homing is an attractive autonomous mobile robot navigation technique, which only uses vision sensors to guide the robot to the specified target location. Landmark is the only input form of the visual homing approaches, which is usually represented by scale-invariant features. However, the landmark distribution has a great impact on the homing performance of the robot, as irregularly distributed landmarks will significantly reduce the navigation precision. In this paper, we propose three strategies to solve this problem. We use scale-invariant feature transform (SIFT) features as natural landmarks, and the proposed strategies can optimize the landmark distribution without over-eliminating landmarks or increasing calculation amount. Experiments on both panoramic image databases and a real mobile robot have verified the effectiveness and feasibility of the proposed strategies.

Suboptimal mid-course guidance algorithm for accelerating missiles

This paper deals with the closed-loop form of mid-course guidance law design for accelerating missile system, whose acceleration is approximately constant. A midcourse guidance algorithm of feedback form is proposed to satisfy the engagement geometry conditions at the burn-out time for terminal homing performance enhancement. The effect of velocity change due to missile acceleration is explicitly considered in the derivation of the guidance law. The terminal constraint update algorithm is proposed under the assumption that the target trajectory is predicted precisely. Simulation results are provided to show the performance and characteristics of the proposed algorithm.

Amino acid and carbohydrate tradeoffs by honey bee nectar foragers and their implications for plant-pollinator interactions

Honey bees are important pollinators, requiring floral pollen and nectar for nutrition. Nectar is rich in sugars, but contains additional nutrients, including amino acids (AAs). We tested the preferences of free-flying foragers between 20 AAs at 0.1% w/w in sucrose solutions in an artificial meadow. We found consistent preferences amongst AAs, with essential AAs preferred over nonessential AAs. The preference of foragers correlated negatively with AA induced deviations in pH values, as compared to the control. Next, we quantified tradeoffs between attractive and deterrent AAs at the expense of carbohydrates in nectar. Bees were attracted by phenylalanine, willing to give up 84 units sucrose for 1 unit AA. They were deterred by glycine, and adding 100 or more units of sucrose could resolve to offset 1 unit AA. In addition, we tested physiological effects of AA nutrition on forager homing performance. In a no-choice context, caged bees showed indifference to 0.1% proline, leucine, glycine or phenylanaline in sucrose solutions. Furthermore, flight tests gave no indication that AA nutrition affected flight capacity directly. In contrast, low carbohydrate nutrition reduced the performance of bees, with important methodological implications for homing studies that evaluate the effect of substances that may affect imbibition of sugar solution. In conclusion, low AA concentrations in nectar relative to pollen suggest a limited role in bee nutrition. Most of the 20 AAs evoked a neutral to a mild deterrent response in bees, thus it seems unlikely that bees respond to AAs in nectar as a cue to assess nutritional quality. Nonetheless, free choice behavior of foraging bees is influenced, for instance by phenylalanine and glycine. Thus, AAs in nectar may affect plant-pollinator interactions and thereby exhibit a selective pressure on the flora in the honey bee habitat.

Analyzing the effect of landmark vectors in homing navigation

The development of an autonomous navigating robot is a challenging task. Motivated by the performance of insects successfully returning to the nest, researchers have studied bio-inspired navigation algorithms for their potential use in mobile robots. In this paper, we analyze landmark-based approaches, especially Distance Estimated Landmark Vector (DELV), Average Correctional Vector and Average Landmark Vector methods, that use landmark vectors for visible environmental landmarks. We evaluated the homing performance of various landmark vector methods with surrounding landmarks under occlusion and found that the occluded or missing landmarks have a significant influence on the performance. We also developed a landmark vector algorithm with a visual compass that uses only retinal images without a reference compass. From our experimental results, we conclude that the DELV shows robust homing navigation performance with missing or occluded landmarks among landmark vector methods.