Skyline-based localisation for aggressively manoeuvring robots using UV sensors and spherical harmonics

Abstract

Place recognition is a key capability for navigating robots. While significant advances have been achieved on large, stable platforms such as robot cars, achieving robust performance on rapidly manoeuvring platforms in outdoor natural conditions remains a challenge, with few systems able to deal with both variable conditions and large tilt variations caused by rough terrain. Taking inspiration from biology, we propose a novel combination of sensory modality and image processing to obtain a significant improvement in the robustness of sequence-based image matching for place recognition. We use a UV-sensitive fisheye lens camera to segment sky from ground, providing illumination invariance, and encode the resulting binary images using spherical harmonics to enable rotationinvariant image matching. In combination, these methods also produce substantial pitch and roll invariance, as the spherical harmonics for the sky shape are minimally affected, providing the sky remains visible. We evaluate the performance of our method against a leading appearance-invariant technique (SeqSLAM) and a leading viewpoint-invariant technique (FAB-MAP 2.0) on three new outdoor datasets encompassing variable robot heading, tilt, and lighting conditions in both forested and urban environments. The system demonstrates improved condition-and tilt-invariance, enabling robust place recognition during aggressive zigzag manoeuvring along bumpy trails and at tilt angles of up to 60 degrees.

Spectral Skyline Separation: Extended Landmark Databases and Panoramic Imaging

“…Several navigational strategies have been unraveled experimentally, which explain the observed navigation abilities of insects, including pheromone trails, path integration and visual landmark navigation. Especially the latter not only receives attention in biology, but also in many technical fields like autonomous driving, localization and route following of driver-less cars and robots [5,6]. In this work, we focus on extracting the skyline from camera images as a basis for visual navigation tasks.…”

“…Local methods have been successfully used to extract the skyline using only the UV channel only on mobile robots [6,20]. However, in these studies, skyline extraction was mainly performed in suburban environments or in forests.…”

“…While local methods can easily be adapted by splitting the image into multiple segments and using an individual threshold for each, the adaption of global methods is not possible without violating the idea of a simple classification as it could be used by insects. Calculating thresholds for multiple segments also reduces the effect of lens flare to the process of skyline extraction, which may lead to errors during the process of skyline extraction [6]. Moreover, for omnidirectional images (e.g., using a fish eye lens), the skyline can be extracted completely and allows the use of existing navigational algorithms for panoramic images [26,27,28].…”

An inexpensive flying robot design for embodied robotics research

“…They can also hold heavy payloads and operate in many diverse environments. Ground robots have demonstrated steering, distance estimation, and obstacle avoidance [27][28]. They have also been used for olfactory-related tasks in order to detect, localize, and navigate toward odor cues [29].…”

“…They have also been used for olfactory-related tasks in order to detect, localize, and navigate toward odor cues [29]. Rugged, wheeled robots have been taken out of the lab and utilized outdoors to replicate the path integration and landmark steering of ants [28]. Hovercrafts have been primarily used to demonstrate the corridor centering response and other reactive control tasks in a hallway [27].…”