Water bodies are challenging terrain hazards for terrestrial unmanned ground vehicles (UGVs) for several reasons. Traversing through deep water bodies could cause costly damage to the electronics of UGVs. Additionally, a UGV that is either broken down due to water damage or becomes stuck in a water body during an autonomous operation will require rescue, potentially drawing critical resources away from the primary operation and increasing the operation cost. Thus, robust water detection is a critical perception requirement for UGV autonomous navigation.
One of the properties useful for detecting still water bodies is that their surface acts as a horizontal mirror at high incidence angles. Still water bodies in wide-open areas can be detected by geometrically locating the exact pixels in the sky that are reflecting on candidate water pixels on the ground, predicting if ground pixels are water based on color similarity to the sky and local terrain features. But in cluttered areas where reflections of objects in the background dominate the appearance of the surface of still water bodies, detection based on sky reflections is of marginal value. Specifically, this software attempts to solve the problem of detecting still water bodies on cross-country terrain in cluttered areas at low cost.
Still water bodies are indirectly detected in cluttered areas of cross-country terrain by detecting reflections of objects in the water bodies using imagery acquired from a stereo pair of color cameras, which are mounted to the front of a terrestrial UGV. Object reflections can be from naturally occurring (e.g. vegetation, trees, hills, mountains, clouds) or man-made entities (e.g. signs, poles, vehicles, buildings, bridges). Color cameras provide a lower-cost solution than specialized imaging sensors (such as a polarization camera) and laser scanners. In addition, object reflections can be detected in water bodies with stereo vision at further ranges than with lidar scanners.
Four methods for detecting object reflections have been implemented: detection in the rectified camera images using cross correlation, detection in stereo range images, detection in a world map generated from range data, and detection using combined stereo range images and rectified camera images. Detection in stereo range images (see figure) exploits the knowledge that 3D coordinates of stereo range data on object reflections occur below the ground surface at a range close to that of the reflecting object.
Any autonomous robotic platform used on cross-country terrain that has restrictions on driving through water could benefit from this software, including military platforms and perhaps some agricultural platforms. The automotive industry could potentially benefit from an application of this technology to detect wet pavement.
This work was done by Arturo L. Rankin and Larry H. Matthies of Caltech for NASA’s Jet Propulsion Laboratory.
In accordance with Public Law 96-517, the contractor has elected to retain title to this invention. Inquiries concerning rights for its commercial use should be addressed to:
Innovative Technology Assets Management
JPL
Mail Stop 202-233
4800 Oak Grove Drive
Pasadena, CA 91109-8099
E-mail: This email address is being protected from spambots. You need JavaScript enabled to view it.
NPO-48494
This Brief includes a Technical Support Package (TSP).
Water Detection Based on Object Reflections
(reference NPO-48494) is currently available for download from the TSP library.
Don't have an account?
Overview
The document titled "Water Detection Based on Object Reflections" discusses innovative techniques developed by NASA's Jet Propulsion Laboratory (JPL) for detecting water bodies, particularly for unmanned ground vehicle (UGV) navigation in cross-country terrain. The primary focus is on leveraging the reflective properties of water surfaces, which behave like horizontal mirrors, to identify water bodies indirectly through the detection of reflections from surrounding objects, such as trees, buildings, and bridges.
The paper outlines a two-stage water detection process. In the first stage, the system analyzes range voids—areas where no range data is available. A key characteristic of water bodies is their level surface; thus, if the perimeter points of a range void are relatively horizontal, it can be flagged as a potential water region. This early warning system prompts the UGV to slow down as it approaches these areas, enhancing safety.
In the second stage, the system detects object reflections in a world map where range data penetrates the surface of the candidate water region. This detection is further refined using stereo range data and cross-correlation techniques, which can confirm the presence of water bodies even when initial stereo data is not available. The document highlights successful evaluations of these methods on a test course, where a significant number of true positive detections were achieved, and the false positive rate was notably reduced.
The evaluation involved processing over 12,000 stereo pairs of images, resulting in a 16.9% increase in true positive detections and a decrease in the false positive rate from 0.196% to 0.068%. The findings demonstrate the effectiveness of combining stereo and cross-correlation detection methods, extending the range of water detection significantly.
Overall, the document emphasizes the importance of robust water detection for UGV autonomous navigation, showcasing how cost-effective color cameras can be utilized for daytime water detection compared to more expensive sensors like lidar and thermal infrared. The research contributes to advancements in autonomous vehicle technology, enhancing their ability to navigate safely in diverse environments by accurately identifying potential hazards such as water bodies.
