Grid-Based Estimation of Surface Traversability Applied to Local Terrain (GESTALT) is a software library that helps a robotic wheeled vehicle choose a safe path across nearby terrain. The input to GESTALT consists mainly of three-dimensional range measurements of the nearby terrain taken by sensors (e.g., stereoscopic cameras or a laser scanner) aboard the robot as it moves along. GESTALT groups the data into grid cells, then geometrically evaluates the cells as separate planar patches, identifying such impediments to traversability as steps, steep slopes, and excessive roughness. Both straight and curved paths from present position of the robot can be evaluated. The safest path that helps the rover move toward its goal is selected. The resulting path evaluations can be used to steer the robot.
GESTALT is an extension of a previously developed algorithm known as Morphin, incorporating additional capabilities: Diagnostic outputs in greater detail are available. Traversability can be evaluated in a direction-specific manner; for example, a hill found to be unclimbable if approached head-on might be found climbable on a sideways approach. Navigation parameters can be reset at any time. Multiple rover configurations can be evaluated in a single map.
This program was written by Mark Maimone of Caltech for NASA's Jet Propulsion Laboratory.
This software is available for commercial licensing. Please contact Don Hart of the California Institute of Technology at (818) 393-3425. Refer to NPO-21233.
This Brief includes a Technical Support Package (TSP).
Software for Analyzing Traversability of Terrain
(reference NPO21233) is currently available for download from the TSP library.
Don't have an account?
Overview
The document presents GESTALT, a software library developed by Mark W. Maimone at NASA's Jet Propulsion Laboratory, designed to assist robotic wheeled vehicles in navigating unknown terrains safely and efficiently. GESTALT processes three-dimensional range measurements collected by the robot's sensors, such as stereoscopic cameras and laser scanners, as the robot moves through its environment. The software organizes this data into a grid of rectangular cells, akin to a checkerboard overlay on the terrain, allowing for detailed analysis of the landscape.
The primary function of GESTALT is to evaluate the traversability of the terrain by identifying potential hazards, including step hazards, slope hazards, and rough terrain. Each grid cell is assessed as a separate planar patch, enabling the software to detect obstacles that may impede the robot's movement. GESTALT can evaluate both straight and curved paths from the robot's current position, ultimately selecting the safest route that leads toward the robot's goal. This path evaluation is crucial for steering and driving the robot effectively.
GESTALT builds upon a previous algorithm known as Morphin, enhancing its capabilities in several ways. It allows for direction-specific traversability evaluations, recognizing that certain obstacles may be approached successfully from different angles. For instance, a hill that is unclimbable when approached head-on might be traversable from the side. The software also offers more detailed diagnostic outputs, the ability to reset navigation parameters at any time, and the capacity to evaluate multiple rover configurations within a single map.
The document emphasizes that GESTALT does not require prior knowledge of the terrain, as it relies solely on real-time sensory data, which may be somewhat noisy. This adaptability makes it suitable for various environments, including extraterrestrial landscapes like Mars.
In summary, GESTALT is a sophisticated tool for robotic navigation, enhancing the safety and efficiency of rovers as they traverse challenging terrains. The software is available for commercial licensing, and interested parties are encouraged to contact Don Hart at the California Institute of Technology for more information. The work was conducted under NASA's sponsorship, highlighting its significance in advancing robotic exploration.
