An improved method of processing information for autonomous navigation of a robotic vehicle across rough terrain involves the integration of terrain maps into a reactive navigation strategy. Somewhat more precisely, the method involves the incorporation, into navigation logic, of data equivalent to regional traversability maps. The terrain characteristic is mapped using a fuzzy-logic representation of the difficulty of traversing the terrain. The method is robust in that it integrates a global path-planning strategy with sensor-based regional and local navigation strategies to ensure a high probability of success in reaching a destination and avoiding obstacles along the way. The sensor-based strategies use cameras aboard the vehicle to observe the regional terrain, defined as the area of the terrain that covers the immediate vicinity near the vehicle to a specified distance a few meters away. The method at an earlier stage of development was described in “Navigating a Mobile Robot Across Terrain Using Fuzzy Logic” (NPO-21199), NASA Tech Briefs, Vol. 27, No. 2 (February 2003), page 5a. A recent update on the terrain classification stage of the method was reported in “Quantifying Traversability of Terrain for a Mobile Robot” (NPO-30744), NASA Tech Briefs, Vol. 29, No. 7 (July 2005), page 56. To recapitulate: The basic building blocks of the method are three behaviors that focus on successively smaller spatial scales and are integrated (in the sense of blended) through gains or weighting factors to generate speed and steering commands. The weighting factors are generated by fuzzy logic rules that take account of the current status of the vehicle.
At the present state of development, the three behaviors are denoted as a map-based seek-waypoint behavior, a sensor-based traverse-terrain behavior, and a sensor-based avoid-obstacle behavior (see figure). Navigation is initiated by a global pathplanning algorithm, which generates a sequence of waypoints that define an optimal path that passes through safe (that is, sufficiently traversable) regions of the terrain from a starting point to a destination. The waypoints are fed to the map-based seek-waypoint behavior, which, as its name suggests, seeks to direct the vehicle safely from the starting location to a waypoint. The sensor-based traverse-terrain behavior determines the safest regional segment to traverse on the basis of information from regional terrain images acquired by the cameras. The sensor- based avoid-obstacle behavior involves the use of local-obstacle information from the images to develop steering and speed commands to maneuver the vehicle around the obstacles.
This work was done by Ayanna Howard, Barry Werger, and Homayoun Seraji of Caltech for NASA’s Jet Propulsion Laboratory. For further information, access the Technical Support Package (TSP) free on-line at www.techbriefs.com/tsp under the Information Sciences category. NPO-30794
This Brief includes a Technical Support Package (TSP).
Integrating Terrain Maps Into a Reactive Navigation Strategy
(reference NPO-30794) is currently available for download from the TSP library.
Don't have an account?
Overview
The document discusses a comprehensive mobile robot navigation scheme designed for traversing hazardous terrains, particularly in the context of planetary exploration. It highlights the importance of integrating terrain characteristics into path planning methodologies to ensure the safety and effectiveness of robotic missions.
The introduction emphasizes the challenges faced by mobile robots when navigating high-risk terrains. To address these challenges, the authors propose a global path planning methodology that incorporates intrinsic terrain properties directly into the navigation logic. This approach aims to minimize risks and enhance mission success by analyzing terrain features before traversal.
A key component of the proposed methodology is the development of a traversability map, which segments the terrain into regions based on their ease of traversal. This multi-valued map representation allows for the assessment of global terrain characteristics and the computation of a traversal cost function, which is crucial for ensuring robot survivability. The traversal cost is then utilized by a global path planner to identify an optimally safe path through the terrain.
The document also reviews traditional path planning methods, which often rely on grid representations and goal-attainment strategies while neglecting intrinsic terrain properties. The authors reference various existing approaches, such as sensor-based motion planning and wavefront propagation algorithms, that have attempted to address these limitations. However, the proposed method distinguishes itself by employing a robust perceptual representation of terrain quality known as the Fuzzy Traversability Index, which categorizes traversability into four fuzzy sets: POOR, LOW, MODERATE, and HIGH.
The methodology is further detailed in subsequent sections, which discuss the software package used for implementation and present experimental results from tests conducted with a Pioneer 2-AT rover. These results demonstrate the effectiveness of the terrain-based path planning algorithm in real-world scenarios.
In conclusion, the document presents a significant advancement in mobile robot navigation by integrating terrain analysis into path planning. This innovative approach not only enhances the ability of robots to navigate challenging environments but also contributes to the broader field of robotics and autonomous systems, with potential applications in various scientific and commercial domains.
