SEARCH · Engineering Papers
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
Reliable detection of non-traversable hazards is a key requirement for off-road autonomous navigation. A detailed description of each obstacle detection algorithm and their performance on the surveyed obstacle course is presented in this paper.
This paper describes machine vision algorithms that enable precision guidance and hazard avoidance during small body exploration through onboard visual feature tracking and landmark recognition.
In January, 2004, the Mars Exploration Rover (MER) mission landed two rovers, Spirit and Opportunity, on the surface of Mars. Several autonomous navigation capabilities were employed in space for the first time in this mission. ]n the Entry, Descent, and Landing (EDL) phase, both landers used a vision system called the, Descent Image Motion Estimation System (DIMES) to estimate horizontal velocity during the last 2000 meters (m) of descent, by tracking features on the ground with a downlooking camera, in order to control retro-rocket firing to reduce horizontal velocity before impact. During surface operations, the rovers navigate autonomously using stereo vision for local terrain mapping and a local, reactive planning algorithm called Grid-based Estimation of Surface Traversability Applied to Local Terrain (GESTALT) for obstacle avoidance. ]n areas of high slip, stereo vision-based visual odometry has been used to estimate rover motion, As of mid-June, Spirit had traversed 3405 m, of which 1253 m were done autonomously; Opportunity had traversed 1264 m, of which 224 m were autonomous. These results have contributed substantially to the success of the mission and paved the way for increased levels of autonomy in future missions.
This paper discusses the Team Raptor passive perception system hardware and software design, implementation, and performance, and describes a road map to faster and improved passive perception.
We show that in conjunction with SAD correlation, our new method improves stereo quality at range discontinuities while maintaining real-time performance.
Successful off-road autonomous navigation by an unmanned ground vehicle (UGV) requires reliable perception and representation of natural terrain. While perception algorithms are used to detect driving hazards, terrain mapping algorithms are used to represent the detected hazards in a world model a UGV can use to plan safe paths. There are two primary ways to detect driving hazards with perception sensors mounted to a UGV: binary obstacle detection and traversability cost analysis. Binary obstacle detectors label terrain as either traversable or non-traversable, whereas, traversability cost analysis assigns a cost to driving over a discrete patch of terrain. In uncluttered environments where the non-obstacle terrain is equally traversable, binary obstacle detection is sufficient. However, in cluttered environments, some form of traversability cost analysis is necessary. The Jet Propulsion Laboratory (JPL) has explored both approaches using stereo vision systems. A set of binary detectors has been implemented that detect positive obstacles, negative obstacles, tree trunks, tree lines, excessive slope, low overhangs, and water bodies. A compact terrain map is built from each frame of stereo images. The mapping algorithm labels cells that contain obstacles as no-go regions, and encodes terrain elevation, terrain classification, terrain roughness, traversability cost, and a confidence value. The single frame maps are merged into a world map where temporal filtering is applied. In previous papers, we have described our perception algorithms that perform binary obstacle detection. In this paper, we summarize the terrain mapping capabilities that JPL has implemented during several UGV programs over the last decade and discuss some challenges to building terrain maps with stereo range data.
Explore the source record for details and available documents.
One goal of the "Demo III" unmanned ground vehicle program is to enable autonomous nighttime navigation at speeds of up to 10 m.p.h. To perform obstacle detection at night with stereo vision will require night vision cameras that produce adequate image quality for the driving speeds, vehicle dynamics, obstacle sizes, and scene conditions that will be encountered. This paper analyzes the suitability of four classes of night vision cameras (3-5 micrometer cooled FLIR, 8-12 micrometer cooled FLIR, 8-12 micrometer uncooled FLIR, and image intensifiers) for night stereo vision, using criteria based on stereo matching quality, image signal to noise ratio, motion blur and synchronization capability. We find that only cooled FLIRs will enable stereo vision performance that meets the goals of the Demo III program for nighttime autonomous mobility.
One goal of the Demo III unmanned ground vehicle program is to enable autonomous nighttime navigation at speeds of up to 10 m.p.h.
The performance of autonomous mobile robots performing complex navigation tasks can be dramatically improved by directing expensive sensing and planning in service of the task. The task-direction algorithms can be quite simple. In this paper we describe a simple task-directed vision system which has been implemented on a real outdoor robot which navigates using stereo vision. While the performance of this particular robot was improved by task-directed vision, the performance of task-directed vision in general is influenced in complex ways by many factors. We briefly discuss some of these, and present some initial simulated results.
The paper examines the use of stereo vision (SV) for obstacle detection in semiautonomous land navigation. Feature-based and field-based paradigms for SV are reviewed. The paper presents stochastic models and simple, efficient stereo matching algorithms for the field-based approach and describes a near-real-time vision system using these algorithms. Experimental results illustrate aspects of the stochastic models and lead to the first semiautonomous traversals of natural terrain to use SV for obstacle detection.
Progress has been made in the development of a family of robotic land vehicles having modular and minimalist design features chosen to impart a combination of robustness, reliability, and versatility. These vehicles at earlier stages of development were described in two previous NASA Tech Briefs articles: "Reconfigurable Exploratory Robotic Vehicles" (NPO-20944), Vol. 25, No. 7 (July 2001), page 56; and "More About Reconfigurable Exploratory Robotic Vehicles" (NPO-30890), Vol. 33, No. 8 (August 2009), page 40. Conceived for use in exploration of the surfaces of Mars and other remote planets, these vehicles could also be adapted to terrestrial applications, including exploration of volcanic craters or other hostile terrain, military reconnaissance, inspection of hazardous sites, and searching for victims of earthquakes, landslides, avalanches, or mining accidents. In addition, simplified versions of these vehicles might be marketable as toys. The most basic module in this family of reconfigurable robots is the Axel rover, which has a cylindrical body with two main wheels and a trailing link. Inside its body are three motors and associated mechanisms for driving the two wheels and for rotating the link 360 around its symmetrical body. The actuated link serves several purposes: It is used as a lever arm to react to the wheels thrust to move Axel in multiple directions. It is used to rotate the Axel housing in order to tilt, to the desired angle, any sensors and instruments mounted on or in the Axel housing. It provides an alternative mobility mode, which is primarily used in its tethered configuration. Turn ing the link into the ground in lieu of driving the wheels causes the Axel housing and wheels to roll as a unit and thereby leads to a tumbling motion along the ground. With a tether mounted around Axel s cylindrical body, the link serves as a winch mechanism to reel and unreel the tether raising and lowering Axel over steep and vertical surfaces (Figure 1). Sensors, computation, and communication modules are also housed inside Axel s body. A pair of stereo vision cameras provides three-dimensional view for autonomous navigation and avoiding obstacles. Inertial sensors determine the tilt of the robot and are used for estimating its motion. In a fully developed version, power would be supplied by rechargeable batteries aboard Axel; at the time of reporting the information for this article, power was supplied from an external source via a cable. In and of itself, the Axel rover is fully capable of traversing and sampling terrains on planetary surfaces. By use of only the two main wheel actuators and the caster link actuator, Axel can be made to follow an arbitrary path, turn in place, and operate upside- down or right-side-up. If operated in a tethered configuration, as shown in Figure 1, it can be made to move down and up a steep crater wall, descend from an overhang to a cave, and ascend from the cave back to the overhang, all by use of the same three actuators. Such tethered operation could be useful in searching for accident victims or missing persons in mines, caves, and rubble piles. Running the tether through the caster link enhances the stability of Axel and provides a restoring force that keeps the link off the ground for the most part during operation on a steep slope. In its extended configuration, two Axel modules can dock to either side of a payload module to form the four wheeled Axel2 rover (Figure 2). Additional payload and Axel modules can dock to either side of the Axel2 to form the Axel3 rover, extending its payload capacity and its mobility capabilities.
Development of unmanned vehicles for local navigation in terrains unstructured by humans is reviewed. Modes of navigation include teleoperation or remote control, computer assisted remote driving (CARD), and semiautonomous navigation (SAN). A first implementation of a CARD system was successfully tested using the Robotic Technology Test Vehicle developed by Jet Propulsion Laboratory. Stereo pictures were transmitted to a remotely located human operator, who performed the sensing, perception, and planning functions of navigation. A computer provided range and angle measurements and the path plan was transmitted to the vehicle which autonomously executed the path. This implementation is to be enhanced by providing passive stereo vision and a reflex control system for autonomously stopping the vehicle if blocked by an obstacle. SAN achievements include implementation of a navigation testbed on a six wheel, three-body articulated rover vehicle, development of SAN algorithms and code, integration of SAN software onto the vehicle, and a successful feasibility demonstration that represents a step forward towards the technology required for long-range exploration of the lunar or Martian surface. The vehicle includes a passive stereo vision system with real-time area-based stereo image correlation, a terrain matcher, a path planner, and a path execution planner.
For real-time stereo vision systems, the standard method for estimating sub-pixel stereo disparity given an initial integer disparity map involves fitting parabolas to a matching cost function aggregated over rectangular windows. This results in a phenomenon known as 'pixel-locking,' which produces artificially-peaked histograms of sub-pixel disparity. These peaks correspond to the introduction of erroneous ripples or waves in the 3D reconstruction of truly Rat surfaces. Since stereo vision is a common input modality for autonomous vehicles, these inaccuracies can pose a problem for safe, reliable navigation. This paper proposes a new method for sub-pixel stereo disparity estimation, based on ideas from Lucas-Kanade tracking and optical flow, which substantially reduces the pixel-locking effect. In addition, it has the ability to correct much larger initial disparity errors than previous approaches and is more general as it applies not only to the ground plane.
The Demo III program has as its primary focus the development of autonomous mobility for a small rugged cross country vehicle. In this paper we report recent progress on both stereo-based obstacle detection and terrain cover color-based classification.
In order to achieve the ambitious objectives of the Mars 2020 (M2020) mission, in particular the ability to autonomously traverse more challenging terrains more efficiently, new surface mobility software was developed for Enhanced Navigation (ENav). That decision was made early in the project, before most of the new surface flight software (FSW) existed, which created a need for a separate framework where the new navigation algorithms could be quickly prototyped and tested, before more realistic FSW-based testbeds became available. The JPL robotics team chose the Robot Operating System [1] (ROS) as the environment in which to test the new ENav algorithms. This made it possible to write the algorithms in the C language required by the FSW, so they could be directly ported over to the flight module later on, while leveraging all the C++ libraries and tools provided by ROS for simulation and testing. The ENav algorithms were developed as a separate C library, and stubs were used to replace any FSW-specific code, such as Event Reporting (EVRs) and data products (DPs). A ROS simulator was developed to generate a rich set of varied 3D terrains representative of the candidate Mars landing sites and simulate the physics of the rover motion, the point cloud perceived by the rover’s stereo vision system, and the new thinking-while-driving (TWD) navigation logic which directs the rover to drive autonomously to user-specified waypoints. To simulate the rover motion and perception, a ROS node was developed that uses a software library called HyperDrive Sim (HDSim), which is a wrapper for the Rover Sequencing and Visualization Program [2] (RSVP). That library provides roverterrain settling, realistic slip modelling, and camera rendering capability based on the rover’s NavCam machine vision models. To simulate the navigation logic, a ROS node was created that initializes and runs the ENav algorithms in a way that mimics the FSW execution, while also providing the capability to load and replay data products, including re-running the recorded inputs through the ENav algorithms for testing. An engineering Graphical User Interface (GUI) was also developed to visualize various elements, such as the rover pose during the drive, the simulated and perceived terrain, the selected local and global paths to the goal, the evaluated candidate paths and the reasons why they were rejected, the keep-in and keep-out zones (KIOZs), etc. Finally, an advanced Monte Carlo (MC) framework that can run many simulations in parallel on the Cloud and automatically generate reports that capture the key ENav performance metrics was developed to evaluate the system in a statisticallymeaningful way. This paper provides an overview of the ROSbased simulator used for testing the M2020 ENav algorithms.
The technology development requirements for various Mars rover range capabilities are discussed, focusing on local navigation of the rover. The capabilities of two methods are compared. In one method, operators on the earth view stereo pictures sent by the rover and determine short traverse paths which the rover follows. The other method achieves more autonomous capability by using computer vision from orbital imagery with approximate long routes commanded from earth. The locomotion, navigation, ground operations, computation, power, thermal control, communications, sample acquisition, and analysis and preservation requirements are examined.
Autonomous Exploration, Inc., has developed an advanced Bayesian statistical inference method that leverages current computing technology to produce a highly accurate surface navigation system. The method combines dense stereo vision and high-speed optical flow to implement visual odometry (VO) to track faster rover movements. The Bayesian VO technique improves performance by using all image information rather than corner features only. The method determines what can be learned from each image pixel and weighs the information accordingly. This capability improves performance in shadowed areas that yield only low-contrast images. The error characteristics of the visual processing are complementary to those of a low-cost inertial measurement unit (IMU), so the combination of the two capabilities provides highly accurate navigation. The method increases NASA mission productivity by enabling faster rover speed and accuracy. On Earth, the technology will permit operation of robots and autonomous vehicles in areas where the Global Positioning System (GPS) is degraded or unavailable.