Andres Huertas

Computer Vision on Mars

Increasing the level of spacecraft autonomy is essential for broadening the reach of solar system exploration. Computer vision has and will continue to play an important role in increasing autonomy of both spacecraft and Earth-based robotic vehicles. This article addresses progress on computer vision for planetary rovers and landers and has four main parts. First, we review major milestones in the development of computer vision for robotic vehicles over the last four decades. Since research on applications for Earth and space has often been closely intertwined, the review includes elements of both. Second, we summarize the design and performance of computer vision algorithms used on Mars in the NASA/JPL Mars Exploration Rover (MER) mission, which was a major step forward in the use of computer vision in space. These algorithms did stereo vision and visual odometry for rover navigation and feature tracking for horizontal velocity estimation for the landers. Third, we summarize ongoing research to improve vision systems for planetary rovers, which includes various aspects of noise reduction, FPGA implementation, and vision-based slip perception. Finally, we briefly survey other opportunities for computer vision to impact rovers, landers, and orbiters in future solar system exploration missions.

Analysis of On-Board Hazard Detection and Avoidance for Safe Lunar Landing

Landing hazard detection and avoidance technology is being pursued within NASA to improve landing safety and increase access to sites of interest on the lunar surface. The performance of a hazard detection and avoidance system depends on properties of the terrain, sensor performance, algorithm design, vehicle characteristics and the overall all guidance navigation and control architecture. This paper analyzes the size of the region that must be imaged, sensor performance parameters and the impact of trajectory angle on hazard detection performance. The analysis shows that vehicle hazard tolerance is the driving parameter for hazard detection system design.

Daytime Water Detection by Fusing Multiple Cues for Autonomous Off-Road Navigation

Abstract : Detecting water hazards is a significant challenge to unmanned ground vehicle autonomous off-road navigation. This paper focuses on detecting the presence of water during the daytime using color cameras. A multi-cue approach is taken. Evidence of the presence of water is generated from color, texture, and the detection of reflections in stereo range data. A rule base for fusing water cues was developed by evaluating detection results from an extensive archive of data collection imagery containing water. This software has been implemented into a run-time passive perception subsystem and tested thus far under Linux on a Pentium based processor.

Passive imaging based multi-cue hazard detection for spacecraft safe landing

Accurate assessment of potentially damaging ground hazards during the spacecraft EDL (entry, descent, and landing) phase is crucial to insure a high probability of safe landing. A lander that encounters a large rock, falls off a cliff, or tips over on a steep slope can sustain mission-ending damage. Guided entry is expected to shrink landing ellipses from 100-300 km to ~10 km radius for the second-generation landers as early as 2009. Regardless of size and location, however, landing ellipses will almost always contain hazards such as craters, discontinuities, steep slopes, and large rocks. It is estimated that an MSL (Mars Science Laboratory)-sized lander should detect and avoid 16-150m diameter craters, vertical drops similar to the edges of 16m or 3.75m diameter crater, for high and low altitude HDA (Hazard Detection and Avoidance) respectively. It should also be able to detect slopes 20deg or steeper, and rocks 0.75m or taller. In this paper we will present a passive imaging based, multi-cue hazard detection and avoidance (HDA) system suitable for Martian and other lander missions. This is the first passively imaged HDA system that seamlessly integrates multiple algorithms - crater detection, slope estimation, rock detection and texture analysis, and multi-cues

Stereo-Based Tree Traversability Analysis for Autonomous Off-Road Navigation

crater morphology, rock distribution, to detect these hazards in real time

Stereo vision and shadow analysis for landing hazard detection

Autonomous off-road navigation through forested areas is particularly challenging when there exists a mixture of densely distributed thin and thick trees. To make progress through a dense forest, the robot must decide which trees it can push over and which trees it must circumvent. This paper describes a stereo-based tree traversability algorithm implemented and tested on a robotic vehicle under the DARPA PerceptOR program. Edge detection is applied to the left view of the stereo pair to extract long and vertical edge contours. A search step matches anti-parallel line pairs that correspond to the boundaries of individual trees. Stereo ranging is performed and the range data within trunk fragments are averaged. The diameters of each tree is then estimated, based on the average range to the tree, the focal length of the camera, and the distance in pixels between matched contour lines. We use the estimated tree diameters to construct a tree traversability image used in generating a terrain map. In stationary experiments, the average error in estimating the diameter of thirty mature tree trunks (having diameters ranging from 10-65 cm and a distance from the cameras ranging from 2.5-30 meters) was less than 5 cm. Tree traversability results from the daytime for short baseline (9 cm) and wide baseline (30 cm) stereo are presented. Results from nighttime using wide baseline (33.5 cm) thermal infrared stereo are also presented.

Attenuating stereo pixel-locking via affine window adaptation

Unmanned planetary landers to date have landed blind, without the benefit of onboard landing hazard detection and avoidance systems. This constrains landing sites to very benign terrain and limits the scientific goals of missions. We review sensor options for landing hazard detection, then identify an approach based on stereo vision and shadow analysis that appears to address the broadest set of missions with the lowest cost. We describe algorithms for slope estimation and rock detection with this approach, develop models of their performance, and validate those models experimentally. Instantiating our model of rock detection reliability for Mars predicts that this approach would reduce the probability of failed landing by at least a factor of 4 compared to blind landing. Conversely, for the safety level desired for the 2009 Mars lander, this approach would increase the fraction of the planet that is accessible for landing from about 1/3 to nearly 100%.

Landing Hazard Detection with Stereo Vision and Shadow Analysis

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 flat 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. We demonstrate the method on synthetic imagery as well as real stereo data from an autonomous outdoor vehicle

Performance evaluation of hazard detection and avoidance algorithms for safe Lunar landings

Under a Phase A Concept Study for the NASA New Millennium Program (NMP) ST9 mission, the Terrain Relative Guidance System (TRGS) project developed an approach to landing hazard detection with descent imagery for missions to most solid-surfaced bodies of the solar system. Unmanned planetary landers to date have landed “blind”; that is, without the benefit of onboard landing hazard detection and avoidance systems. This constrains landing site selection to very benign terrain, which in turn constrains the scientific agenda of missions. In this paper we review sensor options for landing hazard detection and identify an approach based on stereo vision and shadow analysis that addresses the broadest set of missions. We then develop performance models for slope estimation and rock detection with this approach and validate those models experimentally. Instantiating our model of rock detection reliability for Mars predicts that this approach will reduce the probability of failed landing by at least a factor of 4 in any given terrain. Conversely, for the safety level desired for the 2009 Mars lander, this approach would roughly triple the fraction of the planet that is accessible for landing. The key contributions of this work are identifying the most appropriate sensor approach, developing and validating the performance models, and quantifying the impact this could have on missions.

Flight Testing a Real-Time Hazard Detection System for Safe Lunar Landing on the Rocket-Powered Morpheus Vehicle

Unmanned planetary landers to date have landed “blind”; that is, without the benefit of onboard landing hazard detection and avoidance systems. This constrains landing site selection to very benign terrain, which in turn constrains the scientific agenda of missions. Systems for automatic surface reconstruction and for hazard detection, mapping, and assessment are becoming mature. Before they can be put to practical use, it is essential to be able to characterize their performance for the purposes of scientific evaluation and their utility to engineers planning and designing landed missions. It is also important to be able to predict performance for a variety of scenarios. The evaluation metrics need to be simple enough to be readily comprehensible but still to capture the important relevant performance parameters. In this paper we describe the process, metrics, results, and algorithm improvement recommendations from the evaluation of the performance of the hazard detection and avoidance (HDA) algorithms developed in the Autonomous Landing and Hazard Avoidance Technology (ALHAT) Project by means of Monte Carlo simulation of thousands of Lunar landings.1 2