Robots are often employed in hazardous or inaccessible environments, such as disaster sites, extraterrestrial terrains, agricultural fields, and ocean floors. Autonomous operation is crucial in these scenarios to reduce reliance on human operators and enable real-time decision-making. However, robots must balance multiple, often conflicting, objectives. These objectives are subject to change based on new data or evolving conditions. This paper presents a novel approach to dynamic multi-objective trajectory planning. The proposed method leverages the boundary intersection decomposition technique to adaptively plan trajectories that balance multiple evolving objectives. Our approach ensures efficient and effective exploration by continuously optimizing the trade-offs between changing objectives. We show that our method performs on average 34 % better in terms of solution quality on the dynamic multi-objective trajectory planning problem as compared to prior work.

We develop a coverage approach for heterogeneous agents that leverages the different sensing and motion capabilities of a team. Coverage performance is measured using ergodicity, which when optimized balances exploitation versus exploration, where areas of interest are indicated with an information metric. Prior work uses spectral decomposition of a spatial map of information to guide a set of heterogeneous agents, each with different sensor and motion models, to optimize coverage. This work leverages wavelet transforms to decompose the information map rather than the Fourier transform typically applied to ergodic search and demonstrates the importance of selecting a suitable wavelet family to use, based on the information map being explored. Further a sequence of wavelets is used for decomposition to overcome dependency on selecting one suitable wavelet family. Our experimental results show that using wavelet families well-suited to the specific information map for information map decomposition leads to, on average, 43% improvement over a baseline method in terms of a standard coverage metric (ergodicity), while using a wellsequenced set of wavelets for decomposition leads to a 65% improvement in coverage performance across multiple types of information maps.

Coral reefs are an important ecosystem to the local communities and indigenous wildlife that rely on them. However, reefs have greatly degraded in recent decades with the remaining at increasing risk of loss. Quantitatively mapping these reefs would provide a resource for us to monitor changes and understand their health. We explore methods leveraging limited spectral data and resources for efficient global scale modeling of coral reefs. We then evaluate performance on a Deep Neural Network and our previously developed Deep Conditional Dirichlet Model. Regions of high uncertainty based on the model output prediction are used to determine informative in situ sampling. An ergodic planner is implemented to generate a path through these regions to acquire samples that best improve the coral map. The result is a resource efficient learning based pipeline that augments existing spectral data and maps coral reefs globally to improve our understanding of their condition.

Robotic explorers are essential tools for gathering information about regions that are inaccessible to humans. For applications like planetary exploration or search and rescue, robots use prior knowledge about the area to guide their search. Ergodic search methods find trajectories that effectively balance exploring unknown regions and exploiting prior information. In many search based problems, the robot must take into account multiple factors such as scientific information gain, risk, and energy, and update its belief about these dynamic objectives as they evolve over time. However, existing ergodic search methods either consider multiple static objectives or consider a single dynamic objective, but not multiple dynamic objectives. We address this gap in existing methods by presenting an algorithm called Dynamic Multi-Objective Ergodic Search (D-MO-ES) that efficiently plans an ergodic trajectory on multiple changing objectives. Our experiments show that our method requires up to nine times less compute time than a naïve approach with comparable coverage of each objective.

This work introduces range-based GP maps, which directly represent terrain by modeling the range from a LiDAR sensor as a Gaussian process (GP) in spherical space. Such a model aligns the predicted uncertainty from the GP regression with the uncertainty in the underlying sensor observations. Experimental evaluation on simulated natural terrain indicates that local range-based GP maps perform comparably to elevation-based methods when predicting terrain height, with the former producing more stable parameters and providing a better uncertainty representation. An aggregation method is proposed using the pose as an additional input to the GP. Unlike their elevation-based counterparts, range-based GP maps are capable of modeling overhangs and vertical obstacles with ease, demonstrated with examples of maps built on real-world data from a fully 3D subterranean environment.

Spectral mapping, typically performed at the orbital scale, is hard at the rover scale as it necessitates larger coverage and field operable spectrometers. In this work, we propose using RGB cameras to assist spectral mapping. RGB cameras placed on a wide range of robotic platforms, including aerial vehicles, which can explore large regions compared to ground rovers. Our method uses a spectral model which learns the spatial relationship of spectra in a compressed feature space. We show that RGB data can contribute to this feature space and thereby enhance spectral reconstruction accuracy.

Perception using stereo requires external light. In the absence of natural light, active, structured light provides light where it is needed. In this work, we demonstrate how a free moving line striping laser can be used to perceive and model terrains. In this formulation, we do not need to know the position of the laser with respect to the stereo pair which precludes the need for calibrating the laser position. This also saves an actuator as the laser can be passively articulated using a spring. In this paper, we present the algorithms to efficiently extract the laser pixels from the stereo image pair. We also present techniques to use the structure of the light and overcome degenerate cases to build cleaner maps. This perception method benefits micro-rovers specifically those trying to operate in the extreme lighting conditions.

Earth and planetary sciences often rely upon the detailed examination of spectroscopic data for rock and mineral identification. This typically requires the collection of high resolution spectroscopic measurements. However, they tend to be scarce, as compared to low resolution remote spectra. This work addresses the problem of inferring high-resolution mineral spectroscopic measurements from low resolution observations using probability models. We present the Deep Gaussian Conditional Model, a neural network that performs probabilistic super resolution via maximum likelihood estimation. It also provides insight into learned correlations between measurements and spectroscopic features, allowing for the tractability and interpretability that scientists often require for mineral identification. Experiments using remote spectroscopic data demonstrate that our method compares favorably to other analogous probabilistic methods. Finally, we show and discuss how our method provides human-interpretable results, making it a compelling analysis tool for scientists.

Maintaining high levels of productivity for planetary rover missions is very difficult due to limited communication and heavy reliance on ground control. There is a need for autonomy that enables more adaptive and efficient actions based on real-time information. This paper presents an autonomous mapping and exploration approach for planetary rovers. We first describe a machine learning model that actively combines remote and rover measurements for mapping. We focus on spectroscopic data because they are commonly used to investigate surface composition. We then incorporate notions from information theory and non-myopic path planning to improve exploration productivity. Finally, we demonstrate the feasibility and successful performance of our approach via spectroscopic investigations of Cuprite, Nevada; a well-studied region of mineralogical and geological interest. We first perform a detailed analysis in simulations, and then validate those results with an actual rover in the field in Nevada.

In this paper, we present an active SLAM framework for volumetric exploration of 3D underwater environments with multibeam sonar. Recent work in integrated SLAM and planning performs localization while maintaining volumetric free-space information. However, an absence of informative loop closures can lead to imperfect maps, and therefore unsafe behavior. To solve this, we propose a navigation policy that reduces vehicle pose uncertainty by balancing between volumetric exploration and revisitation. To identify locations to revisit, we build a 3D visual dictionary from real-world sonar data and compute a metric of submap saliency. Revisit actions are chosen based on propagated pose uncertainty and sensor information gain. Loop closures are integrated as constraints in our pose-graph SLAM formulation and these deform the global occupancy grid map. We evaluate our performance in simulation and real-world experiments, and highlight the advantages over an uncertainty-agnostic framework.

This work extensively studies and analyses several unsupervised clustering methods for hyperspectral data. We look at unsupervised classification solutions that accomplish adaptive cluster formation in anticipation for new data discoveries. We provide qualitative and quantitative answers to significant problems like high-dimensionality of hyperspectral datasets, multiple sources and relative amounts of existing noise in data and low class separability. The effectiveness of various clustering techniques is illustrated on diverse hyperspectral datasets by intensive experimentation, comparison between techniques and analysis.

Remote sensing measurements can provide crucial information about the material properties of a planetary surface but their application is limited by their spatial resolution, typically tens of meters per pixel, when constituent materials are mixed at much finer scale. Consequently the orbital observations must be validated with in situ measurements from a spectrometer on the ground. In planetary exploration this means that a rover must visit selected locations that jointly improve a model of the environment and satisfy mobility and sampling constraints. Conventional planning methods used in this situation follow sub-optimal greedy strategies that are not scalable to large areas. We show how the problem can be effectively defined in a Markov Decision Process framework and propose a planning algorithm based on Monte Carlo Tree Search, which is efficient but devoid of these drawbacks thereby providing superior performance. We evaluate our approach using hyperspectral imagery of a well-studied geologic site in Cuprite, Nevada.

Plant phenotyping is the measurement of observable plant traits. Current methods for phenotyping in the field are labour intensive and error prone. High throughput plant phenotyping in an automated and noninvasive manner is crucial to accelerating plant breeding methods. Occlusions and non-ideal sensing conditions is a major problem for high throughput plant phenotyping with most state-of-the-art 3D phenotyping algorithms relying heavily on heuristics or hand-tuned parameters. To address this problem, we present a novel model-based optimization approach for estimating plant physical traits from plant units called phytomers. The proposed approach involves sampling parameterized 3D plant models from an underlying probability distribution. It then optimizes, making the mass of this probability distribution approach true parameters of the model. Reformulating the phenotyping objective as a search in the space of plant models lets us reason about the plant structure in a holistic manner without having to rely on hand-tuned parameters. This makes our approach robust to noise and occlusions as frequently encountered in real world environments. We evaluate our approach for plant units taken across simulated, greenhouse and field environments. This work furthers field-based robotic phenotyping capabilities paving the way for plant biologists to study the coupled effect of genetics and environment on improving crop yields.

Planetary exploration involves frequent scientific reformulation and replanning. It is limited by communication constraints and to overcome this limitation, this paper formulates the process as a collaboration in which the human scientist and the robot work together to fill in gaps in knowledge to make discoveries. It introduces the science hypothesis map as the probabilistic structure in which scientists initially describe their abstract beliefs and hypotheses, and in which the state of this belief evolves as the robot makes raw measurements. It discusses how to incorporate path planning for maximizing scientific information gain, which is efficiently computed. As proof of concept, this paper describes a geologic exploration problem where a robot uses a spectrometer to infer the geologic composition of different regions in a mining district at Cuprite, Nevada. It shows that the science hypothesis map can infer geologic units with high accuracy, and that exploration using information gain-based path planning has better performance than exploration with conventional science-blind algorithms.

Efficient exploration of unknown terrains by extraterrestrial rovers requires the development of strategies that reduce the entropy in the geological classification of a given terrain. Without such intelligent strategies, teleoperation of the rover is reliant either on human intuition or on the exhaustive exploration of the entire terrain. This paper highlights the use of low-resolution reconnaissance using satellite imagery to generate plans for rovers that reduce the overall uncertainty in the various geological classes. This becomes pivotal when exploration to collect diverse samples is resource constrained through exploration budgets and transmission bandwidths. We put forward two major contributions - a science-aware planner that uses information gain and a novel method of estimating this information gain. We propose an exploration strategy, based on the Multi-Heuristic A*, to solve the trade-off between optimizing path lengths and geological exploration through Pareto-optimal solutions. We show that our algorithm, which explicitly uses projected entropy-reduction in planning, significantly outperforms science-agnostic approaches and other science-aware strategies like greedy best-first searches. We further propose a feature-space based entropy formulation in contrast to the frequently used differential entropy formulation and show superior results when reconstructing the unsampled data from the set of sampled points.

Automatically correlating plant observable characteristics to their underlying genetics will streamline selection methods in plant breeding. Measurement of plant observable characteristics is called phenotyping, and knowing plant phenotypes accurately and throughout a plant's growth is central to making breeding decisions. In-field plant phenotyping in an automated and noninvasive manner is hence crucial to accelerating plant breeding methods. However, most of the existing methods on plant phenotyping using visual imaging are confined to controlled greenhouse environments. This paper presents an automated method of mapping 2D images collected in an outdoor sorghum field to segmented 3D plant units that are of interest for phenotyping. This method leverages multiple horizontal and vertical viewpoints while capturing 2D images from a robotic platform so as to generate in-field 3D reconstructions of the sorghum plant. We develop and quantitatively evaluate segmentation methods on these 3D reconstructions and also compare against reconstructions obtained from a controlled greenhouse environment. We present analysis that contrasts the role of purely local geometric features and the effect of addition of global context in both datasets. This work furthers capabilities of in-field phenotyping which paves the way forward for plant biologists to study the coupled effect of genetics and environment on improving crop yields.

We extract 3D curb from video sequence, using a single camera equipped with fish-eye lens and located at the front/rear of the vehicle. The challenge in extracting curbs from images lies in their small size and their lack of texture. We show that by appropriately exploiting appearance features, 3D geometry, and temporal information, one can reliably detect and localize the curbs in the 3D scene. The main underlying assumption of our model is that the road surface is flat and that the curb is approximately orthogonal to the road plane. We collected nine videos with ground truth, under day-time sunny weather condition, up to 2m range. Our experimental results compare favorably wrt the current the state-of-the-art on our database -90% precision rate in average and over 85% accuracy in curb height estimation.

Adaptive exploration uses active learning principles to improve the efficiency of autonomous robotic surveys. This work considers an important and understudied aspect of autonomous exploration: in situ validation of remote sensing measurements. We focus on high- dimensional sensor data with a specific case study of spectroscopic mapping. A field robot refines an orbital image by measuring the surface at many wavelengths. We introduce a new objective function based on spectral unmixing that seeks pure spectral signatures to accurately model diluted remote signals. This objective reflects physical properties of the multi-wavelength data. The rover visits locations that jointly improve its model of the environment while satisfying time and energy constraints. We simulate exploration using alternative planning approaches, and show proof of concept results with the canonical spectroscopic map of a mining district in Cuprite, Nevada.

This paper presents a method that applies connected component analysis to plan routes that keep robots continuously illuminated and on traversable slopes while reaching one or more goal locations. Such routes promise to extend the lifespan, range, and scientific return of solar-powered robots exploring environments with changing but predictable lighting conditions, particularly those of the Moon and Mercury. Maps of lighting and ground slope that describe these constraints in position and time are computed, and all distinct interconnected regions that have both direct sunlight and safe slope are found using connected component analysis. These three-dimensional connected components are pruned of roots that violate time constraints and branches that dead-end in discontinuous routes. Each component is the basis for a graph that includes all feasible routes from the initial time to the final time of that component. The shortest feasible route between a pair of start and goal positions within the same component is found using A* search and is characterized by its total length and average speed. Malapert Peak and Shackleton Crater, both near the Moon's South Pole, serve as examples throughout this paper due to their highly-relevant, dynamic, and predictable lighting caused by the Moon's motion relative to the Sun.

In order to be deployed in real-world driving environments, self-driving cars must be able to recognize and respond to exceptional road conditions, such as highway workzones, because such unusual events can alter previously known traffic rules and road geometry. In this paper, we present a set of computer vision methods that recognize, through identification of workzone signs, the bounds of a highway workzone and temporary changes in highway driving environments. Through testing using video data about highway workzones recorded under various weather conditions, our approach was able to perfectly identify the boundaries of workzones and robustly detect a majority of driving condition changes. In addition to these tests, we evaluated, using a mock workzone setup, the usefulness of our workzone recognition systems' outputs for safe-guarding a self-driving car.

This paper presents an algorithm based on ecological models of foraging and that uses uncertainty in scientific observations made by the robot to value future actions. It is the hypothesis of this work that the foraging strategy will be an improvement over strategies based on principles from the design of experiments literature for small budget sizes. The experiment in this paper shows that for small budget sizes the new algorithm performs at least as well as other methods. However the new algorithm does not exhaust its sampling budget by default and thus needs to be modified to outperform other approaches for large sample budgets, which provides opportunity for future work in this area.

We address the problem of adaptive information-optimal data collection in time series. Here a remote sensor or explorer agent throttles its sampling rate in order to track anomalous events while obeying constraints on time and power. This problem is challenging because the agent has limited visibility - all collected datapoints lie in the past, but its resource allocation decisions require predicting far into the future. Our solution is to continually fit a Gaussian process model to the latest data and optimize the sampling plan on line to maximize information gain. We compare the performance characteristics of stationary and nonstationary Gaussian process models. We also describe an application based on geologic analysis during planetary rover exploration. Here adaptive sampling can improve coverage of localized anomalies and potentially benefit mission science yield of long autonomous traverses.

Communication blackouts and latency are significant bottlenecks for planetary surface exploration; rovers cannot typically communicate during long traverses, so human operators cannot respond to unanticipated science targets discovered along the route. Targeted data collection by point spectrometers or high-resolution imagery requires precise aim, so it typically happens under human supervision during the start of each command cycle, directed at known targets in the local field of view. Spacecraft can overcome this limitation using onboard science data analysis to perform autonomous instrument targeting. Two critical target selection capabilities are the ability to target priority features of a known geologic class, and the ability to target anomalous surfaces that are unlike anything seen before. This work addresses both challenges using probabilistic surface classification in traverse images. We first describe a method for targeting known classes in the presence of high measurement cost that is typical for power- and time-constrained rover operations. We demonstrate a Bayesian approach that abstains from uncertain classifications to significantly improve the precision of geologic surface classifications. Our results show a significant increase in classification performance, including a seven-fold decrease in misclassification rate for our random forest classifier. We then take advantage of these classifications and learned scene context in order to train a semi-supervised novelty detector. Operators can train the novelty detection to ignore known content from previous scenes, a critical requirement for multi-day rover operations. By making use of prior scene knowledge we find nearly double the number of abnormal features detected over comparable algorithms. We evaluate both of these techniques on a set of images acquired during field expeditions in the Mojave Desert.

To be deployed in the real-world, automatic and semi-automatic systems should understand traffic rules by recognizing and comprehending contents of traffic signs, because traffic signs inform what driving behaviors should be. In this paper, we present the successful application of methods to improve the traffic sign localization performance. Given a potential sign region, our algorithm represents both the detected sign as a target and candidates in the subsequent frame as probability density functions. Then, our algorithm maximizes the similarity between a target and candidates to localize the sign. Finally, the maximum similarity among candidates is assigned as a tracked sign. The experimental results verify that our algorithm can robustly localize traffic signs in images under various weather conditions and driving scenarios.

Cartographic databases can be kept up to date through aerial image analysis. Such analysis is optimized when one knows what parts of an aerial image are roads and when one knows locations of complex road structures, such as overpasses and intersections. This paper proposes self-supervised computer vision algorithms that analyze a publicly available cartographic resource (i.e., screenshots of road vectors) to, without human intervention, identify road image-regions and detects overpasses.

This paper presents new aerial image analysis algorithms that, from highway ortho-images, produce lane-level detailed maps. We analyze screenshots of road vectors to obtain the relevant spatial and photometric cues of road image-regions. We then refine the obtained patterns to generate hypotheses about the true road-lanes. A road-lane hypothesis, since it explains only a part of the true road-lane, is then linked to other hypotheses to completely delineate boundaries of the true road-lanes. Finally, some of the refined image cues about the underlying road network are used to guide a linking process of road-lane hypotheses.

Grousers, sometimes called lugs, are recognized as a way to improve wheel performance and traction, but there have been, to date, no comprehensive guidelines for choosing grouser patterns. This work presents a quantitative expression for determining appropriate grouser spacing for rigid wheels. Past empirical studies have shown that increasing grouser height and number can improve performance, to a point. The newly proposed grouser spacing equation is based on observations that wheels with an inadequate number of grousers induce forward soil flow ahead of the wheel, and thus rolling resistance. The equation relates geometric wheel parameters (wheel radius, grouser height and spacing) and operating parameters (slip and sinkage), and predicts a maximum allowable grouser spacing (or, equivalently, a minimum number of grousers). Experiments with various grouser heights and numbers demonstrate good correspondence to the proposed equation, as increases in number of grousers beyond the predicted minimum number stop improving performance. A grouser spacing equation is particularly useful for designing efficient wheels. The proposed relation includes slip and sinkage, parameters that cannot be assumed constant or known a priori, but this work shows that wheels designed using the proposed equation are robust to changing operating scenarios even if they degrade beyond estimated nominal conditions.

In order to be deployed in real-world driving environments, autonomous vehicles must be able to recognize and respond to exceptional road conditions, such as highway workzones, because such unusual events can alter previously known traffic rules and road geometry. In this paper, we present a set of computer vision methods which recognize the bounds of a highway workzone and temporary changes in highway driving environments through recognition of workzone signs. Our approach filters out irrelevant image regions, localizes potential sign image regions using a learned color model, and recognizes signs through classification. Performance of individual unit tests is promising; still, it is unrealistic to expect perfect performance in sign recognition. Performance errors with individual modules in sign recognition will cause our system to misread temporary highway changes. To handle potential recognition errors, our method utilizes the temporal redundancy of sign occurrences and their corresponding classification decisions. Through testing, using video data recorded under various weather conditions, our approach was able to perfectly identify the boundaries of workzones and robustly detect a majority of driving condition changes.

This paper describes and evaluates a 3-D kinematic controller for passively-steered rovers. Passively-steered rovers have no steering motors, but rely on differential wheel velocities to change the axle steer angles. This passive steering design is reliable and efficient but more challenging to control than powered steering designs, especially when driving on rough terrain. A controller based on 2-D kinematics fails to accurately maintain the desired trajectory when traversing obstacles. The presented 3-D kinematic controller uses inertial and proprioceptive sensing to modify commanded steer angles and wheel velocities, greatly improving steering accuracy. Validation in simulation and physical experiments is presented. These results are significant because they establish the viability of the passive-steering configuration for precise navigation.

An optical flow odometry method for mobile robots using a single downward-looking camera is presented. The method is robust to the robot's own moving shadow and other sources of error. Robustness derives from two techniques: prevention of feature selection on or near shadow edges and elimination of outliers based on inconsistent motion. In tests where the robot's shadow dominated the image, prevention of feature selection near shadow edges allowed accurate velocity estimation when outlier rejection alone failed. Performance was evaluated on two robot platforms and on multiple terrain types at speeds up to 2 m/s.

Computational aesthetics is the study of applying machine learning techniques to identify aesthetically pleasing imagery. Prior work used online datasets scraped from large user communities like Flikr to get labeled data. However, online imagery represents results late in the media generation process, as the photographer has already framed the shot and then picked the best results to upload. Thus, this technique can only identify quality imagery once it has been taken. In contrast, automatically creating pleasing imagery requires understanding the imagery present earlier in the process. This paper applies computational aesthetics techniques to a novel dataset from earlier in that process in order to understand how the problem changes when an autonomous agent, like a robot or a real-time camera aid, creates pleasing imagery instead of simply identifying it.

Robotic explorers must be capable of autonomous navigation into unknown terrain as well as autonomous science to interpret their observations to guide exploration. In this research we have created a robot able to select science features, direct instruments, collect observations, build maps, and interpret this information to plan actions. We report on field experiments in California's Amboy Crater lava field and demonstrate fundamental capabilities for adaptive exploration in geologic mapping tasks. We show feature detection and instrument visual servoing that enables automated science observation of dozens of targets. Gaussian process models are used to discover spatial and cross-sensor structure including correlations between different locations and sensing scales. The rover develops these relationships on the fly with only on-board computation, reinterpreting remote sensing data in light of the surface materials it observes. The rover learns spatial models of physical phenomena and guides its exploration into informative areas using Maximum Entropy Sampling to improve exploration efficiency. The Amboy Crater experiments show that science autonomy can play a useful role in facilitating geologic survey on kilometer scales.

An autonomous robotic vehicle can drive through and park in a lot more reliably if it is guided by a parking lot map. This specialized map creates structure, including the centerlines of drivable regions and intersection locations in an often unstructured and unmarked environment and enables the vehicle to focus its attention on regions that require detailed analysis. Existing methods of building such maps require manu- ally driving vehicles for collecting sensor measurements. Instead of pursuing a labor-intensive approach, we analyze an aerial image of a parking lot to build a topological map. In particular, our algorithm produces a lane-graph of a parking lot's drivable regions by executing several image processing steps. First it estimates drivable regions by superimposing detection of a parking spots onto the parking lot boundary segmentation. Second, a distance transform is applied to drivable regions to reveal its skeleton. Lastly our algorithm searches the distance map to identify a set of the peak points and connects them to generate a lane-graph that concisely represents drivable regions. Experiments show promising results of real-world parking lot aerial-imagery analysis.

In this paper we present algorithms for automatically generating a road network description from aerial imagery. The road network inforamtion (RNI) produced by our algorithm includes a composite topoloigical and spatial representation of the roads visible in an aerial image. We generate this data for use by autonomous vehicles operating on-road in urban environments. This information is used by the vehicles to both route plan and determine appropriate tactical behaviors. RNI can provide important contextual cues that influence driving behaviors, such as the curvature of the road ahead, the location of traffic signals, or pedestrian dense areas. The value of RNI was demonstrated compellingly in the DARPA Urban Challenge, where the vehicles relied on this information to drive quickly, safely and efficiently.

Registering multiple sets of 3D range data is a crucial capability for robots. The standard method for matching two sets of range data is to convert the ranges to a point cloud representation, and then use on of the many variants of iterative closest point (ICP). We present a set of alternative methods for matching 3D range scans based on a different data representation: evidence grid maps. Evidence grids are robust to noise and variations in point density, can incorporate an indefinite number of ranges, and explicitly encode empty as well as occupied space. While 3D evidence grids can be huge when naively implemented, we use an optimized octree data structure to efficiently store sparse volumetric maps. To register a series of range scans, we build an evidence grid map for each scan, and then register them together using a several different methods. The first two methods are based on a 3D extension of the classic 2D Lucas-Kanade template matching method, and differ only in whether we match a single large region, or multiple small regions that are selected heuristically. Our third method involves extracting surfaces from the evidence grids, and then running ICP to register the surfaces. We demonstrate our methods and compare them to ICP using two datasets collected by two different subterranean robots.

In this paper, we present a user study of a new collaborative communication method between a user and remotely-located robot performing an exploration task. In the studied scenario, our user possesses scientific expertise but not necessarily detailed knowledge of the robot's capabilities, resulting in very little common ground

Selective data return leverages onboard data analysis to allocate limited bandwidth resources during remote exploration. Here we present an adaptive method to subsample image sequences for downlink. We treat selective data return as a compression problem in which the explorer agent transmits the subset of measurements that are most informative with respect to the complete dataset. Experiments demonstrate selective downlink of navigation imagery by a rover during autonomous geologic investigations in the Atacama desert of Chile. Here automatic analysis identifies informative images using classifications based on natural image statistics. Image texture analysis, together with a context-sensitive Hidden Markov Model representation, permits adaptive downlink in response to geologic unit boundaries. Selective data return improves the science content of returned data for this geologic mapping task.

This paper presents locomotion control strategies for a novel, multi-legged hopping robot named ldquorobotic all-terrain surveyorrdquo (RATS). This conceptual robot has a spherical body roughly the size of a soccer ball, with 12 legs equally distributed over its surface. The legs are linear pneumatic actuators (1-DOF), oriented such that their axes of motion are normal to the surface of the body. While the 12-legged robot is still in design, we have experimented with a planar 5-legged prototype to study the control problem in a simpler form. Our control solutions overcame the constraint that the legs are at a fixed orientation with respect to the body by inducing the body to roll. This approach allows the legs to be positioned sequentially at a desired angle with respect to the ground surface, exploiting the symmetric configuration of the system. Successful gaits for running and jumping over obstacles based on the rolling concept are presented in this work. Physical experiments with the 5-legged device validate the control approaches and demonstrate the performance of the system and its potential for the future.

Robotic exploration is an excellent method for obtaining information about sites too dangerous for people to explore. The operator's understanding of the environment depends on the rover returning useful information. Robotic mission bandwidth is frequently constrained, limiting the amount of information the rover can return. This paper explores the tradeoff between information and bandwidth based on two years of observations during a robotic astrobiology field study. The developed theory begins by analyzing the search task conducted by robot operators. This analysis leads to an information optimization model. Important parameters in the model include the value associated with detecting a target, the probability of locating a target, and the bandwidth required to collect the information from the environment. Optimizing the information return between regions creates an image and provides the necessary information while reducing bandwidth. Application of the model to the analyzed field study results in an optimized image that requires 48.3% less bandwidth to collect. The model also predicts several data collection patterns that could serve as the basis of data collection templates for improving mission effectiveness. The developed optimization model reduces the bandwidth necessary to collect information, thus aiding missions in collecting more information from the environment.

Geologists and planetary scientists will benefit from methods for accurate segmentation of rocks in natural scenes. However, rocks are poorly suited for current visual segmentation techniques - they exhibit diverse morphologies and have no uniform property to distinguish them from background soil. We address this challenge with a novel detection and segmentation method incorporating features from multiple scales. These features include local attributes such as texture, object attributes such as shading and two-dimensional shape, and scene attributes such as the direction of illumination. Our method uses a superpixel segmentation followed by region-merging to search for the most probable groups of superpixels. A learned model of rock appearances identifies whole rocks by scoring candidate superpixel groupings. We evaluate our method's performance on representative images from the Mars Exploration Rover catalog.

An artificial satellite independent localization system for a lunar rover using the stars for navigation is described. The system uses a wide field-of-view star tracker and wide range high precision inclinometers to determine location on the Moon using quaternion output from the star tracker along with pitch and roll data from the inclinometer.

During two weeks of study in September and October of 2004, a science team directed a rover and explored the arid Atacama Desert in Chile. The objective of the mission was to search for life. Over the course of the mission the team gained experience with the rover and the rover became more reliable and autonomous. As a result, the rover/operator system became more effective. Several factors likely contributed to the improvement in science effectiveness including increased experience, more effective search strategies, different science team composition, different science site locations, changes in rover operational capabilities, and changes in the operation interface. However, it is difficult to quantify this effectiveness because science is a largely creative and unstructured task. This study considers techniques that quantify science team performance leading to an understanding of which features of the human-rover system are most effective and which features need further development. Continuous observation of the scientists throughout the mission led to coded transcripts enumerating each scientific statement. This study considers whether six variables correlate with scientific effectiveness. Several of these variables are metrics and ratios related to the daily rover plan, the time spent programming the rover, the number of scientific statements made and the data returned. The results indicate that the scientists created more complex rover plans without increasing the time to create the plans. The total number of scientific statements was approximately equal (2187 versus 2415) for each week. There was a 50% reduction in bytes of returned data between the two weeks resulting in an increase in scientific statements per byte of returned data ratio. Of the original six, the most successful proxies for science effectiveness were the time to program each rover task and the number of scientific statements related to data delivered by the rover. Although both these measures have face validity and were consistent with the results of this experiment, their ultimate empirical utility must be measured further.

We report a study of a human-robot system composed of a science team (located in Pittsburgh), an engineering team (located in Chile), and a robot (located in Chile). We performed ethnographic observations simultaneously at both sites over two weeks as scientists collected data using the robot. Our data reveal problems in establishing and maintaining common ground between the science team and the robot due to missing contextual information about the robot. Our results have implications for the design of systems to support human-robot interaction.

This paper describes the application of a RaoBlackwellized Particle Filter to the problem of simultaneous localization and mapping onboard a hovering autonomous underwater vehicle. This vehicle, called DEPTHX, equipped with a large array of pencil-beam sonars for mapping, and autonomously explore a system of flooded tunnels associated with the Zacaton sinkhole in Tamaulipas, Mexico. Due to the three-dimensional nature of the tunnels, we describe an extension of traditional two dimensional evidence grids to three dimensions. In May 2005, we collected a sonar data set in Zacaton. We present successful SLAM results using both the real-world data and simulated data

When searching a remote environment with a robot, the fundamental constraint on the operator is the bandwidth available to the mission. This paper examines how a science team uses a high bandwidth panorama during a two-year astrobiology field test. Bandwidth directly controls the type and amount of information received by the operator. Given the significance of information and bandwidth to mission success, it is important for human robot interaction analyses to consider bandwidth usage during robotic operations. Insight gained from these analyses can help future missions efficiently use their bandwidth to collect the necessary information from the environment. The analysis first shows that the science team preferentially views certain areas of the panorama based on tile elevation but not based on azimuth. This finding led the analysis to look into what tasks the science team completes while using the panorama. In the context of an astrobiology mission, the most important role of the panorama is to determine the robot's position on orbital images. The secondary task is to determine the general geologic context of the environment. Based on the viewing patterns and associated tasks the analysis produces a list of targets important to mission success. Future research efforts will focus on methods for collecting the information contained in these targets from the environment at a reduced bandwidth cost.

The Atacama Desert of northern Chile may be the most lifeless place on Earth, yet where the desert meets the Pacific coastal range desiccation-tolerant micro-organisms are known to exist. The gradient of biodiversity and habitats in the Atacama’s subregions remain unexplored and are the focus of the Life in the Atacama project. To conduct this investigation, long traverses must be made across the desert with instruments for geologic and biologic measurements. In this paper we motivate the Life in the Atacama project from both astrobiologic and robotic perspectives. We focus on some of the research challenges we are facing to enable endurance navigation, resource cognizance, and long-term survivability. We conducted our first scientific investigation and technical experiments in Chile with the mobile robot Hyperion. We describe the experiments and the results of our analysis. These results give us insight into the design of an effective robotic astrobiologist and into the methods by which we will conduct scientific investigation in the next field season.

Mobile robots and robot teams can leverage multiple views of a scene to improve the accuracy of their maps. However non-uniform noise persists even when each sensor's pose is known, and the uncertain correspondence between detections from different views complicates easy "multiple view object detection." We present an algorithm based on expectation/maximization (EM) clustering that permits a principled fusion of the views without requiring an explicit correspondence search. We demonstrate the use of this algorithm to improve mapping performance of robots in simulation and in the field.

Sun-synchronous exploration is accomplished by reasoning about sunlight: where the Sun is in the sky, where and when shadows will fall, and how much power can be obtained through various courses of action. In July 2001 a solar-powered rover, named Hyperion, completed two sun-synchronous exploration experiments in the Canadian high arctic (75/spl deg/N). Using knowledge of orbital mechanics, local terrain, and expected power consumption, Hyperion planned a sun-synchronous route to visit designated sites while obtaining the necessary solar power for continuous 24-hour operation. Hyperion executed its plan and returned to its starting location with batteries fully charged after traveling more than 6 kilometers in barren, Mars-analog terrain. We describe the concept of sun-synchronous exploration. We overview the design of the robot Hyperion and the software system that enables it to operate sun-synchronously. We then discuss results from analysis of our first experiment in sun-synchronous exploration and conclude with observations.

In June and July 1997, Nomad, a planetary-relevant mobile robot, traversed more than 220 kilometers across the barren Atacama Desert in Chile, exploring a landscape analogous to the surfaces of the Moon and Mars. In this unprecedented demonstration, Nomad operated both autonomously and under the control of operators thousands of kilometers away, addressing issues of robot configuration, communication, position estimation, and navigation in rugged, natural terrain. The field experiment also served to test technologies for remote geological investigation, paving the way for new exploration strategies on Earth and beyond. Finally, by combining safeguarded teleoperation with panoramic visualization and a novel user interface, the Atacama Desert Trek provided the general public a compelling interactive experience an opportunity to remotely drive an exploratory robot. Nomad's performance in the Atacama Desert Trek set new benchmark in high performance robotics operations relevant to terrestrial and planetary exploration. This paper presents an overview of the experiment, describes technologies key to Nomad's success, and discusses outcomes and implications.

A terrestrial geologist investigates an area by systematically moving among and inspecting surface features, such as outcrops, boulders, contacts and faults. A planetary geologist must explore remotely and use a robot to approach and image surface features. To date, position-based control has been developed to accomplish this task. This method requires an accurate estimate of the feature position, and frequent update of the robot's position. In practice this is error prone, since it relies on interpolation and continuous integration of data from inertial or odometric sensors or other position determination techniques. The development of vision-based control of robot manipulators suggests an alternative approach for mobile robots. We have developed a vision-based control system that enables our Marsokhod mobile robot to drive autonomously to within sampling distance of a visually designated natural feature. This system utilizes a robust correlation technique based on matching the sign of the difference of the Gaussian of images. We will describe our system and our initial results using it during a field experiment in the Painted Desert of Arizona.

We have designed an architecture that allows a gait planner to effectively control primitive walking behaviors. The behaviors ensure safe and efficient walking, even without planning; the addition of planning improves performance in rough terrain by allowing the robot to anticipate changes in its gait. We have implemented our approach and demonstrated it on a real robot, Dante II, and on a simulated hexapod with more complex kinematics. With the hexapod we can produce a variety of gaits and stably switch among them. We have attained performance improvement by using narrowly focused planning to guide behavior.

The Dante project is developing walking robots to explore inside volcanic craters. These robots face many challenges including generating a walking gait in rough, obstacle-filled terrain. For the walking robot Dante II, we implemented a gait controller to address this situation. Our approach is embodied in a network of asynchronous processes that establish a fundamental gait cycle while maintaining body posture, and reacting to bumps and slips. We describe our implementation, and its relation to similar behavioral approaches, and discuss Dante II's performance during testing and on its descent into Mount Spurr.

Ahstruct-Gait generation is the formulation and selection of a sequence of coordinated leg and body motions that propel a legged robot along a desired path. Approaches to gait generation can be classified into control, behavioral, rule-based, and constraintbased paradigms. We survey these models of gait generation and introduce the Ambler, a hexapod robot that can circulate its legs to produce unique gaits. Then we present kinematic, collision, terrain, support, and stability constraints available for gait generation, and discuss our progress using a constraint-based method to generate the Ambler's gait. I. INTRODUCTION In comparison to wheeled mechanisms, legged rnechnisms require complex design, move slowly, and are difficult to control. However, for locomotion over rough or discontinuous terrain, legged mech'anisms are potentially superior to wheeled mech,uiisms.[ 11 Legged mechanisms make discrete terrain contacts and avoid undesirable footholds while wheeled mechanisms have rollers in continuous contact with the ground. The posture of a wheeled mechanism is dependent upon the terrain, but a legged mechanism can isolate its body from tlie ten ani- ' achieving a more stable stunce' and allowing smooth level motion. Legged locomotion is theoretically more energy efficient because body propulsion does not expend as much energy in soil compaction and body motion can occur in a level plane decoupled from the effects of gravity. Legged mechanisms can actively position their center of gravity to maximize stability. Dead reckoning (estimating position by integrating motion over time) can be more accurate in a legged machine where feet make discrete contacts and do not slip or skid like a wheel. The complexity in the design of legged mechanisms can be achieved, a$ is demonstrated by the working systems examined in this paper. We believe that the slow, smooth motion of legged devices is advantageous in rough terrain where caution (and, sometimes, sensing) will limit the speed of wheeled and legged mechanisms alike. The significant drawback to legs has been the difficulty in control. For a slow-moving walker, this challenge is not in moving the individual legs, a$ this is directly analogous to well-unders tood manipulator control, but in the coordination of leg and body motions. The legged robot, either with a real-time controller, low-level planner, or as an artifact of its architecture, must generate a sequence of leg and body motions, a gait, that will propel it along some path. Gait yeneralion is the formulation and selection of a sequence of coordinated leg and body motions that propcl the robot along a desired path. In this paper we survey several approaches taken toward gait generation aid present some of and discuss our work on gait generation for the Ambler.