Engineering topics
Publications and source records attributed to Smith, David E..
This presentation is about the robotic exploration of Mars using multiple targets command cycle, safe instrument placements, safe operation, and K9 Rover which has a 6 wheel steer rocket-bogey chassis (FIDO, MER), 70% MER size, 1.2 GHz Pentium M laptop running Linux OS, Odometry and compass/inclinometer, CLARAty architecture, 5 DOF manipulator w/CHAMP microscopic camera, SciCams, NavCams and HazCams.
In this paper, we introduce plan-graph based heuristics to solve a variation of the conformant probabilistic planning (CPP) problem. In many real-world problems, it is the case that the sensors are unreliable or take too many resources to provide knowledge about the environment. These domains are better modeled as conformant planning problems. POMDP based techniques are currently the most successful approach for solving CPP but have the limitation of state- space explosion. Recent advances in deterministic and conformant planning have shown that plan-graphs can be used to enhance the performance significantly. We show that this enhancement can also be translated to CPP. We describe our process for developing the plan-graph heuristics and estimating the probability of a partial plan. We compare the performance of our planner PVHPOP when used with different heuristics. We also perform a comparison with a POMDP solver to show over a order of magnitude improvement in performance.
The presence of large impact basins and the low northern plains that might have contained ice or liquid water at an earlier stage of Mars evolution suggests that the global gravity field could have been different in the distant past than it is today. In addition, any significant change in the distribution of mass affects the moments of inertia and consequently and could conceivably change the position of the pole and the length of day. Similar effects could have been produced by large erosional processes, such as the removal of crustal material from the Arabia Terra region and subsequent re-deposition in the Chryse region of the northern plains. We have endeavored to estimate the magnitudes of material that might have been involved in these processes and their possible effect on the gravity and dynamics of Mars. We have used present-day topography and gravity field as a starting point, recognizing that both the result of the processes that we are trying to study rather than the state at the times of interest.
During the last 8 years the US has successfully sent 2 major spacecraft to Mars 1z?der= planet. The Mars Global Surveyor spacecraft (IGS), which started operating at Mars in Sept 1977 and Mars Odyssey spacecraft that started operating in Jan 2002, are both operational and collecting data, including 1.5 meter pixel imagery, them spectrometry, magnetometry, altimeter/radiometry, gamma ray and neutron counts. imaging spectrometry, and gravity. The two orbiters are conceptually similar to Earth orbiting spacecraft in that their primary purpose is to observe, monitor, and measure a variety of parameters about the planet and report any changes that might be occurring on the surfaces and in the atmosphere. As a result we now have detailed physical descriptions of the shape, topography, magnetism. surface roughness, thermal inertia, atmospheric pressure and temperature, the distribution of hydrogen in the top meter of the crust, models for the planetary interior based upon data, and an image set that portrays much of the history of the past 3 billion years. understand how the planet came to be this way.
Although there are many approaches for compiling a planning problem into a static CSP or a scheduling problem, current approaches essentially preserve the structure of the planning problem in the encoding. In this pape: we present a fundamentally different encoding that more accurately resembles a scheduling problem. We sketch the approach and argue, based on an example, that it is possible to automate the generation of such an encoding for problems with certain properties and thus produce a compiler of planning into scheduling problems. Furthermore we argue that many NASA problems exhibit these properties and that such a compiler would provide benefits to both theory and practice.
There has been considerable work in AI on planning under uncertainty. However, this work generally assumes an extremely simple model of action that does not consider continuous time and resources. These assumptions are not reasonable for a Mars rover, which must cope with uncertainty about the duration of tasks, the energy required, the data storage necessary, and its current position and orientation. In this paper, we outline an approach to generating contingency plans when the sources of uncertainty involve continuous quantities such as time and resources. The approach involves first constructing a "seed" plan, and then incrementally adding contingent branches to this plan in order to improve utility. The challenge is to figure out the best places to insert contingency branches. This requires an estimate of how much utility could be gained by building a contingent branch at any given place in the seed plan. Computing this utility exactly is intractable, but we outline an approximation method that back propagates utility distributions through a graph structure similar to that of a plan graph.
The addition of durative actions to PDDL2.1 sparked some controversy. Fox and Long argued that actions should be considered as instantaneous, but can start and stop processes. Ultimately, a limited notion of durative actions was incorporated into the language. I argue that this notion is impoverished, and that the underlying philosophical position of regarding durative actions as being a shorthand for a start action, process, and stop action ignores the realities of modelling and execution for complex systems.
For a given problem, the optimal Markov policy over a finite horizon is a conditional plan containing a potentially large number of branches. However, there are applications where it is desirable to strictly limit the number of decision points and branches in a plan. This raises the question of how one goes about finding optimal plans containing only a limited number of branches. In this paper, we present an any-time algorithm for optimal k-contingency planning. It is the first optimal algorithm for limited contingency planning that is not an explicit enumeration of possible contingent plans. By modelling the problem as a partially observable Markov decision process, it implements the Bellman optimality principle and prunes the solution space. We present experimental results of applying this algorithm to some simple test cases.
The dynamic role of volatiles on the surface of Mars has been a subject of longstanding interest. In the pre-Viking era, much of the debate was necessarily addressed by theoretical considerations. A particularly influential treatment by Leighton and Murray put forth a simple model relying on solar energy balance, and led to the conclusion that the most prominent volatile exchanging with the atmosphere over seasonal cycles is carbon dioxide. Their model suggested that due to this exchange, atmospheric CO2 partial pressure is regulated by polar ice. While current thinking attributes a larger role to H2O ice than did the occasional thin polar coating this model predicted, the CO2 cycle appears to be essentially correct. There are a number of observational constraints on the seasonal exchange of surface volatiles with the atmosphere. The growth and retreat of polar CO2 frost is visible from Earth-based telescopes and from spacecraft in Mars orbit, both at visible wavelengths and in thermal IR properties of the surface. Recently, variations in Gamma ray and neutron fluxes have also been used to infer integrated changes in CO2 mass on the surface. Measurements made by Viking's Mars Atmospheric Water Detector experiment were sensitive to atmospheric H2O vapor abundance. Surface condensates and their transient nature were detected by the Viking landers. The study here is motivated by recent data collected by the Mars Global Surveyor, affording the opportunity to not only detect the lateral distribution of volatiles, but also to constrain the variable volumes of the reservoirs. We elaborate on a technique first employed by Smith et al. By examining averages of a large number of topographic measurements collected by the Mars Orbiter Laser Altimeter (MOLA), that study showed that the zonal pattern of deposition and sublimation of CO2 can be determined. In their first approach, reference surfaces were fit to all measurements in narrow latitude annuli, and the time dependent variations about those mean surfaces were examined. In their second approach, height measurements from pairs of tracks that cross on the surface were interpolated and differenced, forming a set of crossover residuals. These residuals were then examined as a function of time and latitude. The initial studies averaged over longitude to maximize signal and minimize noise in order to isolate the expected small signal. In this follow-up study we now attempt to extract the elevation change pattern also as a function of longitude, and we focus on the crossover approach.
Many NASA planning problems are over-subscription problems - that is, there are a large number of possible goals of differing value, and the planning system must choose a subset &it car! be accomplished within the limited time and resources available. Examples include planning for telescopes like Hubble, SIRTF, and SOFIA; scheduling for the Deep Space Network; and planning science experiments for a Mars rover. Unfortunately, existing planning systems are not designed to deal with problems like this - they expect a well-defined conjunctive goal and terminate in failure unless the entire goal is achieved. In this paper we develop techniques for over-subscription problems that assist a classical planner in choosing which goals to achieve, and the order in which to achieve them. These techniques use plan graph cost-estimation techniques to construct an orienteering problem, which is then used to provide heuristic advice on the goals and goal order that should considered by a planner.
Analysis of the orbits of spacecraft can be used to infer global tidal parameters. For Mars, the Mars Global Surveyor (MGS) spacecraft has been used to estimate the second degree Love number, k2 from the tracking DSN tracking Doppler and range data by several authors. Unfortunately, neither of the spacecraft presently in orbit are ideally suited to tidal recovery because they are in sun-synchronous orbits that vary only slightly in local time; and, further, the sub-solar location only varies by about 25 degrees in latitude. Never-the less respectable estimates of the k2 tide have been made by several authors. We present an updated solution of the degree 2 zonal Love number, compare with previous dues, and analyze the sensitivity of the solution to orbital parameters, spacecraft maneuvers, and solution methodology. Estimating the k2 Tidal Gravity Love Number of Mars.
The Mars Orbiter Laser Altimeter (MOLA) is routinely making radiometric observations of Mars at a wavelength of 1064 nm. Although the altimeter function is no longer operational, the MOLA detector continues to measure the reflectivity of the surface. Observations have been obtained almost continuously since the beginning of the Mars Global Surveyor (MGS) mapping mission in February 1999, and are providing measurements relevant to understanding the seasonal cycling of CO2 surface frost.
Radio tracking of the Mars Global Surveyor spacecraft has revealed temporal changes in the long-wavelength gravity field of Mars that correlate, to first order, with the pattern expected for the seasonal redistribution of carbon dioxide between the atmosphere and surface. Detecting these gravity field changes requires isolating very small perturbations in the velocity of the spacecraft and estimating the very low degree zonal coefficients of the field. A comparison of these coefficients determined every 5 days for a period over 2 Mars years shows annual and semi-annual variations that are similar to those predicted by a General Circulation Model simulation. These changes result from the redistribution of the mass of the planet by the exchange of carbon dioxide between the surface and the atmosphere through deposition and sublimation of CO2 in the polar regions. A simple time-dependent model for the icecaps enables an estimate to be made of the mass of carbon dioxide at each pole as a function of the seasonal parameter, Ls.
An airborne profiling laser is used to monitor multiple resources related to landscape structure, both natural and man-made, across regions encompassing hundreds of thousands of hectares. A small, lightweight, inexpensive airborne profiling laser is used to inventory Delaware forests, to estimate impervious surface area statewide, and to locate potentially Suitable Delmarva Fox Squirrel (Scrotum niger cinereus) habitat. Merchantable volume estimates are within 14% of US Forest Service estimates at the county level and within 4% statewide. Total above-ground dry biomass estimates are within 19% of USES estimates at the county level and within 16% statewide. Mature forest stands suitable for reintroduction of the Delmarva Fox Squirrel, an endangered species historically endemic to the eastern shores of Delaware, Maryland, and Virginia, are identified and mapped along the laser transacts. Intersection lengths with various types of impervious surface (roofs, concrete/asphalt) and open water are tallied to estimate percent and areal coverage statewide, by stratum and county. Laser estimates of open water are within 7% of photointerpreted GIS estimates at the county level and within 3% of the GIS at the state level.
Having judged dozens of science fairs over the years, I am repeatedly disturbed by the ground rules under which students must prepare their entries. They are almost invariably required to follow the "scientific method," involving formulating a hypothesis, a test of the hypothesis, and then a project in which this test is carried out. As a research scientist for over 40 years, I consider this approach to science fairs fundamentally unsound. It is not only too restrictive, but actually avoids the most important (and difficult) part of scientific research: recognizing a scientific problem in the first place. A well-known example is one of the problems that, by his own account, stimulated Einstein's theory of special relativity: the obvious fact that when an electric current is induced in a conductor by a magnetic field , it makes no difference whether the field or the conductor is actually (so to speak) moving. There is in other words no such thing as absolute motion. Physics was transformed by Einstein's recognition of a problem. Most competent scientists can solve problems after they have been recognized and a hypothesis properly formulated, but the ability to find problems in the first Place is much rarer. Getting down to specifics, the "scientific method" under which almost all students must operate is actually the experimental method, involving controlled variables, one of which, ideally, is changed at a time. However, there is another type of science that can be called observational science. As it happens, almost all the space research I have carried out since 1959 has been this type, not experimental science.
First generation spaceborne altimetric approaches are not well-suited to generating the few meter level horizontal resolution and decimeter accuracy vertical (range) resolution on the global scale desired by many in the Earth and planetary science communities. The present paper discusses the major technological impediments to achieving few meter transverse resolutions globally using conventional approaches and offers a feasible conceptual design which utilizes modest power kHz rate lasers, array detectors, photon-counting multi-channel timing receivers, and dual wedge optical scanners with transmitter point-ahead correction.
Asteroid 951 Gaspra appears to be in an obliquity resonance with its spin increasing due to the YORP effect. Gaspra, an asteroid 5.8 km in radius, is a prograde rotator with a rotation period of 7.03 hours. A three million year integration indicates its orbit is stable over at least this time span. From its known shape and spin axis orientation and assuming a uniform density, Gaspra's axial precession period turns out to be nearly commensurate with its orbital precession period, which leads to a resonance condition with consequent huge variations in its obliquity. At the same time its shape is such that the Yarkovsky-O'Keefe-Radzievskii-Paddack effect (YORP effect for short) is increasing its spin rate. The YORP cycle normally leads from spin-up to spin-down and then repeating the cycle; however, it appears possible that resonance trapping can at least temporarily interrupt the YORP cycle, causing spin-up until the resonance is exited. This behavior may partially explain why there is an excess of fast rotators among small asteroids. YORP may also be a reason for small asteroids entering resonances in the first place.
Since the arrival of the Mars Global Surveyor (MGS) at Mars in September 1997 and the subsequent beginning of observations of the planet there has been a constant stream of surprises and puzzling observations that have kept scientists looking at new 'out of the box' explanations. Observations of the shape and topography have shown a planet with one hemisphere, the southern, several kilometers higher than the north and a northern hemisphere that is so flat and smooth in places that it's difficult to imagine it was not once the bottom of an ocean. And yet the ocean idea presents some enormous difficulties. The measurements of gravity derived from the tracking of MGS have shown that several Mars volcanoes are enormous positive gravity anomalies much larger than we see on Earth and revealed small errors in the orbit of Mars and or Earth. And the magnetic field is found to be composed of a number of extremely large crustal anomalies; but as far as can be ascertained there is no main dipole field such as we have on Earth. Understanding these diverse observations and placing them in the sequence of the evolution of the planet will be a long, challenging but rewarding task.