Cover Image

Not for Sale



View/Hide Left Panel
Click for next page ( 332


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 331
CHAPTER 19 THE AUTOMATED BIOLOGICAL LABORATORY D. A. GLASER, JOHN MCCARTHY, AND MARVIN MINSKY INTRODUCTION The state of evolution of the Earth is very complex to describe, let alone to discover. Mars may be in a much simpler state, but we cannot count on it. Even if Mars is in a much simpler evolutionary state than the Earth, there is much work to be done before Mars is nearly so well understood as our own planet. We are now considering what can be done with a single unmanned lander weighing several thousand pounds. Our present ability to make small scientific equipment already permits us to include a wide variety of techniques within our weight limit. The problem we shall face in this chapter is to suggest how the various devices can be coordinated into an automated biological laboratory (ABL) that will give us a good chance of determining whether there is life on Mars, and in any case, of giving an estimate of the state of its chemical evolution. In our opinion, the key to making the automated laboratory effective is to make it a computer with sensors and effectors rather than a collection of isolated experiments. It should be possible to use a piece of apparatus such as a television camera or a mass spectrometer in a number of dif- ferent ways in experiments aimed at answering different questions. More- over, we want to maintain as much control of the experimental program 331

OCR for page 331
332 MARTIAN LANDINGS: UNMANNED from the Earth as the 5 to 25 minute round-trip time for signals between Earth and Mars will allow. Only the maintenance of this control will give us much chance of getting a reasonable picture of Martian evolution from a single mission or even from a small number of missions. Few biologists have thought much about the computer control of ex- periments, and there is a temptation to put the idea aside as too compli- cated for an early mission and settle for adapting to predicted Martian conditions a few experiments that would be simple if performed on Earth. This puts a heavy burden on our ability to predict Martian conditions, and we must face the fact that many of the experiments planned would turn out to be inappropriate. A much greater chance of success is offered by a co-ordinated laboratory that can be ordered to change the experi- ments from the Earth after the first results are returned. Computer Programs for Controlling Experiments A computer program is a sequence of instructions in the memory of the computer. The ABL computer should have room in its main memory for, say, 50,000 instructions, and substantial secondary storage, such as magnetic tape. It executes instructions one after another. Some of these instructions do arithmetic operations involved in computing the next value of the magnetic field for the mass spectrometer, some compute where to point the television camera, or when to end a titration. Other instructions select the instructions to be executed next according to whether an ex- perimental operation is complete or whether an iterated computation has been carried out the right number of times, or whether a signal has come from Earth indicating that a new program is being transmitted. Other instructions turn on or off experimental apparatus such as the motor that rotates the camera in azimuth or the motor that extends the sample col- lection arm. Other instructions cause information to be transmitted to the Earth after it has been edited into a compressed form that will make best use of limited transmission bandwidth. Time Scales It is important to understand the time scales involved. The computer executes an instruction every few microseconds. A simple mechanical operation involving the experimental apparatus takes between one tenth and ten seconds. The round trip time for a signal from Earth is between 300 and 1500 seconds. Thus the computer can execute about 105 in-

OCR for page 331
The Automated Biological Laboratory 333 structions in the time required for a mechanical operation, and we can perform, say, 2000 mechanical operations in the time required to look at the result of some complex of operations and decide what to do next. When the ABL is on the opposite side of Mars from the Earth it will be on its own for 12 hours and could be shut off if we cannot program a useful strategy. These times are the key to understanding the possi- bilities and problems of computer control of the automated biological laboratory. First of all, 105 computations in a mechanical operation time means that the computer can control, say, 100 mechanical devices at a time and still execute an average of 1000 operations in deciding what each device is to do next. This means that the procedure for deciding what each device is to do next may be quite elaborate, if this is desirable. Secondly, if we want to use our device with full effectiveness, we must delegate to the computer program control of up to 2000 elementary actions of each device while we decide on the next compound action. Thus, we should program complex actions such as: a complete sequence of separation actions such as solvent extractions, titrations, and scans of the mass spectrum, including the decisions about endpoints or when the mass spectrometer has been at a given charge-to-mass ratio long enough. More elaborately, if we can, we should program the computer to collect objects of a kind we are interested in. Kinds of Experiments Let us try to classify the experiments that might be performed in the following way: a) Observations. Most important will be pictures on scales ranging from telescopic panoramas to photomicrographs. Also there may be temperature, pressure, atmospheric chemical content, sound, and radia- tion measurements. b) Analysis of samples. For example, we may use a computer-con- trolled shovel to pick samples, grind them, dissolve them in chemicals, use solvent extraction and chromatographic methods to concentrate frac- tions of high optical activity, finished off by mass spectrometer analysis of the final concentrate. The remaining concentrate may have to be stored while scientists on Earth decide what further tests shall be performed. Physical as well as chemical analyses will be made. c) Growth experiments. A specimen may be put in a variety of en- vironments, and effects of growth, or other signs of life, looked for peri- odically.

OCR for page 331
334 MARTIAN LANDINGS: UNMANNED Limitations of Programming The limits of what can be programmed for the ABL are not easily set. A large class of useful operations is well within the state of the art. For example, it is not difficult to program a fractionation process to select for further analysis the fraction that shows optical activity or shows frag- ments at given mass numbers on the mass spectrometer. It would also be easy to program the machine to transmit only the parts of pictures that differ from previously stored pictures of the same scene. It is fairly easy to program a computer to make a hardness map of a mineral specimen by poking it with a needle and transmitting this together with a picture of the specimen. It is difficult, but probably possible, to take pictures of a desert scene and, after looking at them, program the computer to transmit pictures of cacti that differ from the already classified types of cacti, as these are encountered in the ABL's travels. If biologists are to be able to ask for this kind of performance, they will need the support of extensive earth- based computer facilities and programming groups. It is not now within the state of the computer art to program a com- puter to control the dissection of a mammal, much less to perform an operation on a mammal such as might be involved in a physiological experiment. By 1971 this situation may change if a determined effort is made, but it would be unwise to count on it. It would also be unwise to make decisions that preclude it. When we cannot program a kind of decision, the experiment is slowed up because we must send a picture of a tray of samples, or mass spectro- grams of fractions to Earth for decision as to what objects should be ground up or what fractions should be further treated and how. Fortu- nately, the automated biological laboratory can provide us with complete flexibility in this respect. If a particular decision that has to be made on Earth is slowing our progress, we can test on Earth suitable programs for making the decision, and when we think they are correct transmit the programs to Mars. Danger of Thinking too Small We must confess to the following fear: At present, the art of pro- gramming computers to select objects of a given kind by looking at a picture of collection of objects against a background, is in a rather primi- tive state. On this basis, it might be decided that although the ABL is to be provided with the ability to take pictures and transmit them to Earth, the Mars computer will not be able to look at the pictures. (The com-

OCR for page 331
The Automated Biological Laboratory 335 puter looks at a picture by having an instruction that allows it to read the optical density at a point on the picture with given co-ordinates; pro- grams can be written, using this operation, that track the light-dark boundaries and recognize objects). We believe that it is extremely im- portant to make the ABL completely flexible. This requires that all appa- ratus be subject to computer control, and that all information collected by the sensors be readable by the computer. It is also important that the necessary computer programming and checkout facilities be available on Earth to allow the quick changing of computer programs to meet changed experimental conditions. If our view of what we will be able to program and of the benefits of flexibility proves over-optimistic, little will have been lost. The computer is still the best way to control even relatively simple processes, as industrial experience is showing. On the other hand, if through lack of imagination a decision is made for a preprogrammed system, or even if the computer and its programming are set up in a way that makes changes difficult or risky, or if not all sense information is available to the programs, a tre- mendous opportunity will be lost. Summary In the succeeding sections of this chapter we shall treat the following topics: the state-of-the-art in computer control; description of a simple automated laboratory; control of the laboratory from the Earth; tele- vision systems, transmission of pictures, and the problems and uses of computer picture pattern recognition; sample collection and the computer controlled hand; the advantages and the problem of making the ABL mobile; some recommendations for research and development projects that may be undertaken now to provide support for the ABL. The automated biological laboratory provides a marvelous focus for research and development in computer control systems. The potential technical benefits for the control of scientific experiments and other proc- esses on Earth seem as great as that for any other aspect of the space program. By itself it may repay the cost of the entire Mars exploration. THE STATE OF THE ART OF COMPUTER CONTROL The art of computer control of external devices is advancing rapidly. If this were 1955, what we propose in this report would be almost im- possible, and if it were 1975, what we have to say would be regarded as obvious by every scientist. We are now at the point where the tools are

OCR for page 331
336 MARTIAN LANDINGS: UNMANNED comfortably available, but we shall have to work fast to make good use of them. Computers in Airplanes and Spacecraft Modern fighter planes contain computers for navigation and fire con- trol. They are usually magnetic drum computers, and their programs are rarely changed. They are reliable enough for their present use and com- pact enough, even for the Mars mission. However, they are not fast enough, they are not easily programmed, they do not have sufficiently large memo- ries, and they are probably not sufficiently reliable for use in the ABL. The M.I.T. Instrumentation Laboratory has designed, and IBM is build- ing, a computer to be carried on the Apollo spacecraft. This computer is probably fast and reliable and small enough for the ABL, but it uses a read-only memory for programs and does not have enough memory. The proposed supersonic transports are to be controlled by digital com- puters. Several American companies have designed computers for in- clusion in spacecraft, but we believe that the ABL computer can and should be more powerful than these. Computer Control of Industrial Processes Chemical plants, bakeries, atomic power plants, and nuclear particle accelerators have been controlled by computers. Most of these programs have been rather simple, certainly simpler than we shall want for the ABL. Time-Sharing The ABL computer must be able to manage many pieces of apparatus at the same time. This is possible because the computer is nearly 100,000 times faster than the apparatus it controls. The art of making a computer carry out a large number of separate tasks at the same time without con- fusion is called time-sharing. Systems that allow a computer to interact simultaneously with tens of people and external devices are in use today. They have the property that an error in one user's program cannot result in interference with the programs of any other user. This property is essential for the ABL if we are to dare to allow scientists to change pro- grams after the machine is on Mars. Picture Recognition Some work has been done on programming computers to classify pic- tures into a number of categories. These programs even learn the cate-

OCR for page 331
The Automated Biological Laboratory 337 gories from examples. The number of categories and their complexity are quite limited so far. Other work has concentrated on the more relevant problem of picking out objects of given categories from a background and measuring their positions and dimensions. The work in recognizing nu- clear events in bubble chamber and spark chamber pictures is relatively advanced. The apparatus for this recognition work is just becoming avail- able, and rapid advances may be expected because the problem is being pursued energetically. Artificial Intelligence This is the problem of making computers perform tasks which, when performed by people, are considered to require intelligence. Modest suc- cesses have been achieved, but progress is likely to be slow. This work has led to an ability to identify those tasks that are readily assigned to a computer and those that still require human intervention. Computer Performance In our opinion the ABL can profitably use a computer of large scale by present standards. Great advances are being made in minaturizing computers. However, we do not yet know whether a large scale computer (with, say, 216 words of one microsecond memory) can be reduced to 200 Ib, in time, or whether we will have to compromise in this area. The result of a compromise would be to reduce the complexity and number of processes that can be controlled simultaneously and to increase the time required to change the course of the experiments. A SIMPLE AUTOMATED LABORATORY In order to clarify the problem of automating a biological laboratory, we shall consider a simple one. We mention specific apparatus not to express an opinion about what should be included—much more can be included than is listed here—but merely to make the control problem concrete. Let us assume the following equipment: 1. A television camera and a storage tube. The camera has a variety of lenses for magnifications from telescopic to microscopic. An arm per- mits the camera many positions: on a tower for looking at the landscape; attached to a microscope for looking at slides; overlooking the immediate foreground for controlling an arm used for picking up samples or for

OCR for page 331
338 MARTIAN LANDINGS: UNMANNED controlling the motion of the ABL over the ground; and a position that allows the camera to look inside the ABL in order to see the positions of movable parts. Several cameras may be taken if the workload or relia- bility requires it. The computer can transfer information from the storage rube into its memory either en masse or point by point. Computer pro- grams compress picture information for digital transmission and use the same information to make decisions. 2. One or more mechanical arms like those used for handling radio- active materials are under the control of the computer. They can be positioned to computed positions when the computer knows the precise sequence of motions desired, or can be controlled via the picture infor- mation by the computer when a servo-mechanical type of operation is required. The arms can use tools such as shovels, coring drills, and a variety of clamps for holding objects of various shapes. 3. A wet chemical laboratory. Reagents may be added to samples, and operations such as titration, centrifuging and filtration may be performed. 4. Optical spectrometry. 5. Mass spectrometry. CONTROL FROM THE EARTH The ABL will be carrying out simultaneously a wide variety of experi- ments in a number of different fields. Many of these experiments are of kinds that, on Earth involve continuous supervision by the experimenter. The round trip signal time precludes continuous supervision and so we have emphasized computer control. Nevertheless, we want to make human supervision as effective as possible, and this requires very sophisticated Earth-control. We envisage the following kind of system: 1. There are several groups of scientists, each pursuing its own line of investigations. 2. Each group has consoles for the display of information coming from Mars, and for transmitting instructions to that part of the computer program on Mars carrying out the group's investigations. They have com- puter facilities on Earth for analyzing data and for correcting new pro- grams to be sent to Mars. 3. The consoles are attached to a computer on Earth that coordinates their communication with the experiment on Mars. The allocation of resources among the groups is decided by directorate and administered by programs in the Earth computer and to a lesser degree in the Mars computer. These decisions include:

OCR for page 331
The Automated Biological Laboratory 339 a. The rate of which each group can obtain pictures and other data from Mars. b. Allocation of expendable supplies. c. Decisions about when and where the lander will move. d. Allocation of the services of the arm, the chemical analyzer, and the cameras. One might argue that it would usually be better to do one experiment at a time, and this might be true if we could program all decisions for the Mars computer. However, if a particular experiment has to be car- ried out in such a way that a short operation is first performed, and then the result sent to Earth for decisions, there will be much time wasted if only one experiment is done at a time. 4. Each group will strive to program the decisions needed to carry out its experiments. For example, suppose an experiment requires the selec- tion of objects from a shovelful for subsequent chemical analysis. At first, it may be necessary to have a television picture of the shovel re- turned to Earth in order to select the objects. However, the experiment will go faster once the selection criterion can be programmed for the Mars computer. The programs to do this will be verified on Earth-bound copies of the Mars computer operating Earth-bound copies of the ABL. 5. Because of the limited time the lander will operate, the results ob- tained up to a given time should be available in raw form to the whole scientific community. This will enable suggestions to be made and even new groups to start new research programs using the ABL, if their pro- posals seem to warrant it. THE CHEMICAL LABORATORY It is too soon to say what the chemical analysis facilities should be in detail. However, some general remarks can be made. A chemical analysis procedure is a strategy involving the following kinds of operations. 1. Physical preparation of the sample. Grinding, etc. 2. Mixing reagents with the sample. 3. Controlling the physical environment. Temperature, pressure, illumination. 4. Separation. Filtration, centrifuging, solvent extraction, chromatog- raphy. 5. Physical measurements. Presence (e.g., did anything precipitate), weight, color, reflection spectrum, form (flocculent precipitate; if we want the criterion, we need a computer program to recognize it); spectrum; mass spectrometry; optical activity; density; viscosity.

OCR for page 331
340 MARTIAN LANDINGS: UNMANNED 6. Storage. Some fractions may be put aside for later use. In general, the results of the physical measurements determine what mixing and separation operations will be performed next and what frac- tions will be put aside or discarded. Besides reliability, the following con- siderations should determine the methods made available: 1. Generality. As few assumptions as possible about the chemical en- vironment of Mars should be made. Some long shot guesses can be ac- commodated by including special reagents. 2. Economy. The consumption of expendable supplies per experiment should be very low. 3. Speed. Automated mechanical movements can be very fast; five operations per section are readily achieved. This means 3 X 10s elementary chemical operations may be performed in the life of the ABL. If one milligram of supplies is consumed per operation we will need 300 kg of supplies. The above figures represent our guess as to the order of magni- tude of the quantities involved, and perhaps they suggest that expendable supplies will be the limiting factor on how much chemistry can be done. Much present chemistry depends on having large excesses of certain reagents, especially water. Perhaps, in order to get by with milligram amounts of reagents one should use microgram amounts of sample. It should be pointed out that the mechanical movements can be speeded up to 100 per second, if small enough masses have to be moved. One may ask whether there would be any use for 3 X 108 chemical events. Would not some smaller number, say 3 X 104, do? We believe that the larger number is quite likely to be wanted because the reactions will be combined into procedures, and each procedure may involve hundreds of chemical events. The reactions themselves should usually proceed on the 1/10 second time scale although one second reaction times can be tolerated if one vessel can be put aside to react, while others are manipulated. The problem of cleanliness is a large one. Perhaps disposable liners for the reaction vessels will solve the problem. Difficult-to-clean vessels like stills, and perhaps continuous processes generally, may turn out to be impractical. PICTURES AND VISUAL CONTROL OF EXPERIMENTS In discussions of research techniques one takes vision for granted. However, the beginner in biology is often bewildered by the expert's sure identification of the important object in what appears to be a very com-

OCR for page 331
The Automated Biological Laboratory 341 plicated or indistinct picture. We are all beginners so far as Mars is con- cerned, but we would like to become experts. In this section we deal with two topics: 1. Returning pictures to Earth to develop our understanding of what things on Mars look like. 2. Programming the identification of objects by the ABL computer so that it can avoid obstacles, select samples of desired kinds for analysis, send back pictures of previously unseen objects, etc. What is there to look at? Much more of Mars will be visible from the ABL than can be inspected with any other sense. A camera boom on the ABL, and the ability to move the ABL to scenic lookouts will increase what can be seen. What can be seen may be divided into topography and objects. Some general information about topography will already be available from the environmental flights required to assure the safe landing of the ABL. The additional topographical information obtained by the ABL will be useful, if correlated with the objects. The possible varieties of objects are too numerous to catalog. They include craters, vegetation, mineral outcroppings and many objects that may be difficult to classify when first seen. In order to extract the maximum information from distant objects, the ABL needs telescopes of various magnifications, with emphasis on the maximum usable magnification. Color information may provide useful clues about the composition of the surfaces seen. This suggests that we include the ability to photograph a scene through an arbitrary spectral window and that we develop the ability to infer composition from such reflection spectra. The near scene also requires photography at various magnifications. Next we come to photography of objects that we can manipulate. Here are some examples: 1. Lichen on a rock. 2. Objects under an over-turned rock and the bottom of the rock. 3. The stratification of a hole we have made. 4. Fragments of a broken or breakable abject. 5. Sections of a sectionable object. 6. A precipitate or polymer resulting from a chemical process. The picture handling system should include the following: 1. Optical instruments—telescopes, microscopes. 2. A television camera with a storage tube. 3. A picture storage system, e.g., video tape in the ABL. 4. The ability of the computer to look at points in pictures on the storage tubes.

OCR for page 331
342 MARTIAN LANDINGS: UNMANNED 5. Computer programs for digitizing and compressing picture infor- mation. 6. Computer programs for recognizing objects of various kinds. 7. Transmission facilities for sending the pictures to Earth. Pictures are likely to require more bandwidth than any other information trans- mitted. Programming Computers to Recognize and Handle Objects First we shall list the relevant research. 1. H. A. Ernst programmed the TX-O computer to control a mechanical hand to pick up blocks and stack them. (1961 M.I.T. Sc.D. Thesis in Electrical Engineering). 2. L. Hodes and T. Evans at M.I.T., while working under Minsky, pro- grammed the IBM 7090 to find geometrical objects when partially over- laid with other objects. 3. A number of physics groups have programmed computers to find events of given sorts in pictures of spark chambers. The group at Argonne National Laboratories has used their system to count the number of chromosomes of each of several types occurring in a picture. 4. A programmable film reader that reads radar traces and graphs from film and writes magnetic tapes with the information in digital form is marketed by Information International. 5. A large amount of work has gone into the classification of whole pictures. This is not very relevant for the present purpose that requires the identification of objects in a picture in a manner that will allow the manipulation of the objects. 6. A system called FIDAC has been developed by R. S. Ledley (Science, Oct. 9, 1964) that scans pictures and reads about 106 bits of information into the memory of an IBM 7090 computer for analysis. A programming system for picture analysis has also been developed, and is being used to classify chromosome pictures. Much of the work on picture recognition uses the following technique: The computer controls the position of a spot on the face of a cathode ray tube, i.e., there is a computer instruction that says position point of light at image coordinates (x,y). An optical system projects the light point through a photographic transparency and onto a photomultiplier cathode. An analog-to-digital converter makes the photomultiplier cathode cur- rent, which is proportional to the transparency of the photograph, avail- able to the computer. Thus the basic computer instruction is to determine the optical density at a point on the film with given co-ordinates. Several variants of this apparatus have been used or proposed to in-

OCR for page 331
The Automated Biological Laboratory 343 crease the speed of various recognition schemes. The "PEPR" apparatus for measuring bubble chamber pictures displays a bar whose length and orientation, as well as position, are controlled by the computer. This is useful for detecting tracks of particles. The Argonne apparatus scans a rectangle chosen by the computer and returns the co-ordinates of all points where a change in density occurs from one level to another (64 levels are used in this case). Given this basic facility, the computer can be programmed to find objects of various kinds, for example by tracing their outlines. The apparatus is only now beginning to be available, and all the present facili- ties are dedicated to very specific applications. Besides the above-mentioned work, much attention has been given to apparatus and programs that classify pictures as a whole, e.g., this is a picture of an A. It is difficult to see how these methods can help with the present problem, but the advocates of perceptions, and adelines, etc. will speak for themselves when the time comes. All the above-mentioned work, except Ernst's which used photocells and mechanical sensors, has been concerned with photographs. Direct recognition of objects requires a television system, in which the computer can ask for the intensity at a given point on an electrical image of the scene. Systems of this kind have been designed, but are not yet built. A number of additional techniques have been proposed, such as using the magnitude of the high frequency component of the intensity in a scan as a measure of whether the camera is in focus and using focus to measure distance. We believe that a useful capability for recognizing and manipulating objects can be available in time for the ABL if a prompt and serious effort is made. Mechanical Manipulation We believe it is reasonable to consider basing much of the ABL's me- chanical activities upon a set of general purpose computer-controlled manipulators. Each manipulator would consist of a fast, firm positioner, with several degrees of freedom, and an attachment that can hold a variety of special tools, graspers, or sensors. Interchangeability would mean much more flexibility and capability than could be obtained by the same number of actuators installed in particular experiments for fixed purposes. The most straightforward design would have motions of the positioner based on a fixed coordinate system relative to some points on the ABL vehicle. We should also study the possibility of making the manipulator

OCR for page 331
344 MARTIAN LANDINGS: UNMANNED along the general lines of the human arm and hand. With computer con- trol and visual monitoring (by computer) this may be practical. Computer-controlled manipulation exists today chiefly in the form of automatic machine-tool control systems. Visual control of manipulators has not been developed, but we believe the state-of-the-art is just ready for such a development. The computer-controlled, human-like arm, de- veloped by H. A. Ernst, used only simple tactile sensors, but it could find a number of scattered blocks, stack them up in a tower, and then put them in a box that it had to find. The advantages of general-purpose manipulators include: Reduction in weight, as compared to many special activators. Great flexibility in programming sample-collection and material trans- fers. Adjustment of physical layouts of experiments. Assembly of parts into many configurations. Adjusting parameters of experiments. Some possibilities of repairs on site, or at least replacement of parts. In particular, a manipulator could serve to control TV cameras, outside and within the spacecraft. It might be feasible to use it to lay out and phase a large, efficient, outside antenna. It could operate micro-tools, through a reduction device. It is possible that basing operations on a few reliable manipulators could yield a substantial gain in overall reliability, since it would then be possible to simplify most experiments. It may be preferable to adjust the mechanical parameters of an experiment by moving a simple stud or tab, instead of installing and depending on a special motor or actuator for that task. It is not possible to say now which system would be most reliable and compact. We shall want arms of several sizes. Very small arms can move very quickly. MOBILITY We believe that the effectiveness of the ABL can be greatly enhanced by making it mobile. There is a substantial chance that it will land in an unsuitable place such as a small crater, and there may be very little to observe. A negative conclusion about the existence of life would be quite suspect, if based on a single site or even a few sites. The problem of providing mobility has three aspects: 1. The power available will be quite small. For example, the Beagle

OCR for page 331
The Automated Biological Laboratory 345 study estimates that a total 300 watts will be available for all activities of a 5000 Ib package. Other studies estimate up to 2 kilowatts. 2. The terrain is unknown. 3. Step-by-step control from the Earth is hampered by the signal round- trip time. The answer to the power problem is to go slowly. A velocity of one meter per second, the maximum that could be hoped for, would permit covering 60,000 km in two years. This amounts to two circumnavigations of Mars. Even one cm per sec would permit 600 km of travel. If the ABL could find vantage points along its route, permitting 10 km of side visibility, this allows the exploration of 12,000 square kilometers in the sense that a number of visually interesting objects could be approached and examined in detail. A one meter per second velocity means that the ABL would cover about one kilometer in a round trip signal time. This precludes detailed control from Earth and requires a computer program that can use tele- vision information to steer a course. On the other hand the one cm per second rate permits only 10 meters to be covered, so instructions to the vehicle can be based on a human look at the terrain to be covered. The decision on what mobility system is best is a complicated one. In this chapter we shall mention only two complementary systems that suit the low power available. The first system is to have a long arm that can extend a drill that can make a hole and that can attach an anchor. A winch is then used to haul the ABL with whatever power can be spared. Three cable-anchor combinations are needed. This system can deal with almost any solid terrain, including cliffs. More suitable for flat ground with unavoidable obstacles not more than one meter high is a system of eight legs, two at each corner. One set of legs is lifted, advanced and set down, the second is lifted, advanced and set down, and finally the body of the ABL moves forward. The legs can extend to different lengths and are as light as possible. We minimize up- and-down motion of the ABL in order to reduce the power used. Research and Development Projects The purpose of this section is to identify some research and develop- ment projects that should be started soon if the ABL is to be maximally effective. 1. Identification of substances by reflection spectra. The ABL will be

OCR for page 331
346 MARTIAN LANDINGS: UNMANNED able to see much more than it can touch. 2. Computer recognition and manipulation of objects. 3. Computer controlled wet chemistry on as small a scale as possible. 4. Computer control of a vehicle. One should work towards systems that have at least the human ability to tolerate variations in terrain.