---

Contents

---

1  Aims of the project

This is a joint work of the Dipartimento di Ingegneria dell'Informazione and of the Dipartimento di Scienze della Terra of Università degli studi di Siena.

2  Research group

3  External collaborators

4  Students

5  Research activity in 1998-1999

5.1  Mechanical design of the rover base

Figure 1: Snowmobile SKI-DOO

The vehicle consists of a 2m. x 1m. rectangular base mounted on two differentially driven tracks. Its design sketch is reported in fig. 2.

Figure 2: Track-based configuration of the rover.

The base of the rover is under construction. Mechanical details of the two tracks are reported in fig. 3 and fig. 4. Both tracks are powered by independent DC motors.

Figure 3: Rover base.

Figure 4: Rover Base (lateral view).

Placed upon the base of the rover there will be the control unit, the data pre-processing unit, and the sensors (including the visual system) necessary for navigation and detection of meteorites and obstacles.

The platform will also pull a trailer with an expected load of 200 Kg. The trailer will carry other sensors used for detecting meteorites both on the surface (digital telecamera) and underground (GPR: Ground Penetrating Radar, able to detect meteorites beneath the snow up to a depth of several metres).

This track-based configuration of the rover was chosen because of two factors:

1. simplicity in control and in driving, given the possibility of changing direction with ease and of steering in a very limited space;
2. large surface contact with the ground, especially useful for the icy stretches the vehicle will have to cross.

1. keeping the track at the proper tension by means of a complex system of springs and gear-wheels;
2. building the suspension system needed to tackle even the most uneven terrain without causing any damage to the vehicle itself or to the instruments carried aboard.

Figure 5: Track detail

5.2  Rover localization through GPS

The sperimental hardware setup consists of a GPS receiver "Zodiac Jupiter" by Rockwell, cf. fig. 6,

Figure 6: Inside view of the Rockwell Jupiter.

endowed with a passive and an active antenna and a station for the differential correction in RTCM format. The difference between the active and passive antennas is that the former, being amplified, offers a greater gain compared to the passive antenna. The GPS receiver with the active and passive antennas are reported in fig. 7.

Figure 7: GPS receiver with active and passive antenna. The passive antenna is placed on the receiver.

The Rockwell Jupiter is connected to a PC (serial port) which monitors data received. The two GPS receivers working in differential configuration communicate by means of a radio link.

A C procedure running on the PC decodes the incoming messages from the serial port, which are codified according to the Zodiac Binary protocol. It implements a conversion from hexadecimal to decimal to obtain intelligible data. The control law for mobile robot navigation will be  based on such data.

Each Zodiac binary message consists of a header part and a data part with own checksum. Each input/output message starts with a synchronization word. The second word in the header contains the numeric message ID. The third word contains the word count for the data part of the message. The fourth word is a 16-bit field allocated to protocol. These flag bits control ACK/NACK requests and implement message logging requests. Word five of the message header is the data checksum, used to validate the header portion of the message. The data portion of the binary message can be variable in length, as specified by the data word count found in the header.

The goal of this research is to execute a navigation experiment for the rover based on sensor information provided by the GPS. The aim of the experiment is to test and validate the practicability of GPS autonomous navigation on the anctartic plateau.

6  PNRA'99

6.1  Rationale

Within the issue of exploring the planet Earth, the Antarctic plateau has become of great interest in the last few decades. The Antarctic is the most productive region of the whole planet for the search of meteorites. The great number of findings is mostly due to its polar desert environment, which preserves extraterrestrial material for more than some hundred thousand years, and to the meteorite against ice color-contrast which allows an easy spotting in the field. Even more importantly, meteorites fallen on the ice cap are concentrated by ice dynamics in blue ice fields, the so-called 'meteorite traps'. The discovery of the Antarctic meteorite concentrations dates back to 1969, and since then more than 17,000 meteorite fragments have been recovered. This tremendous yield of extraterrestrial material has provided the scientific community with new insights on the origin and evolution of the Solar System, including planet Earth, Mars and the Moon.

In such a context, scientific and technological know-how of the partners involved in the present research program are joined to tackle the issue of exploring Antarctica by means of autonomous rovers.

With respect to the previous version, the present version of the project shows new features especially in the sense of clarifying synergies with the ENEA project MORICO3, (principal investigator: Dr. Moriconi). In fact, although the projects share some common features, their objectives are somewhat different and more importantly, cooperation of the two autonomous vehicles could make easier the achievement of both project targets. Specifically, the following considerations apply:

1. The variety of tasks and functionality requested for an autonomous vehicle working in Antarctica clearly show that cooperation of a `large' rover (4-5 meters, 5-6 tons) like the MORICO3 vehicle with a `small' rover (1-1.5 meters, 0.2-0.3 tons) like the one referred to in this project , would allow to optimize the execution of the different specific tasks (see subsection 7b.3).

The basic idea behind this project is to build an autonomous rover executing a number of tasks within the range of some kilometers around the base-station which may be either the base-camp or the `large' rover MORICO3.
2. Testing of the various subsystems and possibly of the whole rover could be performed jointly with the rover MORICO3 (in Norway, December 2000-March 2001). This will allow to verify both functionality of the single rovers and the cooperation strategies (localization, velocity estimation, on line map building).

It is worth while to recall that the University of Siena supported a pre-project (50 ML) during 1998, aimed to a preliminary investigation on mobile robotics for Antarctica exploration and the design of a pre-prototype (www-dii.ing.unisi.it/~ control/research/Antartide.html). The rover supporting frame, moved by two independent caterpillars, is under construction and should be completed by the end of next September.

6.2  Research design

Two possible scenarios are taken into account. In the first one, the mobile robot executes its tasks in the neighbourhood of the base camp within the range of some kilometers. In the second configuration, the mobile robot cooperates with other rovers to perform its tasks: the robot is taken to the area of interest by a robot of larger size, for instance the robot MORICO3-ENEA, then, the robot starts to explore the area in the neighbourhood of the robot MORICO3-ENEA and cooperates with it in acquiring and processing multi-sensorial data.

Some of the functional capabilities of the single rover and of the rover team are here reported:

Crevasse-warning

In this configuration the rover can serve as pilot for snow-mobile convoys. Through RES (Radio Echo Sounding) and video measures it will be able to detect obstacles, alarm the caravan and re-plan a new path free from collisions. Such a function is necessary for wide-range missions in Antarctica and becomes mandatory for cooperating rover tasks (second scenario). The advantage of using a small and light robot as pilot in the exploration task is apparent: the larger robot (e.g. the robot MORICO3), following the pilot trajectory, navigates in safe mode and can recover the light rover pilot in case of accidents.

Maps of meteorites

This application is mainly focused on the scientific research on meteorites. It consists in collecting data on the presence of meteorites on the surface of the plateau or buried in the ice. Beside the possibility of sample recovery, the detection of meteorites buried in the ice will provide 3D meteorite distribution maps. Such maps of meteorites will allow sample recovery and reliable estimates of the fluxes of extra-terrestrial material to Earth over time.

 

 
 
 

Although the rover should be able to independently build meteorites maps, this function could be carried out in a framework of cooperating rovers as well. Note that the cost of building a map with given accuracy could be substantially reduced by using a team of rovers to explore the area of interest. Moreover such a configuration would help to reduce the energy consumption and the level of pollution: a two level solution suggests to use the robot MORICO3 only for large trips (10 KM and more) to transfer the 'small' rover to the areas of interest.

Ground stratigraphy

Palaeo-climatic reconstruction and climatic forecast can be inferred by studying the ice stratification of the Antarctic plateau. This research project deals with both the development of GPR techniques (optimal filtering of data) and seismic reflection experiments. In this case the two rovers cooperate: the smallest one brings the geophone receivers around while rover MORICO3 generates pulses on the ground.

Maintenance

In order to reduce the risk of manned missions for maintenance operations in case of hard atmospheric conditions, the mobile robot, provided with a CCD camera, a laser finder and a simple robotic manipulator, can be used to execute autonomous or teleoperated maintenance procedures.

Robust localization and map building

This function is relevant to both the individual and cooperating rovers scenarios. In order to improve the knowledge of the environment, a team of robot can integrate their sensor capabilities through procedures fusing sensor data coming from CCD cameras, GPS's, laser finders and other sensor equipment. In particular, the small rover of the present project and that of MORICO3 could interact to increase the estimation accuracy of their positions and velocities.

 

 
 
 

The D.I.I. research group and that of Prof. Cipolla at the Lab. of Robotics and Artificial Intelligence in Cambridge are jointly working on artificial vision sensing. Specifically, researches on advanced techniques for estimating the structure of the scene and the rover camera motion relative to the scene through bidimensional images are in progress.

A joint researh with the LAAS-CNRS in Tolosa on localization and map building techniques for unstructured environments has also started up. A PhD student will spent one year at the LAAS-CNRS with the group of Prof. Lacroix.
 

6.3  Objectives

1. self-localizazing with respect to a world frame through procedures able to fuse sensor data; tracking pre-computed paths joining the base camp and important targets for maintenance and research tasks;
2. detecting the presence of meteorites on the surface or buried in the ice up to a certain depth and on board processing of sensor data to detect meteorites and optimally adapt trajectories in order to maximize information contained in the 3D maps of meteorites;
3. coring and collecting tools to pick up meteorites lying on the plateau surface or buried up to some hundreds meters;
4. obstacle avoidance and alarm docking base procedures;
5. cooperation with other rovers to optimize problem solving  procedures.

1. Construction and adaptive updating of navigation maps. This problem is usually approached starting from an initial and rough workspace description which is dynamically refined and continuously updated through sensor measurements collected during the mobile robot motion. Suitable measures of map reliability are associated with the map. These are given as probability distributions or hard bounds on the worst-case error. The main goal consists in generating optimized algorithms for handling uncertainties according to a deterministic approach. Reasearch on this topic is being presently developed at the D.I.I. and will be further reinforced through the recently started cooperation with the research group of Prof. Lacroix at the LAAS-CNRS of Tolouse.
2. Dynamic vision and filtering in autonomous navigation. The basic issue in this research area is to estimate the structure of the scene and the camera motion relative to the scene through bidimensional images. The specific goal pursued in this project is to study and test algorithms for tracking moving object contours, finalized to solving typical problems in autonomous navigation, such as adaptive obstacle recognition, ''time to contact'' estimation, etc..
3. Detection limit for masses buried in the ice by RES investigations. Tests aiming at evaluating the RES detection limit for masses buried within the ice have been carried out since Dec.'97. By using test samples such as metallic bodies deep into shallow bores, tests proved that the available RES techniques can detect meteorite-like masses of 1 kg at a depth of about 10m. In this project it is expected to improve the RES detection limits (reducing the minimum mass detection limit; increasing the depth of investigation) on an experimental basis.

 

6.4  Methodology

1. Description of the technological problems associated with a rover for the autonomous exploration of Antarctica. Particular attention will be devoted to the technological problems related to hard meteorological conditions. Definition of the technical specifications of some parts of the rover, like the engine, the sensors, the radar device for the localization of meteorites and crevasses, the processing unit, the radio communication device, and others.
2. Complete the mechanical base of the mobile robot under construction at the Lab. of D.I.I., to be used as a prototype for explorations in Antarctica-like environments. Design and build the electronic components and the robot sensing. The main actions to be performed are:
1. Adapting the robot propulsion system to the Antarctica environment.
2. Realizing a suitable thermal insulation and providing a thermostatic control for all the parts of the robot which are sensitive to a temperature below -20 celsius degrees.
3. Equipping the robot with all the sensors necessary to perform the required operations, and in particular with a high sensitivity radar device for the localization of meteorites and crevasses.
4. Providing devices for signal acquisition, conditioning and real time data processing.
5. Equipping the robot with a radio link to the base camp, other transmitting stations or rovers.
3. This last stage is devoted to designing and performing experiments, in order to test the prototype.

 

 
 
 
 
 

6.5  International cooperations

7  Bibliography

7.1  Robust estimation and filtering: theory and algorithms

1. A. Garulli, A. Tesi and A. Vicino (eds.), Robustness in Identification and Control, Springer-Verlag, 1999.
2. A. Garulli, A. Vicino and G. Zappa, 'Optimal induced-norm and set membership state smoothing and filtering for linear systems with bounded disturbances', Automatica, vol. 35, pp. 767-776, May 1999.
3. L. Chisci, A. Garulli and G. Zappa, 'Recursive state bounding by parallelotopes', Automatica, vol. 32, pp.1049-1056, 1996.
4. A. Vicino and G. Zappa, 'Sequential approximation of feasible parameter sets for identification with set membership uncertainty', IEEE Transactions on Automatic Control, No. 6, pp. 774-785, June 1996.

 

7.2  Autonomous mobile robot navigation

1. M. Di Marco, A. Garulli, D. Prattichizzo, A. Vicino, 'Constructing and updating navigation maps with uncertainty', submitted to 38-th IEEE Conf. on Decision and Control, Phoenix, 1999.
2. A. Garulli and A. Vicino, 'Uncertainty sets for dynamic localization of mobile robots', submitted to 38-th IEEE Conf. on Decision and Control, Phoenix, 1999.
3. G. Fusai and A. Vicino, 'Costruzione e aggiornamento di mappe di avigazione', Tech. Rep., Dipartimento di Ingegneria dell'Informazione, Universita' di Siena, 1998.
4. I. Palmas, A. Vicino, 'Costruzione ed aggiornamento di mappe di navigazione per robot mobili tramite triangolazioni', Tech. Rep., Dipartimento di Ingegneria dell'Informazione, Universita' di Siena, 1997.
5. S. Betge'-Bresetz, P. Hebert, R. Chatila, M. Devy, 'Uncertain map making in natural environments', Proc. IEEE Int. Conf. on Robotics and Automation, pp. 1048-1053, 1996.

 

7.3  Control of robotic manipulators

1. D. Prattichizzo and A.Bicchi, 'Dynamic analysis of mobility and graspability of general manipulation systems', IEEE Transactions on Robotics and Automation, 14(2), April 1998.
2. D. Prattichizzo, P. Mercorelli, A. Bicchi, and A. Vicino, 'Geometric control techniques for manipulation systems', in Proc. IFAC Conference on Control of Industrial Systems, Belfort, France, May 1997 in Invited Session on Control of robotics systems.
3. D. Prattichizzo, P. Mercorelli, A. Bicchi, and A. Vicino, 'On the decoupling and output functional controllability of robotic manipulation', in Proc. IFAC Symposium on Robot Control, Nantes, France, September 1997.
4. D. Prattichizzo, P. Mercorelli, A. Bicchi, and A. Vicino, 'On the geometric control of internal forces in power grasps', in Proceedings 36th IEEE International Conference on Decision and Control, San Diego, California, 1997.
5. D. Prattichizzo, P. Mercorelli, A. Bicchi, and A. Vicino, 'Active suspensions decoupling by algebraic feedback', in Proc. of the the 6th IEEE Mediterranean Conf. on Control and Systems, Sardinia, Italy, June 1998.
6. D. Prattichizzo, P. Mercorelli, A. Bicchi, and A. Vicino, 'Geometric disturbance decoupling control of vehicles with active suspensions', in Proc. IEEE Int. Conf. on Control Applications, 1998.
7. D. Prattichizzo, P. Mercorelli, and A. Vicino, 'Robust decoupling control of contact forces in robotic manipulation', in Proc. of 2nd IFAC Symposium on Robust Control Design, Budapest, Ungary, June 1997.
8. D. Prattichizzo, J. Salisbury, and A. Bicchi, 'Contact and grasp robustness measures: Analysis and experiments', in O.Khatib and J. Salisbury, editors, Experimental Robotics IV, Lecture Notes in Control and Information Sciences 223, pages 83-90, Springer Verlag London, 1997.

7.4  Dynamic vision

1. A. Garulli, D. Prattichizzo, A. Vicino, 'A set theoretic approach for time to contact estimation in dynamic vision', 37-th IEEE Conf. on Decision and Control, Tampa, December 1998.
2. A. Garulli, D. Prattichizzo, A. Vicino and G. Zappa, 'Uncertainty interval evaluation for time-to-contact estimation problems', Proc. 3rd IFAC Symposium on Intelligent Autonomous Vehicles, Madrid, March 25-27, 1998.
3. G. Chesi, E. Malis and R. Cipolla, 'Collineation estimation from two unmatched views of an unknown planar contour for visual servoing', Tech. Report, University ofCambridge, 1999.
4. R. Cipolla, Active Visual Inference of Surface Shape, Lecture Notes in Computer Science, Springer-Verlag, 1995.
5. S. Maybank, Theory of Reconstruction from Image Motion, volume 28 of Information Sciences Series, Springer Verlag, 1992.
6. A. Blake and A. Yuille, Eds., Active Vision, MIT Press: Cambridge, MA, 1992.
7. Y. Bar-Shalom and T. E. Fortmann, Tracking and Data Association, Academic Press, 1988.

7.5  Meteorites

1. N. Perchiazzi, M. Mellini, 'Revisione e riclassificazione della collezione di meteoriti del Museo di Storia Naturale dell' Universita di Pisa', Atti Soc. Tosc. Sc. Nat., 1996.
2. L. Folco, M. Mellini, 'Crystalchemistry of meteoritic kirschsteinite', European J. Mineralogy 9, 969-973, 1997.
3. L. Folco, M. Mellini, C. T. Pillinger, 'Equilibrated ordinary chondrites: constraints for thermal history from iron-magnesium ordering in orthopyroxene', Meteoritics and Planetary Science, 32, 567-575, 1997.
4. M. Mellini, 'The need for electron crystallography in mineral sciences. In 'Electron Crystallography', (ed. Dorset ) Kluwer Academy Press, 1997.
5. W. Cassidy, R. Harvey, J. Shutt, G. Delisle and K. Yanai, 'The meteorite collection sites of Antarctica', Meteoritics 27, 490-525, 1992.
6. N. Perchiazzi, L. Folco, M. Mellini (1999), 'Volcanic ash bands in the Frontier Mountain and Lichen Hills blue-ice fields, northern Victoria Land", Antarctic Science
7. A. Capra, M. Chiappini, L. Folco, M. Frezzotti, M. Mellini, I. Tabacco (1999), 'La trappola per meteoriti di Frontier Mountain (Terra Vittoria settentrionale): morfologia del basamento roccioso, flussi del ghiaccio e velocita' di ablazione. Convegno Naz. Glaciologia Antartica e Paleoclima, Padova.
8. L. Folco, M. Mellini (1999), 'Meteoriti antartiche: un bilancio di quattro spedizioni e programmi per il futuro', Convegno Naz. Glaciologia Antartica e Paleoclima, Padova.