6TH EUROPEAN CONFERENCE FOR AERONAUTICS AND SPACE SCIENCES (EUCASS)

Mobile platform for a drilling system M. Ciszewski*, T. Buratowski*, A.Gallina*, K. Seweryn**, W. Teper***, A. J. Zwierzyński***, T. Uhl* * Department of Robotics and Mechatronics, Faculty of Mechanical Engineering and Robotics, AGH University of Science and Technology, Al. Mickiewicza 30, 30-059 Krakow, Poland ** Space Research Centre, Polish Academy of Sciences, ul. Bartycka 18A, 00-716 Warszawa, Poland *** Department of Drilling and Geoengineering, Faculty of Drilling, Oil and Gas, AGH University of Science and Technology, Al. Mickiewicza 30, 30-059 Krakow, Poland

Abstract In this paper design process of a mobile platform intended for an ultralight mobile drilling system (UMDS) is presented. The main goal of the system is to extract soil probes from hardly accessible places and environment difficult to explore, thus appropriate mobility must be a significant feature of the system. The paper consists of a concept of drilling operation, followed by design workflow for the mobile platform, leading to creation of a prototype that was tested in a laboratory. The rover features a special rocker design with four wheels, central differential mechanism and roll angle compensation by adjustments of height of rocker arms.

1. Introduction The paper presents a complex mechatronic system designed for different purposes related to drilling processes, sample extraction and its delivery to a designated area. All subsystems are designed to operate in terrestrial conditions and can be adapted to space environment. The device must have possibly low mass to enable transport by an Unmanned Aerial Vehicle (UAV) to a place designated for drilling. High level of autonomy of the drilling system must be provided in order to realize drilling tasks without human supervision. In terrestrial applications, drilling performance is usually optimized due to economic reasons. This leads to usage of high power motors and effective but heavy drilling bits to increase the rate of penetration. In space conditions the time of drilling is not an important factor, while minimization of mass and power consumptions has the highest priority. In addition, the planetologists indicate the need to sufficiently secure the samples gathered by drills in terms of cross contamination in the borehole as well as influence of the heat dissipation during the drilling process[1]. Therefore, our design is focused on minimization of power which needs to be delivered to regolith or rock to allow the system to work in volume drilling regime. Such systems are unique with respect to terrestrial applications, but there exist some applications dedicated for specific research missions realized in space. The Mars Exploration Rovers, Spirit and Opportunity, are both equipped with a Rock Abrasion Tool (RAT) [2]. Similarly, the Curiosity rover is equipped with a drilling system, however, all of them are rather abrasive tools. The system in Curiosity rover has a grinding wheel, with a diameter of 16 mm, designed to remove dust and weathered surface layers to a depth of 5 mm [3]. In North America the most advanced drilling systems are built by Honeybee Robotics. In the frame of experiment called MARTE (Mars Astrobiology Research and Technology Experiment) the system operated for 23 days in simulated mission conditions and the drill reached 6 m according to [4]. One of the most successful devices that were utilized is Apollo Lunar Surface Drill [5]. This device that possessed power exceeding 450 W was utilized to make holes up to 3 m deep. Nevertheless, the device was operated by astronauts. The deepest hole produced by an unmanned drill was done on the Moon, where Luna 16 robotic probe managed to collect specimen from 350 mm depth [6]. Laboratory tests for deep drilling were performed by several research teams. The system presented in [7] has the potential to drill a hole up to 2 m, however the size is significantly bigger. In ESA, a drilling system is developed that would be installed on upcoming ExoMars mission [8]. The system utilizes a 6-wheel rover equipped with a core drill designed to extract samples from the depth up to 2 m. The majority of presented drilling systems utilize large scale rover featuring 6 wheel designs that can host various equipment for investigation of an unknown environment. In this paper, we focus on an ultralight mobile platform for an autonomous drilling system that would be able to produce holes reaching depth of 2 m. Firstly, general overview of the system is presented and major functionality of subsystems is outlined in each phase of the entire mission. Next, the mobile robotic platform is described in details, along with multi-body simulations, mathematical models and stability analyses. Finally, tests of the prototype are presented.

Copyright  2015 by M. Ciszewski et al. Published by the EUCASS association with permission.

M. Ciszewski, T. Buratowski, A. Gallina, K. Seweryn, W. Teper, A.J. Zwierzyński, T. Uhl

2. Overview of the ultralight mobile drilling system The main subsystems (i.e., mobile robot – MR, support module – SM and drilling subsystem – DR) of the mobile drilling system are shown in Figure. 1. The mobile robot is a light (4.5 kg) subsystem which provides mobility in an unknown terrain. The mass of UMDS as a system is 22 kg. It was designed to operate at the distance not exceeding 100 m from the base station. Drilling can be performed in regolith or rock with compressive strength below 35 MPa and depth not exceeding 2 m. The system is capable of drilling a single hole in the current version, however, multiple bore hole drilling is the aim of future development.

a)

b)

Figure. 1. Mobile drilling system: overview of the subsystems: MR – Mobile Robot. SM – Support Module, DR – Drilling subsystem, b) drilling mission scenario Drilling and collecting samples is a very complex process, featuring multiple operations. The mission is divided into 8 phases that are ilustrated in Figure. 1 b. In the first part of mission, the rover autonomously gets to the designated drilling point. In the next phase, the Support Module unfolds from transport position to the anchoring position and afterwards the anchoring module activates. Anchoring process is a crucial step in the mission, indispensable for properly performed drilling process. When the system is successfully anchored, the subsystem SM is positioned at an angle that allows to drill parallel to the gravity vector. The SM is a planar manipulator with base rotary joint that allows to prepare the borehole at angle in range -15° to 45° relative to the normal of the surface. Next phase of the mission is the drilling process in which the clamping force and vertical movement of the tool is provided by SM. In order to ensure the stability of the borehole during penetration of loose layers of regolith, the DR extends tubular C-shape coilable tape. The drilling module (DR) is able to collect a 50 mm length core at a time, due to dimensional restrictions of the system. Additionally, cuttings from the drilling process are being discharged from the borehole bottom and are collected in a dedicated container. Because of that reason the drilling process on 2 m depth borehole is divided into 40 small steps. After each 50 mm progress in depth, the DR is pulled to the surface, where the sample is inserted to a special vessel. Subsequently, the DR is moved by the SM to the position where the container filled with cuttings is being emptied. Operations of placing a sample into the sample container and removal of cuttings are followed by return of the SM to drilling position and afterwards, start of the next core drilling. After collecting samples from the maximum depth, according to actual design, the SM leaves the anchoring module and the drum with C-shape tabular tape, used for securing the borehole . Afterwards, the SM is folded to a transport position. Finally, the rover can return to the start point and deliver the samples.

2

MOBILE PLATFORM FOR A DRILLING SYSTEM

3. Mobile robotic platform 3.1 Mechanical model MR is a platform intended for transport of SM and DR modules and has dedicated space for power supply unit, on-board computer and drive controllers. In Figure 2 we may see a general view of a CAD model of the mobile platform. The MR has four wheels powered independently by brushless DC motors, equipped with integrated encoders and gear transmissions. The 4-stage planetary gear heads used in the design have reduction ratio of 850:1. This design provides high torque along with mass reduction of the drives. The wheels have special design, featuring ultra light aluminum rims with integrated hubs on which titanium hoops with inclined grousers are mounted. Left and right pairs of wheels have opposite grouser angles that were optimized for climbing obstacles of various shapes. A suspension system enables independent setting of extension of the left and right rocker of the rover with respect to the ground by usage of two self-locking linear drives as shown in Figure 2 a. The rockers have rigid, light-weight structure, adapted to minimize thrust force of linear drives. Furthermore, the rockers are mounted on a balancing mechanism that ensures symmetric extension with respect to shafts assembled to central differential mechanism and equalizes pitch angle of the rover.

a)

b)

Figure 2. CAD model of the mobile robot (MR) platform: a) with extended rockers. b) with retracted rockers The suspension system provides adjustment of height of the MR platform by 100 mm or can compensate roll angle of the rover up to 13.3° as presented in Figure 3 a. Roll angle compensation is an important feature in terms of stability of the entire structure during motion and drilling operations as depicted in Figure 3 a. The mobile robot may negotiate transversely inclined surfaces by angles up to 30°. The robot is equipped with a central differential mechanism that connects left and right side of the suspension, averaging pitch of the pairs of rockers by ±10° and enabling smooth operation on uneven surfaces (Figure 3 b). This is a well-known solution that significantly improves the rover mobility [9]. It is important to indicate that an anchoring system mounted on the Support Module assures that forces and torques occurring during drilling are only partially transferred to the MR.

a)

b)

Figure 3. MR platform adjusting to various terrain: a) compensation of roll angle, b) negotiating uneven terrain by usage of differential mechanism. Optimization of energy consumption showed that velocity of the rover should not exceed 50 mm/s. Compensation of roll angle and height adjustment by rockers is a slow process and takes approximately 10 minutes between limiting values of extension.

3

M. Ciszewski, T. Buratowski, A. Gallina, K. Seweryn, W. Teper, A.J. Zwierzyński, T. Uhl

3.2 Kinematic model of the rover Kinematic models of the complete suspension system of the mobile platform are currently being developed. Fundamental element that must be analyzed is the rocker suspension. In this case, the suspension consists of two rockers with wheels that allow height adjustment of the rover. Thus, we may note that the distance rca presented in Figure 4a is variable. The main objective of kinematic modeling is to obtain the required angular velocity ratios between the wheels to maintain no slip conditions during entire motion of the rover [10]. Inappropriately selected wheel velocities would violate kinematic constraints.

a)

b)

Figure 4. Kinematic model of the rover: a) coordinates of important points of the rover, velocities of important points of rocker suspension. In the rover, front and rear rockers are considered to be rigid bodies, therefore, the distance between wheel axle (point a) and frame axle (point e) is constant. We may write the expression using (1). (1) Analogous equation can be written for the rear rocker. To determine velocity of the rover frame, we need to take into consideration, the composition of velocities that result from wheel contact with uneven ground. In Figure 4b, velocities of important points of the structure were presented, whilst the rover is in motion on an uneven surface. At this stage we consider only one half of the suspension. We may write that the relative velocity between the two wheels along the direction of line joining centers (points a, c), when height of the suspension is not altered, is equal to zero. (2) We express the velocities of wheels 1 and 2 in vector form: vector joins centers of these wheels. When we substitute the values of angular velocities and wheel radii obtain constant velocity ratio expressed by (3):

, and the distance by cartesian coordinates, we

(3) Both wheels are connected to the rockers of constant length, therefore, the velocity of point e, attached to the rocker frame ( ) can be expressed using (4) and (5): (4) . Distance vectors and represent lines of the front and rear rocker and assembly with respect to the point e.

4

(5) is the angular velocity of rocker

MOBILE PLATFORM FOR A DRILLING SYSTEM

When we perform substitution of velocities and distance vectors and solve (4) and (5), we obtain angular velocity ratios by (6) and dependencies of velocity components in the coordinate system expressed by (7) and (8). (6) (7) .

(8)

The presented kinematic model allows calculation of position and velocity of the rover frame, with assumption that rockers on both side are symmetrically oriented. In case when the central differential mechanism operates, we need to take into consideration additional parameters that alter orientation of the frame on a more complicated terrain.

3.3 ADAMS multi-body simulations A multi-body model of the rover was built in MSC.Adams. Although the components’ geometries are simplified, their sizes and masses are very close to the ones of the physical model and the full mobility of the rover suspension is included. The presence of the SM and DR module is taken into account by adding a concentrated mass. A solid-tosolid wheel-soil contact model was employed. The contact model reduced the contact to one point and calculates the normal and friction forces exchanged between the two bodies. The friction force is calculated by a classical Coulomb model, that is similar approach to [9]. Multi-body simulations aim at predicting torque and normal force acting at the wheels’ hub for different scenarios and with the rover traveling on hard soil. The torque and force under considerations are two critical parameters in the rover design. Two simulation scenarios are presented: a) Rover negotiating an obstacle (Figure 5a) b) Rover travelling on an inclined rough surface ( Figure 5b - right)

a)

b)

Figure 5. Analyzed scenarios in multi-body simulations: rover negotiating an obstacle (left), rover travelling on an inclined rough surface (right) 3.3.1 Rover negotiating an obstacle The obstacle is represented by a semi-cylindrical bump. The bump has height which is half of the rover’s wheel radius. This is in agreement with a project requirement stating that the rover shall negotiate an obstacle whose height is half of the rover’s wheels diameter. The surface on which the rover travels is a 10° slope according to the project requirements. The friction coefficient of the contact model was set to a value large enough to make the rover overcome the obstacle without significant slip. The torque at the four wheels hubs have been calculated by MSC.Adams, assuming that the wheels rotate at constant velocity. The diagrams of the measured quantities are shown in Figure 6.

5

M. Ciszewski, T. Buratowski, A. Gallina, K. Seweryn, W. Teper, A.J. Zwierzyński, T. Uhl

a)

b)

a)

b)

.

Figure 6. Loads on right wheels during simulation with passing of an obstacle: a) front right wheel radial force, b) front and rear right wheel moment For very short instants of time a value of 4.5 Nm is reached. This value is also the maximum torque for the selected drive unit (motor + planetary gearhead) recommended by the unit manufacturer. Radial forces at the wheel hub are also measured in the simulation. They show values constantly lower than maximum allowable radial force. 3.3.2 Rover driving on an irregular surface In this analysis an irregular surface is modeled as a random field with specified correlation length. The harshness level is such that the height from top to valley of the dunes reaches the rover’s wheel diameter. As in the previous case, the calculated torques and forces are within the motor specifications. The results are shown in Figure 7. The scatter present in the measured signals is a result of numerical problems and the fact that the structure is completely rigid as no spring or damper is located at the joints. Therefore, we may assume that the torque values do not exceed 2.2 Nm and the forces are lower than 120 N.

a)

b)

Figure 7. Loads appearing on right wheels during simulation on an inclined rough surface: a) front and rear right wheel moment, b) front right wheel radial force

3.4 Stability of the mobile robot Support Module and thus resultant position of the DR system have major influence on the robot stability. During robot motion, the SM module should be set to a compact position to minimize the risk of overturning. Stability may be also increased by lowering height of the robot with usage of linear drives mounted in the rockers. Nevertheless,

6

MOBILE PLATFORM FOR A DRILLING SYSTEM

the drives are indispensable to provide obstacle passing capabilities and give a possibility to increase ground clearance of the robot from 90 mm to 200 mm. Longitudinal stability of the mobile platform is mainly dependent on slope inclination. Due to alteration of position of the center of gravity in different configurations of the mobile robot, we should distinguish two cases: when the robot moves uphill and downhill. Specimens gathered by the DR module that are carried in a container located relatively high with respect to the robot platform may have a total mass of 4 kg, that increases total weight of the device by 20 % and shifts position of the center of gravity. In order to provide optimal operation of the SM manipulator in terms of overall system stability, expressions (9) and (10) were utilized.  (9)  .

(10)

where: angle of rotation of arm I with respect to the SM base angle of rotation of arm I with respect to arm II Value of angle exceeds when the robot has to perform drilling operation on an inclined surface. In order to determine the influence of extension of the SM manipulator on the longitudinal stability of the robot, positions of center of gravity were acquired for the prepared CAD model. Horizontal shift of the center of gravity was investigated and it is shown in Figure 8. We may observe that the change reaches respectively 140 mm or 170 mm for low and high MR platform settings. The ability of the robot to negotiate inclined surfaces was the main objective of the previously performed analyses. Basing on the results, it was decided that the MR platform should move in the low position.

Figure 8. Horizontal shift of center of gravity with respect to MR axis.

In Figure 9a we may see the robot positioned on an inclined surface, climbing a hill. The theoretical longitudinal stability is maintained, based on projection of the center of gravity on wheel base, however the inclination of 40° may be impossible to climb due to excessive wheel slippage. Exact values of inclination may only be determined by experiments carried out on a prototype. Mechanical structure of the mobile platform provides a possibility to compensate roll angle (sideways tilt) during motion and to level the device for drilling as presented in Figure 9 a. The platform can be maintained in a horizontal position for angles in range ±13.3° by usage of linear drives located in robot arms. The theoretical roll angle during motion of the robot can reach 30°, basing on the position of center of gravity, however we have take into

7

M. Ciszewski, T. Buratowski, A. Gallina, K. Seweryn, W. Teper, A.J. Zwierzyński, T. Uhl

consideration the slippage of wheels on granular surface such as gravel or sand. Maximum roll angle correction for drilling operation is depicted in Figure 9c.

a)

b)

c)

Figure 9. Stability assessment of the mobile platform: a) uphill motion, b) roll angle compensation during forward motion, c) roll angle correction for drilling.

4. Environment exploration with usage of the Wavefront algorithm Trajectory generation of a robot in space consists of definition of position and orientation in reference to the base coordinate system. Navigation of the robot is based on two concepts. First concept, related to Earth exploration consist of data from Inertial Measurement Unit (IMU) and Global Positioning System (GPS). Additionally, measurements from encoders mounted on wheel drives may be used as complementary information when it is possible. The second concept of navigation intended for extraterrestrial environments would be based on visual topological Simultaneous Localization and Mapping (SLAM). Such techniques are related to identification of some objects in environment to estimate robot position and orientation. Visual data of the objects can be collected by satellite imagery. There are several ways to plan robot trajectory. In our project Wavefront algorithm is utilized. The algorithm in the case of this rover is implemented using a specific approach. It is assumed that when a rover moves on a rough terrain such as gravel or sand, non-holonomic constraints given by equation (11) are not met. (11) where:

The area quantized for calculations in Wavefront algorithm must be decomposed with usage of constant cluster size grid (Constant Grid Decomposition), stretched at height corresponding to the maximum size of obstacle present in the area of exploration. The result is a projection of the grid on an uneven surface as presented in Figure 10.

Figure 10. Grid used in Wavefront algorithm adapted to exploration environment of the rover. The principle of operation of the Wavefront algorithm is presented in [11]. First, obstacles are marked with a 1 and the destination point is marked with 2. One can optionally surround the entire area of interest with squares numbered

8

MOBILE PLATFORM FOR A DRILLING SYSTEM

by 1 as well to tell the robot to avoid those squares, and/or "expand" the size of the obstacle to avoid collisions due to dead-reckoning errors. After those operations the work environment will look like the one depicted in Figure 11.

a)

b)

Figure 11. Trajectory generation with usage of the Wavefront algorithm: a) initial assignment of clusters, b) trajectory generated using Wavefront algorithm In order to create the "wave" of values, the calculations must start at the destination, and a distance of 3 must be assigned to every square adjacent to the goal. Then a distance of 4 is assigned to every square adjacent to the squares of distance 3 (Figure 11a). The operation should be continued until the start point is reached. When the first stage is completed, the numbers need to be followed in reverse order, and obstacles marked by 1 should be avoided. A possible path has been marked with red color in Figure 11b. The goal and start point are labeled as 2 and 16 consecutively. This is just one of the many paths that can be used to reach the goal, and any path that follows the descending numbers correctly will be acceptable.

5. Tests of the prototype A prototype of the mobile drilling system was manufactured, according to the CAD models. In Figure 12 the mobile platform subsystem is depicted in position with extended rockers during motion on rough and flat surface.

a)

b)

Figure 12: Prototype of the mobile platform during motion on different surfaces: a) rough terrain, b) flat surface. Motion tests of the MR prototype were performed to analyze conformity of the mechanisms with assumptions made at the design stage. In Figure 12 we may observe the mobile platform during motion in an environment with obstacles.

Figure 13: Mechanical integration of MR, SM and DR subsystems, transport position.

9

M. Ciszewski, T. Buratowski, A. Gallina, K. Seweryn, W. Teper, A.J. Zwierzyński, T. Uhl

The experiments showed that blades situated on the wheel hoops are so efficient, that the robot may negotiate obstacles with height exceeding wheel radius without difficulty. The high level of mobility was achieved as well owing to the central differential mechanism. Finally, an initial integration of all subsystems was performed. Figure 13 shows the complete system assembled in transport position. At this stage only mechanical integration was performed. The system was not equipped with power supply unit and control electronics. Laboratory tests of the subsystems were performed as separate experiments.

6. Conclusions and further work The mobile robotic platform presented in this paper represents an innovative approach in building ultralight mobile drilling devices. The main objective of design of particular subsystems of the device was to assure functionality of drilling rigs that are significantly bigger and heavier, whilst maintaining the lowest possible mass. Suspension system of the mobile platform proved to be efficient in negotiating obstacles and uneven terrain. The design aids such as multi-body simulations and CAD analyses provided indispensable resources for structural optimization. Future development would focus on final integration of all subsystems and installation of equipment, including power supply, sensors and controllers that are crucial for autonomous operation of the mobile drilling system.

References [1] D. Sears, C. Allen, D. Britt, D. Brownlee, A. Cheng, C. Chapman, B. Clark, B. Drake, R. Fevig, I. Franchi, A. Fujiwara, S. Gorevan, H. Kochan, J. Lewis, M. Lindstrom, K. Nishiizumi, M. Race, D. Scheeres, E. Scott, G. Taylor, and H. Yano, Near-earth asteroid sample return, in The Future of Solar System Exploration (20032013) - First Decadal Study contributions, ser. Astronomical Society of the Pacific Conference Series, M. Sykes, Ed., vol. 272, Aug. 2002, pp. 111–140. [2] S. P. Gorevan, T. Myrick, K. Davis, J. J. Chau, P. Bartlett, S. Mukherjee, R. Anderson, S. W. Squyres, R. E. Arvidson, M. B. Madsen, P. Bertelsen, W. Goetz, C. S. Binau, and L. Richter, Rock abrasion tool: Mars exploration rover mission, Journal of Geophysical Research: Planets, vol. 108, no. E12, pp. n/a–n/a, 2003. [Online]. Available: http://dx.doi.org/10.1029/2003JE002061 [3] J. Grotzinger, J. Crisp, A. Vasavada, R. Anderson, C. Baker, R. Barry, D. Blake, P. Conrad, K. Edgett, B. Ferdowski, R. Gellert, J. Gilbert, M. Golombek, J. Gomez-Elvira, D. Hassler, L. Jandura, M. Litvak, P. Mahaffy, J. Maki, M. Meyer, M. Malin, I. Mitrofanov, J. Simmonds, D. Vaniman, R. Welch, and R. Wiens, EnglishMars science laboratory mission and science investigation, EnglishSpace Science Reviews, vol. 170, no. 1-4, pp. 5–56, 2012. [Online]. Available: http://dx.doi.org/10.1007/s11214-012-9892-2 [4] H. N. Cannon, C. R. Stoker, S. E. Dunagan, K. Davis, J. Gomez-Elvira, B. J. Glass, L. G. Lemke, D. Miller, R. Bonaccorsi, M. Branson, S. Christa, J. A. Rodriguez-Manfredi, E. Mumm, G. Paulsen, M. Roman, A. Winterholler, and J. R. Zavaleta, Marte: Technology development and lessons learned from a mars drilling mission simulation, Journal of Field Robotics, vol. 24, no. 10, pp. 877–905, 2007. [Online]. Available: http://dx.doi.org/10.1002/rob.20224 [5] N. Navarathinam, R. Lee, K. Borschiov, and B. Quine, Northern light drill for mars: Performance evaluation, Acta Astronautica, vol. 68, no. 7-8, pp. 1234 – 1241, 2011. [Online]. Available: http://www.sciencedirect.com/science/article/pii/S0094576510003978 [6] NASA. Luna 16. http://nssdc.gsfc.nasa.gov/nmc/masterCatalog.do?sc=1970-072A. [Online]. Available: http://nssdc.gsfc.nasa.gov/nmc/masterCatalog.do?sc=1970-072A [7] K. Zacny, G. Paulsen, Y. Bar-Cohen, L. Beegle, S. Sherrit, M. Badescu, B. Mellerowicz, O. Rzepiejewska, J. Craft, S. Sadick, F. Corsetti, Y. Ibarra, X. Bao, H. J. Lee, and B. Abbey, Wireline deep drill for exploration of mars, europa, and enceladus, in Aerospace Conference, 2013 IEEE, March 2013, pp. 1–14. [8] M. Van Winnendael, P. Baglioni, and J. Vago, Development of the esa exomars rover, in Proc. 8th Int. Symp. Artif. Intell., Robot. Automat. Space, 2005, pp. 5–8. [9] F. Zhou, R. E. Arvidson, K. Bennett, B. Trease, R. Lindemann, P. Bellutta, K. Iagnemma, and C. Senatore, Simulations of mars rover traverses, Journal of Field Robotics, vol. 31, no. 1, pp. 141–160, 2014. [Online]. Available: http://dx.doi.org/10.1002/rob.21483 [10] S. Parakh, P. Wahi, and A. Dutta, Velocity kinematics based control of rocker-bogie type planetary rover, in TENCON 2010 - 2010 IEEE Region 10 Conference, Nov 2010, pp. 939–944. [11] T. Buratowski, B. Dabrowski, T. Uhl, and M. Banaszkiewicz, The precise odometry navigation for the group of robots, Schedae Informaticae, vol. 19, pp. 99–111, 2010.

10