Gezici mikro ay robotunun hareketlilik, manipülasyon ve görüntü işleme sistemleri tasarımı

The research area for the lunar micro-rovers (LMR) is very extensive such as navigation, obstacle avoidance, path planning, rover platform design, field-testing, motion kinematics, slip models, and so on. Since there was no rover platform to work on within our faculty, the project started with designing a micro-rover platform to implement and test certain technologies not only during this project but during future projects as well. Therefore, the thesis presents a simple and versatile rover platform. According to the future missions, the geology of the Moon, potential resources and the distribution of them are gaining attention internationally since the Moon itself is an important planetary body for exploration by both human and robotic missions. For this reason, lunar micro-rovers must be designed to make scientific research, which can support these missions by investigating the lunar environment and mission parameters. Multiple scientific areas such as planetary sciences and geology, plasma physics, astrobiology, astronomy, space science will be benefiting from the scientific outcomes of the future lunar missions. Therefore, performing scientific researches with an LMR and addressing multiple scientific goals are essential objectives. Landing on the lunar surface can be expressed as another challenge; however, it is not one of the topics of this project. Many studies suggest orbital missions need to be coupled with surface investigations in order to validate the interpretation of data sets. In addition, surface data will show suitability for manned missions as well as validity of the models on the significant surface characteristics. The mission is proposed to land on a determined area in South Polar region of the Moon since longer periods of solar illumination is available by reason of the low sun inclination is coupled with topographic highs. The main objective is to obtain a surface sample from the Moon and store it within the micro-rover to study in mini-laboratory, which is expected to be on the top of the LMR. The sample is desired to have a volume between 1-3 cm3; therefore, the robotic arm tool is designed based on a handling mechanism with a surface geometry, which enables to tip tool grab a solid rock sample from the ground; however, this tool can also be replaced with a drill mechanism for subsurface sampling. The main interest of the research is chosen as investigating the ejecta areas on the lunar surface containing both meteor samples as well as Moon's own material. The rover must be able to move to points of interest to make measurements with its onboard instruments and transmit data. An LMR must be lightweight with high mobility, resistant to high radiation, extreme temperature variations, lunar dust etc. In design process, the most important considerations were mobility and weight limitations. The mass and inertia properties are derived from Solidworks after material selection. Most of the time in an LMR, power is generated using the solar panels and stored in the batteries while onboard computer would be performing multiple task and the reset capability in emergencies. All subsystems of LMR have to be protected from high radiation, vacuum and extreme temperature variance. First, a small lightweight rover provides significant savings in the launch and lunar landing thruster capabilities; therefore, size and weight have to be limited. For this reason, there is not enough room for complex ideas, so everything must be as simple as possible. In addition, the energy limitations require actuators to be chosen those consume very little power since there will be limited electrical power from solar panels. Lunar surface temperature varies from -240oC at night to 121oC during the day, and micro-rover will be operating in micro gravity environment. Onboard computer and battery is expected to keep the electronics warm in order to operate in that low temperature. The body structure is designed to have low volume; however, it must provide sufficient space for the onboard computer, scientific instruments, actuators, cameras, sensors etc. The uneven lunar surface requires LMR to sense the obstacles as well as other hazards in order to get closer to the crater region, and it must be able to identify interesting sample candidates during the motion. Therefore, the primary technologies to be implemented within the LMR are chosen as local mapping, manipulation and sample handling, and path planning. The surface sampling activities will be commanded from Earth during the communication windows while the instruments within the mini laboratory analyze samples, and data of the environmental properties such as radiation, ionization, temperature will be gathered as well. The formal mission lifetime is expected to be couple of months if the LMR survives the lunar night that is approximately 13 days. The first phase of the mission will contain primary scientific activities and measurements before the lunar night period; therefore, if the LMR does not survive the lunar night, it will gather and transmit data before it malfunctions. At the beginning, the design of the lunar micro-rover is chosen as a 6-wheel system with a rocker-bogie suspension that has a mechanism as passive rotary joint to attach two links called rocker and bogie. As wheels are free to move up and down on an uneven terrain with the rotary joint, each wheel carry the weight of the LMR equally, but it is much heavier than a 4-wheeled suspension, and each extra actuator means more energy consumption and higher hardware requirements. By only giving input for the links' motions, the joints' torques are derived for the given system, and for the poses, which robotic manipulators cannot reach, the singularity of the Jacobian matrix is observed. The computational complexity of each iteration remained constant, and it grows in a linear fashion with the number of joints increases. In addition, a closed-form dynamic model is obtained which is the same as the one obtained by using DH parameters. The inertia tensor matrix shows all mass and inertial properties of the robotic system, and it is very important to be derived for understanding the dynamics of the system since the inertia of the manipulator has to be surpassed in order to move the links. Therefore, the effect of Moon and Earth's gravity ratio is observed in the torques. Since the robotic arms' motion is determined to be slow, Coriolis Effect is not strong. Moreover, the profiles of the both torque graphs seem to be very similar; however, the magnitude of the torque's ratio is very close to the ratio of the gravity, which is approximately 1/6. Active laser triangulation system operates by sensing the reflected red laser light from a surface; however, the ambient light variations and the reflectance properties of different surfaces are the main challenges of this system. The ambient light variations can be overcome by using optical filters according to the wavelength of the laser emitter to be used in outdoor environments more accurately. In addition, active laser triangulation system is desired to be low-cost, low power consuming while having high response speed and high sampling frequency, and it will be used for detecting hazards and mapping the rough terrain. In addition, a passive triangulation system such as stereo-based planner is a reliable choice, and it will be added to the LMR in the future; however, it cannot detect craters and shadow areas. When the LMR uses its active laser triangulation system and captures a new image, the LMR's field of view is updated in the image space. The information of outside of the LMR's field of view is kept in memory, and the map data are implemented in the algorithm as 2D grid that shows the coordinates of the nodes and the height values corresponding to those nodes. Then, the updated image space is used to renew 3D main map. The LMR's field of view will be updated on the main map, and the rest information will be preserved in the memory unchanged. If the captured image is used to create a local map, and then it is used to update the current map, the new nodes cannot match the main map's nodes exactly. Therefore, the image space must be updated first in mm metric, and then the main 3D map must be updated. Finally, greedy search algorithm and A* search algorithm are implemented in the main algorithm to compare their path planning performance. Different tasks are given to the path planning algorithms, and if there are no obstacles or hollows, which the robot cannot overcome, both algorithms found the reasonable path. On the other hand, the complication level is increased for the tasks, and A* search algorithm found the optimal path each time, if it exists at all. Therefore, A* search is chosen as the path planning algorithm instead of greedy search. The next step for this project is to build prototype for verification of the performance and to have a micro-rover platform to work on within our faculty.