Colombia is the biggest producer of palm oil in Latin America. The production of palm oil alone makes up for 6% of the agricultural gross domestic product of the country. However, growing oil palm trees is a risky business since this kind of tree only starts producing its fruits when it reaches the age of three years, and during that time the trees are very prone to contract several diseases, which may completely damage the tree. The most common, and devastating, disease that can affect oil palm trees is the bud rot disease, which destroys the young tissue of the palms. Experts indicate that when the disease is detected in its early stages, a simple pruning of the affected area together with a chemical control may be enough to fight the disease. Therefore, there is a strong need for continuous monitoring of the trees. Some of the main challenges for the monitoring of oil palm plantations are the large areas that must be covered, the difficult environmental conditions with high temperatures and relative humidity, and the irregularities of the terrain. One way to make this monitoring task more cost effective, as compared to doing it only by means of human workers, is with the assistance of mobile robots. There are different kinds of mobile robots that have been used in precision agriculture applications such as tracked robots and wheeled robots. Although these robots offer several advantages such as simple mechanics and good payload capacities, in many cases they may not be the best choice, for example in very uneven or muddy terrain. This work proposes the study, design and construction of a spherical robot, intended for remote monitoring of oil palm plantations. Spherical robots provide an interesting alternative for monitoring of plantations thanks to their excellent maneuverability and adaptability to various terrain conditions. Spherical robots can rotate in place to change direction, can move over muddy terrain and they can even float and move over water. They also have some disadvantages such as difficulties to pass over an obstacle and lower payload capacity. This work was carried out in 4 stages, the first one was the selection of a locomotion architecture that adapts to the proposed design conditions. Then, the mathematical model of the robot was obtained, which allowed simulating the dynamic behavior of the robot and designing a control system to guarantee the stability needed for the guidance of the robot. Finally, the prototype was built and validated.