Good Vibes
Motivation
Vibration is normally an undesired effect in Mechanical Systems. It can cause undesired motions and forces, wear of mechanical parts and in extreme cases failure of parts or the entire system when resonance occurs. However, what if we take advantage of vibration in some specific scenarios, such as repetitive tasks performed by robots in industry? In these tasks, the actuators must continuously accelerate and decelerate the links. Therefore, energy is typically transferred from the motors to the links during the acceleration phase to increase their kinetic energy and is then dissipated during braking. However, this is highly inefficient, because the energy could be recovered. One strategy for this recuperation is to add springs, to convert the kinetic energy of the links into potential elastic energy to be used in the next cycle. Thus, the robot would naturally vibrate between the two positions it has to achieve for example for picking and placing objects.
Goal
The objective of this project is to study the feasibility of adding elastic elements to a parallel robot and exploiting the natural dynamics of the ensuing system to reduce the energy consumption and the control effort during typical pick-and-place tasks. An estimation of the performance of the modified system is also expected.
Approach
Two steps are contemplated: simulation of the system using multibody models and experimental validation. Specifically, the multibody models are used for several purposes. First, the concept is proven by finding springs that match the free-vibration of the system with the desired nominal task, such that the robot naturally oscillates between the nominal positions for pick and place. Then, optimization algorithms are used in combination with the multibody model to find the optimal trajectory that the robot must follow between any two given positions, even if they differ from the nominal ones, to minimize the energy consumption. Next, a controller is added to guarantee that the manipulator follows the optimal trajectory in the presence of disturbances such as dissipative forces or uncertainty in the mechanism properties. Finally, the performance of the robot with springs is estimated and compared with that of the original robot without springs. An experimental validation of the methods and the results is required.
Partner
Universidad de los Andes (Colombia)
Funding
DAAD - German Academic Exchange Service
DFG Deutsche Forschungsgemeinschaft
AROP – Advanced Research Opportunities Program (RWTH)