M.Sc Student Aranyi Paz Optimization of a Hybrid Robot's Weight Lifting Ability Department of Mechanical Engineering Professor Emeritus Yehoshua Dayan Professor Yizhar Or Abstract

In general, robotic manipulators can be divided into two main groups. The first group is serial manipulators, and the second is parallel manipulators. However, there is an additional group called hybrid manipulators, which combines the benefits of both groups. Usually, a hybrid manipulator consists of a parallel base on which a serial arm is mounted, although an inverted arrangement is also possible. It is important to note that serial and parallel manipulators have been studied thoroughly, in numerous researches and were compared with hybrid manipulators as well. None of the researches who dealt with hybrid manipulators ever considered the weightlifting capabilities of these robots.

In this work we consider a specific hybrid robot and analyze its ability to lift weight, in a minimal amount of time. The final aim, though, is to maximize the weight lifting ability of this robot, while satisfying constraints of physical bounds (e. g., maximum allowable motors’ torque, current or power).

It is important to note that the serial part of our robot has four degrees of freedom (4 DOFs), and the parallel part has 3 DOFs. Such a structure may (vaguely) resemble a weightlifting athlete.

First, we calculated, analytically, the kinematics and dynamics of the serial and parallel parts separately. For the dynamics of the parallel part, we used constrained Lagrange equations. Next, we verified our calculations with two numeric simulations: one based on the equations we formulated, and the other based on MATLAB's “SimMechanics” simulator. Obtaining the same results from two independent methods has proven that the mathematical analysis is correct.

Following the verifications of the kinematics and dynamics for both parts of the hybrid robot, we connected the two parts in the “SimMechanics” simulator, and gained a fully verified simulator of the entire hybrid robot. We then used Matlab’s constrained optimization function, Fmincon, to find the optimal path in which a weight can be lifted in minimum time while satisfying physical constraints on the motors’ power, while avoiding configurations of kinematic singularity.

Finally, in order to evaluate the contribution of the parallel base, we compared the obtained solution to the solution of the same problem in which only the serial part is participating. The results of this comparison showed that the hybrid robot has 9.57% better performances than the serial robot, and the best performance in both cases was obtained from paths which were similar to those applied by weightlifters, when they use one arm to lift the weights.