Colloquia

Summary

In the field of agro-robotics, much research is performed on ‘soft’ or adaptable grippers, for example to harvest certain fruits or grasp other food products. The focus of this thesis is an underactuated, flexure-based gripper consisting of two fingers with two phalanxes each. The objective of  this research is to analyse the mechanics of such grippers and to optimize the range of object sizes that can be grasped, while meeting constraints on the stress and the support stiffness.

In the approach, different Spacar models are used to analyse the geometric behaviour of the gripper and to obtain values for the stress and the support stiffness. By limiting the analysis to five-point grasps and by separating the elastic and geometric behaviour of the compliant gripper, the dimensionality of the problem is reduced to make the procedure less computationally expensive.

An optimization algorithm is proposed, which is able to optimize 39 parameters of the gripper in typically a few hours. The proposed algorithm is compared to a Nelder-Mead algorithm and is used to optimize two grippers for two different use cases.

Two experiments are conducted to validate Spacar models and the force calculations. A natural frequency analysis with Spacar model shows agreement with a measured frequency response for the frequency in the drive direction and for the first three support directions. The required actuator forces for a five-point grasp are also validated and only show small deviations from the calculated values.

A Spacar model with reduced complexity typically shows less than 5% deviation in forces from a detailed Spacar model. The maximum deviation equals 11.7%, for which suggestions are given to reduce this. Therefore, the principle of separating the elastic and geometric behaviour is also concluded applicable. The results from the experiments are in good agreement with the Spacar models and other calculations, which are concluded representative.