The demonstrator and imitator robots

This post describe another project that I have assigned to our students at Aalesund University College while teaching mechatronics. In particular, this project addresses the problem of body mapping in robotic imitation where the demonstrator and imitator robots may not share the same embodiment (degrees of freedom, body morphology, constraints and so on).

Main idea

The main idea consists in realising a flexible architecture to control different imitators with dissimilar kinematic structure by using the same demonstrator robot. The challenge in designing such a system consists of implementing a flexible mapping method between different kinematic models.

The demonst...

Possible approaches

• End effector mapping. Using the forward kinematics, the demonstrator end effector position is calculated. The position vector can be scaled to be adapt ed to the imitator working space. Then the imitator joint parameters can be calculated using the corresponding inverse kinematic model. This approach requires the inverse kinematic model of each arm to be controlled.
• Joint-to-joint mapping. This approach is based directly on similarities between the demonstrator and the imitator. Some joints can be coupled or simply locked in order to have the same number of input and output parameters.
• Neural network. The demonstrator working space is mapped on the imitator working space using a neural network.

Neural Netw...

Current implementation

For the current implementation, I gratefully acknowledge the contribution of the students Longva Ole Martin, Skinnes Marius Pieroth, Nilsen Christian Steiner, Strand Jan Ove.