Robotic grippers using direct drive motors
In the field of robotics, creating grippers that can perform tasks similar to the capabilities of a human hand, is still a challenge yet to be solved.
To get human-like performance of a mechanical hand is a difficult task, and many different gripper solutions exists, each have their own benefits and drawbacks.
In this masters project we seek to explore if and how direct drive (torque-) motors can be used to create and control one or more fingers. Traditionally most motors used in robotics has been geared motors working at high numbers of revolutions per minute. Being geared, the actuators can achieve high precision and efficiency in repeated tasks. However as robots are moving out of industrial facilities and into the human and natural environment, this calls for robots with different properties and goals. There are tasks where high torque, speed and relative low momentum is more desirable than exact position. Effectively picking up an egg from a table with a simple, general purpose gripper may be one such application, sorting clean laundry from a washing machine would be another.
To achieve this type of performance there are certain steps that we would like to explore.
One task we aim to explore is to explore how to control the position of a mechanical finger utilizing one or more direct drive motors, possibly in conjunction with other actuators. Machine learning algorithms will be applied to tune the system.
An other relevant task is to study how loosing grip (such as if someone tries to pull an object out of your hand) influences current drawn from the motor, and thus create a system that can learn and adjust the torque used to hold on to objects without crushing them. Machine learning can be applied to both detect when something is slipping, and finding the proper torque or force to stop the object from slipping within the required time frame to avoid loosing the object.
Secondary tasks may be to suggest and come up with new innovative designs or improvements that may benefit or be solved using machine learning in combination with direct drive motors.
The project will be done in cooperation with Halodi Robotics, and the student(s) taking part in this project, will spend some time at their facilities in Moss, and have an external supervisor from Halodi. At the time of writing this project description, we do not have a name of the external supervisor, but we do have confirmed their interest and support of this project.
Halodi robotics is a relatively new company based in Moss, Norway, which utilizes their own direct drive motors to create humanoid robots. In this thesis the students will likely use their motors in addition to grippers that will be prototyped at ROBIN labs in IFI.