What and where we will use stuff...

[vicentepedro]

@vvv-school/vvv17-team-orange Feel free to add the rest of the stuff that we talked...

Vision

• Detect the object position (3D) using IOL
• Detect the board
• Detect the empty spaces

Machine Learning and Deep Learning

• Learn to distinguished between O and X

Kinematics

• Inverse kinematics (from the 3D position of the objects)
• Closing the fingers to Grap the object
• Release the object in the right position

Dynamics

• Controlling the arm from the initial joint position to the final joint position received from the kinematics module.
• Torque Control

GIT

• Everything we commit and push =P

  YARP

• In every module we have created

[gmmarconi]

I would put Deep Learning under Machine Learning. Also, I'm more and more skeptical about dynamics, do we really need torque control? This question is aimed at the dynamics guys in particular.

[vicentepedro]

For our demo: We can do everything with kinematics. It's simpler. For the sake of using everything in the demo, it could be a good idea to use Dynamics to control the arm.

I'm not sure if the Blacky iCub has accuracy enough to exploit torque control for the reaching movement... maybe after dinner, we can clarify this issue.

[DanielePucci]

@vicentepedro @gmmarconi let me remind you that one of the aims of the challenge is to combine all the topics you have been through during the week. So, if you guys won't take into account any dynamics related quantity at all, this will be clearly a weakness of your project. Let me also pinpoint that taking into account dynamics may also mean taking into account external forces/wrenches applied to the robot, not necessarily controlling the robot with torque control.

[gmmarconi]

@DanielePucci Yes, in the end we discussed about that with Ugo and Vadim and they told us how to get some useful information about hand dynamics when grasping the glasses! Thank you for the heads up.