Finally my Masters research is complete. Here is a experiment video highlighting my work
embedded by Embedded Video
This method takes joint torques(generated by position controller) and converts them into the force domain so that they can be limited(to prevent tipping, twisting, sliding, etc) and finally realized via force control.
So essentially the position controller becomes a reference force generator.
The advantages of this method are:
- No ZMP trajectory is needed: ZMP moves freely to realize the desired motion but the forces at the foot are constrained to prevent tipping(ZMP going to edge of support)
- Prevent sliding&twisting of the foot and prevent excessive GRF’s after jumping: Limit all 6-axes of the force wrench to prevent unstable/undesirable situations.
- Improved robustness: It is proved that this method has the same positioning performance as pure position control(as long as force limiting is not needed). Any position controller can be used since the inputs are the desired joint torques. Therefore motion response depends purely on the position controller used.
- Independent joint control: The joints can be moved separately as long as the degrees of freedom dont exceed 6 and the manipulator is not in a singular position.
Here is a block diagram of the method:
Please read my Masters research page for more details.
Now that the intuitiveness of position control has been combined with the versatility of force control. I look forward to advancing the field of robotics in my PhD 