We have a project where we are controlling a robotic arm using a wearable glove component. I have to simulate one of the features we are implementing which includes a flex sensor (input), servo motor (output) and force sensor (feedback). The servo motor will be connected to the robot's clamp and is controlled by the flex sensor input between 0-90 degrees. The clamp should be able to grip onto objects with a specific amount of pre-defined force. I'm having trouble implementing this logic in my code. The logic should work for firm objects of different sizes. This means that the initial impact in terms of degrees may be different for each object. Essentially, the servo should take continuous input from flex sensor up until say a 50g reading on the force sensor to which further input from the flex sensor no longer adds additional input to the servo regardless of the further degrees bent until the flex is relaxed and the force sensor reads a value below 50g again. I know using a force sensor atm may not be ideal as its non-linear however it is only meant for simulation and we may later implement a PID instead when presented with more time in the summer.
I'd use the flex sensor to indicate the desired force instead of the servo angle, at least once the force sensor signals contact with an object.
A servo can not apply specific force. Once it cannot reach the desired angle or linear extent it will apply its full torque to the clamp or claw. Here I'd add a flexible extension of a rotational servo arm that translates the servo angle into a force that can be measured by another flex sensor.
Or do you want to mimic compressible fingers around a stiff bone, or spiral springs, to apply a certain force by deformation?
A sketch of your intended construction were really helpful.
We're going to be using an old Robix Rascle that's provided to us. The clamp is quite small and I think you may have been thinking of a bigger size clamp. I'm honestly not entire sure how the force feedback feature will be implemented if it is more of a mechanical limitation. If there's more ideal solutions I'm open to suggestions. At the moment, its more for applying force to solid objects that may require different amount of grip/force to pick them up. The current implementation may work with compressible objects however it would increase complexity and may cause the system to jitter too much due to somewhat inaccurate readings of the sensors. This of course could later be made robust with better software.
This topic was automatically closed 180 days after the last reply. New replies are no longer allowed.