Objective: To remotely trigger and perform the feedback control of a pick and place operation with a 5-axis robotic arm

In robotics, trajectory is the time-bound path profile that the robotic tool or end-effector has to traverse generally to execute a manipulation task such as a pick and place operation. Making the manipulator follow a given trajectory is nothing but controlling the manipulator joint actuators in such a way that the desired path profile can be traversed with the highest possible accuracy.

Therefore suitable control schemes need to be designed.

While developing control strategies two things are done in the first place:

1.      The robotic manipulator to be controlled is mathematically modeled based on the physics governing the system.

2.     The manipulator is fitted with sensors and actuators to enable it to feed information from its environment to its controller and to respond according to the instructions of its controller (usually a microcontroller) respectively.

This control strategy can be open-loop control or closed-loop (feedback) control.

Open loop control is the simplest form of trajectory planning. In this scheme, the control algorithm is fed as input to the robot controller. The controller then sends the control signal to the actuators and the end-effector moves accordingly. This system of planning and control is inaccurate as there is no way in which the configuration attained by the end-effector at the end can be verified.

Due to difficulties in exactly computing the manipulator parameters like link length, link inertia, etc. and presence of certain uncertain/variable factors like payload (object) to be manipulated, friction, backlash, etc. the end-effector cannot execute the exact trajectory if open loop control is employed.

For this reason closed-loop control or feedback control is employed. Here after the manipulator has moved to a commanded configuration the sensors attached to them verify whether there is any error between the path to follow (desired trajectory) and the actual path that has been traversed. In case of error the sensors feedback this difference (error) to the robot controller which then sends the necessary control signal again till the desired configuration has been acquired by the manipulator to a high level of accuracy.

This employed feedback control scheme can again be linear as well as non-linear.   

For a large number of practical robot operations the robot moves rather slowly. In such a case, the joints of the robot can be controlled independently and the robotic system can be modeled mathematically by a set of linear differential equations with constant coefficients. Thus linear control schemes can be applied.

However if the robot moves fast then joints need to move simultaneously. So they cannot be controlled independently. Such a system can be modeled mathematically only by non-linear differential equations because of large dynamic coupling of joint torques.