Shared Autonomy Control of Telepresence Robot
A method of shared control is presented considering the environment conditions of a telepresence robot. A grid is considered in which the robot moves from cell to cell following the operator’s command. Cells are classified as untraversable on the basis of sensor information. A set of motion primitives are used as the trajectories that the robot can follow. We check the collisions of the trajectories with the untraversable cells and select the collision free motion primitives. Approximate feedback linearization control technique is applied on a unicycle model to track the trajectories that avoids the obstacles and move the robot towards the prescribed direction by the operator.
Figure 1 shows the basic outline of our approach. Operator performs the teleoperation, while the robot sensors scans the environment for obstacles. If an obstacle is detected, an obstacle avoidance algorithm overrides the operator’s command. Figure 2 shows the outline of the obstacle avoidance algorithm. We generate some motion primitives at each point and check whether they are in collision with an obstacle. We select the motion primitive which is collision free and track the trajectory to move the robot.
Kinematic model of a differential drive robot was studied to design the controller which can track the trajectories. To design a controller the error dynamics equation is linearized around the desired or reference trajectory.
Project Code can be found here
