A new simpler vision based robot control system is proposed characterized with position specificrnartificial neural network (ANN) and end-effecter integrated camera system. Position specific ANNrnavoids the difficulty of covering the whole joint space with changing parameters using one set ofrnANN, and end-effecter integrated camera system makes the image of an object consistent whenrnthe end-effecter approaches the object. The object coordinate can be directly used as feedback. rnMost vision-based robot positioning techniques rely on analytical formulations of the relationshiprnbetween the robot pose and the projected image coordinates of several geometric features of thernobserved scene. Feature matching algorithms, camera calibration, models of the camera geometryrnand object feature relationships are also necessary for pose determination. These steps are oftenrncomputationally intensive and error-prone, and the complexity of the resulting formulations oftenrnlimit the number of controllable degrees of freedom. rnThis thesis presents controlling mechanism of a parallel robot based on deep neural learning andrnposition based visual servoing that overcomes many of these limitations. ROS/Gazebo simulatorrnis used to model delta 3 parallel robot. From the model training data set is collected and a multilayerrnfeedrnforwardrndeeprnneuralrnnetworkrnisrnusedrntornlearnrntherncomplexrnimplicitrnrelationshiprnbetweenrnrnthernposerndisplacementsrnofrntherndeltarn3rnrobotrnandrnjointrnangles.rnThreernnetworksrnwithrnthreernhiddenrnrnlayersrnrnbut different number of neurons per hidden layer were trained and their performance isrnevaluated. Based on the simulation result it is shown that a network with higher number of neuronsrnper hidden layer shows better performance. rnThe trained network may then be used to move the robot from arbitrary initial positions to a desiredrnpose with respect to the observed scene with MSE less than 0.05. Simulation result shows that thernsystem works smoothly, and converges in limited steps. The algorithm simplifies the model ofrnvision based robot manipulator control system, and improves the control accuracy and responserntime.