System Integration And Testing Of A High-speed Delta Parallel Robot

In this thesis, the Delta robot, a parallel manipulator with three translationaldegrees of freedom is investigated, for the purpose of improving the PPO (Pick andPlace Operation) motion performance.Firstly, the robot’s fundamental concept and characteristics as well as itsstate-of-the-art technology and industrial applications are introduced. In kinematicanalysis, the freedom of system is discussed; the derivation of the inverse and forwarddisplacement solutions is presented. Based on these results, a representation of theworkplace and the instantaneous Jacobian matrix are then obtained. A simple methodbased on virtual work principle is adopted for dynamics modeling, which is thefoundation of trajectory plan.Second, detailed investigation has been done on the operation trajectories ofDelta robot’s end-effector including the profiles of both the velocity and theacceleration on the path. The transitions between the vertical and horizontal segmentsof the operation trajectory are smoothed with Lamé curves, thus determining theoverall shape of the trajectory in Cartesian space. High order polynomials are adoptedto regulate the one-dimensional displacement along the curve and generate theposition, velocity, acceleration information necessary for the interpolation points.Using the dynamics model of Delta robot, energy-based cost functions are definedand utilized to optimize Lamé curve’s geometric parameters. Simulations show thatthe optimized trajectory could reduce the total energy cost in a typical pick-and-placecycle, as well as the residual vibrations at the terminal positions of the path.Thirdly, a machine vision system which is used to guide the Delta robot sortingmovement is developed. The works include camera modeling, inside and outsideparameters calibration, position relationship acquisition between the robot and thevision system using the known size of the rod is fixed at the end of the robot, andmethod of objects recognition based on template learning and matching.Finally, an experimental prototype is fabricated. A method on parameter tuningfor PID parameter is presented, which takes the mean square root error of the servomotors as the objective. The trajectories planned are tested on the platform. Theresidual vibrations as the robot’s end-effector reaches the terminal are recorded by theacceletion sensors bounded on the move-plate and then compared under different Lamé curve parameters. These results verify that the optimized trajectory has a betterperformance in terms of the motion quality. A vision system has been integrated withthe robot, which renders the platform the intelligence to recognize different shapes ofarticles. Utilize a turntable as a conveyor, make the Delta robot identify and pickdifferent objects at frequency of40,80,120cycles/min. The operation speed andprecision of the overall system is thus attained.