Model Independent Visual Servoing CMPUT 610 Literature Reading Presentation Zhen Deng

Introduction w Summaries and Comparisons of Traditional Visual Servoing and Model independent Visual Servoing emphasizing on the latter. w Works are mostly from Jenelle A. Piepmeier’s thesis and Alexandra Hauck’s thesis

Visual Servo w Visual servo control has the potential to provide a low-cost, low-maintenance automation solution for unstructured industries and environments. w Robotics has thrived in ordered domains, it has found challenges in environments that are not well defined.

Traditional Visual Servoing w Precise knowledge of the robot kinematics, the camera model, or the geometric relationship between the camera and the robot systems is assumed. w Need to know the exact position of the end- effector and the target in the Cartesian Space. w Require lots of calculation.

Rigid Body Links

Forward Kinematics w The Denavit-Hartenberg Notation: i-1 T i = Rotz(    ransz(d)  Rotx(   Trans(a) w Transformation 0 T e = 0 T 1 1 T 2 2 T 3 … n-1 T n n T e

Jacobian by Differential w Velocity variables can transformed between joint space and Euclidean space using Jacobian matrices   x = J *    J \  x  J ij =  i  x j

Calibrated Camera Model

Model Independent Visual Servoing w An image-based Visual Servoing method. w Could be further classified as dynamic look- and-move according to the classification scheme developed by Sanderson and Weiss. w Estimate the Jacobian on-line and does not require calibrated models of either of the camera configuration or the robot kinematics.

History w Martin Jagersand formulates the visual Servoing problem as a nonlinear least squares problem solved by a quasi-Newton method using Broyden Jacobian estimation. w Base on Martin’s work, Jenelle P adds a frame to solve the problem of grasping a moving target. w me ? …

Reaching a Stationary Target  Residual error f(  ) = y(  y *.  Goal: minimize f(  )   f = f k - f k-1  J k = J k-1 + (  f-J k-1  T /  T    k+1  k  J -1 k f k

Reaching a Fixed Object

Tracking the moving object w Interaction with a moving object, e.g. catching or hitting it, is perhaps the most difficult task for a hand-eye system. w Most successful systems presented in paper uses precisely calibrated, stationary stereo camera systems and image-processing hardware together with a simplified visual environment.

Peter K. Allen’s Work w Allen et al. Developed a system that could grasp a toy train moving in a plain. The train’s position is estimated from(hardware- supported) measurements of optic flow with a stationary,calibrated stereo system. w Using a non-linear filtering and prediction, the robot tracks the train and finally grasps it.

“Ball player” w Andersson’s ping-pong player is one of the earliest “ball playing” robot. w Nakai et al developed a robotic volleyball player.

Jenelle’s modification to Broyden  Residual error f( ,t) = y(  y*(t).  Goal: minimize f( ,t)   f = f k - f k-1  J k = J k-1 + (  f - J k-1  y*(t)  t  t  T /  T    k+1  k  J k T J k ) -1 J k T (f k  y*(t)  t  t 

Convergence w The residual error converges as the iterations increasing. w While the static method does not. w The mathematics proof of this result could be found in Jenelle’s paper.

Experiments with 1 DOF system

Results

6 DOF experiments

Future work ? w Analysis between the two distinct ways of computing the Jacobian Matrix. w Solving the tracking problem without the knowledge of target motion. w More robust … ?

Literature Links w ivs/vsweb2.html w A Dynamic Quasi-Newton Method for Uncalibrated Visual Servoing by Jenelle al w Automated Tracking and Grasping of a Moving Object with a Robotic Hand-Eye System. By Peter K. Allen

Summary w Model Independent approach is proved to be more robust and more efficient.