Probabilistic state estimation and calibration for robot manipulators

Simple item page
dc.contributor.advisorVan Daalen, Corne E. en_ZA
dc.contributor.advisorBurke, Michaelen_ZA
dc.contributor.advisorMakondo, Ndivhuwoen_ZA
dc.contributor.authorSijovu, Zimkhithaen_ZA
dc.contributor.otherStellenbosch University. Faculty of Engineering. Dept. of Electrical and Electronic Engineering.en_ZA
dc.date.accessioned2024-03-04T08:45:17Zen_ZA
dc.date.accessioned2024-04-26T22:29:41Zen_ZA
dc.date.available2024-03-04T08:45:17Zen_ZA
dc.date.available2024-04-26T22:29:41Zen_ZA
dc.date.issued2024-03en_ZA
dc.descriptionThesis (MEng)--Stellenbosch University, 2024.en_ZA
dc.description.abstractENGLISH ABSTRACT: A fundamental task in all robotic applications is the ability of a robot to determine its position and orientation in the environment. Such localisation tasks are used in a variety of robotic applications like manufacturing, pick and place, medical applications, and many others. Achieving high localisation accuracy in these applications is difficult and important to achieving reliable and full robotic autonomy. This thesis presents a state estimation approach for robot manipulators subject to uncertainty. The robot manipulator is mounted on a tracked mobile base but only considers the case when the base is stationary. The proposed method combines two sources of data to improve the accuracy of the position of the manipulator's end effector: one from the joint encoders and one from the robot-mounted camera. First, the measurements from the joint encoders are transformed using the kinematic equations of the robot to estimate the position of the end effector. Then, 2D camera measurements are obtained by observing a marker attached to the manipulator end effector. The measurements obtained from these two sources are associated with uncertainties. Also, the mathematical equations for kinematics to transform from the joint angles to the end-effector position, and the camera model used for projecting the end-effector position to the image plane are generally non-linear. A probabilistic framework is developed for the systematic integration of the two probability distributions, using Bayes' theorem to calculate the posterior distribution of the end-effector position. This method uses a well-known technique called unscented transform to approximate the uncertainty in the manipulator end-effector position. The presented approach is initially verified in a simulation environment to test its performance compared to a Monte Carlo approach. Then the estimation algorithm is verified using the real-robot data obtained from the robot's joint encoders and the robot camera mounted on the shoulder of the manipulator. The experimental results indicated that the unscented transform is a good state estimation algorithm to handle uncertainties and show empirically that incorporating camera measurements into joint encoder measurements significantly improves the end-effector positioning accuracy. The numerical analysis demonstrated good accuracy, with an error of around four centimetres when compared to the Vicon motion capture data that is used as ground truth. The estimation accuracy is also quantified by the Mahalanobis distance metric, which also shows that about 95-97% of all the observed end-effector values fall within the three standard deviations.en_ZA
dc.description.abstractAFRIKAANSE OPSOMMING: Geen opsomming beskikbaar.af_ZA
dc.description.versionMastersen_ZA
dc.format.extentxiii, 79 pages : illustrationsen_ZA
dc.identifier.urihttps://scholar.sun.ac.za/handle/10019.1/130572en_ZA
dc.language.isoen_ZAen_ZA
dc.language.isoen_ZAen_ZA
dc.publisherStellenbosch : Stellenbosch Universityen_ZA
dc.rights.holderStellenbosch Universityen_ZA
dc.subjectProbabilistic state estimation and calibration for robot manipulatorsen_ZA
dc.subject.lcshManipulators (Mechanism) -- Automatic controlen_ZA
dc.subject.lcshElectric power systems -- State estimationen_ZA
dc.subject.lcshRobot manipulatoren_ZA
dc.subject.lcshEnd-effectoren_ZA
dc.subject.lcshRoboticsen_ZA
dc.subject.lcshUCTDen_ZA
dc.titleProbabilistic state estimation and calibration for robot manipulatorsen_ZA
dc.typeThesisen_ZA
Files
Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
sijovu_probabilistic_2024.pdf
Size:
20.59 MB
Format:
Adobe Portable Document Format
Description:
Download
Collections
Masters Degrees (Electrical and Electronic Engineering)