Guidance control for automatic guided vehicles employing binary camera vision

摘要

This thesis presents guidance control schemes for a class of Automatic Guided Vehicles (AGVs) employing binary camera vision. These control schemes enable the AGV to follow a given track profile accurately. The accuracy in following tracks has been set based on the current industrial needs for assembly automation. For guidance, the binary camera is continuously focussed on an illuminated track provided on the floor prepared by taping or otherwise. An efficient image processing technique that is capable of identifying the track and recognizing a set of road signs is developed. The image processing scheme provides the instantaneous position and orientation of the longitudinal axis of the vehicle relative to the track and this information is used for guidance control.;This thesis presents three new approaches for guidance control of the vehicle to achieve track following. The first approach employs a camera at the front end of the vehicle. This assures accurate track following of a small section of the vehicle. The steering command is computed using the instantaneous position and orientation offsets of the longitudinal axis of the vehicle relative to the track. The second approach employs an additional camera at the rear end to provide accurate track following of the front and rear ends of the vehicle. In this scheme, the steering command is calculated using control laws that make use of the instantaneous position and orientation offsets as measured by the front and rear sensors of the vehicle. This scheme employs control equations that are dependent on the presence of track curvature. Further, the choice of control equations is also dependent on the relative position of the front and rear ends of the vehicle with respect to the track. In addition to this, the gains of the control equations are changed in real-time depending on the relative position of the front and rear ends of the vehicle with respect to each other as well as the track. The third approach is a feed forward control scheme for accurate track following of any section of the vehicle as defined by a driving criterion. This is an intelligent method as the vehicle guidance is based on the identified geometry of the track in real-time. The geometry of the track is identified using the motion parameters available upto that instant. The driving criterion defines the positioning of the AGV body on the track, i.e. which section(s) of the AGV has to be on the track. The steering command is computed based on the predicted track profile ahead of the vehicle and the driving criterion.;A methodology to select a suitable wheelbase configuration and combinations of driving and steered wheels without redundancy for optimal kinematic control of AGVs is presented. A criterion is derived for this purpose by considering a generic kinematic model. A prototype AGV with a differential drive and a diamond shaped wheelbase configuration has been built for experimental study. Results of simulations and experiments indicate satisfactory performance.