Task oriented robotics

  • Paul Strickland

    Student thesis: Doctoral Thesis


    An alternative between product dedicated automation and general purpose robots is presented. with the new approach a given robot is customised to fulfil the requirements of the manufacturing tasks to be automated, is programmed in terms of production tasks or can be truly automated. This allows exploitation of the natural relationship between production tasks and robot systems.
    Current construction of industrial robots relies on a one-to-one relationship between robot and controller. The perceived way forward with this constraint has been the functionally related general purpose Industrial Robot. paradoxically, industrial robots are 'bolted' to task specific environments which have fixed repertoires of materials and tools to act upon.
    Computer integration of these functional machines involves human interaction to constrain the general purpose robots to relatively simple production tasks. This increases the overall cost, levels of expertise required to program the robots and lead-times in reprogramming. Unquestionably these factors have led to a reluctance towards exploitation of industrial robots.
    The research undertaken endeavoured to provide an alternative to this method of automation. The research completed allows robots to be programmed in terms of production tasks and dissolves the necessity to specifically design a robot controller for a given robot configuration. A modular robotic framework which consists of a number of generic modules has been employed. A loosely coupled transputer computer network has been implemented to encompass task, robot coordination and robot axis levels.
    At task level a 'production orientated programming environment' reflects the corresponding manual production activities. Information is gathered from this environment to allow 'task related rules' to be formulated. These 'task rules' have been utilised to fully automate, allowing product specifications to be translated to machine actions.
    The robot coordination level translates global coordinates to joint actions. A set of closed loop inverse kinematic equations have been generalised to ensure that the robot controller is not dependent upon a given robotic structure. These generic equations are customised to the localised constraints of each modular robotic element of the robot structure. Robust axis control is utilised to decouple robot control at joint level. 'Mix-and match' hardware and software techniques have been created which facilitate customisation of a given robot axis, and in turn, the ascending levels of the system. Hardware design capitalised on new advances in compact components which allowed self contained modular robotics elements to be formed.
    Date of Award1993
    Original languageEnglish
    Awarding Institution
    • University of Portsmouth
    SupervisorJim Hollis (Supervisor), John Billingsley (Supervisor) & David Sanders (Supervisor)

    Cite this