Human-robot interaction using a behavioural control strategy

Abstract

A topical and important aspect of robotics research is in the area of human-robot interaction (HRI), which addresses the issue of cooperation between a human and a robot to allow tasks to be shared in a safe and reliable manner. This thesis focuses on the design and development of an appropriate set of behaviour strategies for human-robot interactive control by first understanding how an equivalent human-human interaction (HHI) can be used to establish a framework for a robotic behaviour-based approach. To achieve the above goal, two preliminary HHI experimental investigations were initiated in this study. The first of which was designed to evaluate the human dynamic response using a one degree-of-freedom (DOF) HHI rectilinear test where the handler passes a compliant object to the receiver along a constrained horizontal path. The human dynamic response while executing the HHI rectilinear task has been investigated using a Box-Behnken design of experiments [Box and Hunter, 1957] and was based on the McRuer crossover model [McRuer et al. 1995]. To mimic a real-world human-human object handover task where the handler is able to pass an object to the receiver in a 3D workspace, a second more substantive one DOF HHI baton handover task has been developed. The HHI object handover tests were designed to understand the dynamic behavioural characteristics of the human participants, in which the handler was required to dexterously pass an object to the receiver in a timely and natural manner. The profiles of interactive forces between the handler and receiver were measured as a function of time, and how they are modulated whilst performing the tasks, was evaluated. Three key parameters were used to identify the physical characteristics of the human participants, including: peak interactive force (fmax), transfer time (Ttrf), and work done (W). These variables were subsequently used to design and develop an appropriate set of force and velocity control strategies for a six DOF Stäubli robot manipulator arm (TX60) working in a human-robot interactive environment. The optimal design of the software and hardware controller implementation for the robot system has been successfully established in keeping with a behaviour-based approach. External force control based on proportional plus integral (PI) and fuzzy logic control (FLC) algorithms were adopted to control the robot end effector velocity and interactive force in real-time. ii The results of interactive experiments with human-to-robot and robot-to-human handover tasks allowed a comparison of the PI and FLC control strategies. It can be concluded that the quantitative measurement of the performance of robot velocity and force control can be considered acceptable for human-robot interaction. These can provide effective performance during the robot-human object handover tasks, where the robot was able to successfully pass the object from/to the human in a safe, reliable and timely manner. However, after careful analysis with regard to human-robot handover test results, the FLC scheme was shown to be superior to PI control by actively compensating for the dynamics in the non-linear system and demonstrated better overall performance and stability. The FLC also shows superior performance in terms of improved sensitivity to small error changes compared to PI control, which is an advantage in establishing effective robot force control. The results of survey responses from the participants were in agreement with the parallel test outcomes, demonstrating significant satisfaction with the overall performance of the human-robot interactive system, as measured by an average rating of 4.06 on a five point scale. In brief, this research has contributed the foundations for long-term research, particularly in the development of an interactive real-time robot-force control system, which enables the robot manipulator arm to cooperate with a human to facilitate the dextrous transfer of objects in a safe and speedy manner.