| This thesis addresses issues relating to the development of a robotic capability to assist a human in performing a complex task. It describes research and experiments involving a robot that assists a human to herd an animal. The novel focus of this work is that the robot is able to perceive the human's intentions by interpreting the human's movements without any explicit communication from the human. Furthermore, the robot is able to detect if the human is absent or unable to herd, and in that case, it herds the animal autonomously. A herding framework is developed based on low-stress herding techniques that enable the robot to start and stop herding, herd the animal forward, and turn the animal. Experiments were conducted to demonstrate the autonomous and assisted herding behavior of the robot. A conclusion is presented showing promising results that validate the approach for designing a robot with assisting and herding capabilities. |
