OSCAR: A Low-Cost, Open-Source Robotic Platform Design for Cognitive Research

Abstract

Research in cognitive robotics has been traditionally carried out in high-end research robots. This poses a barrier to many research institutions that cannot handle the costs associated with purchasing and operating this type of robot. In this paper, the design of a low-cost and open-source robot called OSCAR is presented. This robot was purposely designed to adapt to manipulation tasks in the context of cognitive robotics. The morphology, sensors equipped and software architecture of the robot are explained. Moreover, an initial implementation based in the Gazebo simulator is used to carry out an experiment involving a manipulation task using the Multilevel Darwinist Brain cognitive architecture. Results show that OSCAR is capable of learning a task and to obtain general information of its environment when paired with a cognitive architecture.