We present a new method of learning control policies that successfully operate under unknown dynamic models. We create such policies by leveraging a large number of training examples that are generated using a physical simulator. Our system is made of two components: a Universal Policy (UP) and a function for Online System Identification (OSI). We describe our control policy as universal becaus...

Designing a robot that can autonomously traverse a variety of terrain types is difficult. For this reason, one may refer to nature for inspiration and produce robots that mimic biological organisms. EduBot is one such device, a hexapedal robot that resembles a cockroach. Building a robot that mimics a living creature offers numerous advantages, including design efficiency and locomotive stabili...

A modified double inverted pendulum – modified by connecting the mass carrying the pendulum with another mass through a spring makes the general inverted pendulum become a more interesting problem. The system is defined as a linear spring connected double inverted pendulum as proposed by Hou et al. [1],[2]. The system is highly nonlinear and unstable. However, the system can be simplified to a ...

This paper includes a review of three physics simulation packages that can be used to provide researchers with a virtual ground for modeling, implementing and simulating complex models, as well as testing their control methods with less cost and time of development. The inverted pendulum model was used as a test bed for comparing ODE, DANCE and Webots, while Linear State Feedback was used to co...

A growing cognitive model of motor balance skill (CMMBS), which simulates the manners of human beings and animals in motor balance skill learning, is presented. CMMBS, whose architecture is similar to the reflex arc of the biologic nerve, adopts a growing algorithm from reference to Growing Cell Structures in order to perform the pattern classification in the nerve center. This growing mechanis...

Abstract: In this study, a dynamic surface controller based on a nonlinear disturbance observer is investigated to control mobile wheeled inverted pendulum system. By using a coordinate transformation, the underactuated system is presented as a semi-strict feedback form which is convenient for controller design. A dynamic surface controller together with a nonlinear disturbance observer is desi...

Control of Rotary Inverted Pendulum (RIP) is a classic control problem. It represents a broader class of underactuated systems. The dynamical equations of RIP are nonlinear coupled and complex. The nonlinear feature and the complex internal dynamics makes the design of the control challenging. This paper presents swing up and stabilization both via sliding modes. Tracking control of rotary actu...

Robots can help to demonstrate and prove concepts on human locomotion such as concepts based on springlike leg behavior. Hopping is a fundamental requirement for running in two basic functions: bouncing and balancing. In addition, for robust hopping against perturbations, swinging the leg in flight/swing phase is observed. These three elements need to be integrated for stable/robust locomotion ...

A number of foot placement strategies for walking have been proposed that make use of widely varying model complexities. Although a number of successful demonstrations have been individually shown in simulation and on physical robots, it is difficult to make a direct performance comparison due to the large differences in hardware, gait generation strategy, control system gains, actuator saturat...

Humans face the task of balancing dynamic systems near an unstable equilibrium repeatedly throughout their lives. The task of inverted pendulum (stick) balancing (Fig. 1a) is an increasingly popular paradigm of studying human control behavior in such situations [1]. In particular, much research has been aimed at understanding the mechanisms of discontinuous, or intermittent control in the conte...