This paper presents a new technique for assessing the influence that clearance of spatial joints has on the kinematics and dynamics of multi-body systems. The technique is developed for the revolute joint in which the basic elements are the journal and bearing. Under certain working conditions, these two mechanical elements collide with each other being the contact treated as coupled forces. Th...

It is well known that a suitably designed unpowered mechanical biped robot can walk down an inclined plane with a steady gait The characteristics of the gait e g velocity time period step length depend on the geometry and the inertial properties of the robot and the slope of the plane A passive motion has the distinction of being natu ral and is likely to enjoy energy optimality Investiga tion ...

The focus of this work is a systematic study of the passive gait of a compass-like, planar, biped robot on inclined slopes. The robot is kinematically equivalent to a double pendulum, possessing two kneeless legs with point masses and a third point mass at the "hip" joint. Three parameters, namely, the ground-slope angle and the normalized mass and length of the robot describe its gait. We show...

Finding an optimal policy in a reinforcement learning (RL) framework with continuous state and action spaces is challenging. Approximate solutions are often inevitable. GPDP is an approximate dynamic programming algorithm based on Gaussian process (GP) models for the value functions. In this paper, we extend GPDP to the case of unknown transition dynamics. After building a GP model for the tran...

This paper presents an algorithm for finding the minimum time trajectory of an actual dynamic system by using on-line measurements of the state trajectory. The algorithm is shown to be extremely robust to mismatch between the model and the system. It is a projected gradient method that uses the measured terminal state error of the actual system and gradients based on the theoretical state equat...

In this article, we propose a new reinforcement learning (RL) method based on an actor-critic architecture. The actor and the critic are approximated by Normalized Gaussian Networks (NGnet), which are networks of local linear regression units. The NGnet is trained by the on-line EM algorithm proposed in our previous paper. We apply our RL method to the task of swinging-up and stabilizing a sing...

In human walking, the center of mass motion is similar to an inverted pendulum. Viewing double support as a transition from one inverted pendulum to the next, we hypothesized that the leading leg performs negative work to redirect the center of mass velocity, while simultaneously, the trailing leg performs positive work to replace the lost energy. To test this hypothesis, we developed a method ...

Previous double pendulum models of the golf swing could have included some over-simplifications (only two model segments, 2D analysis of 3D motion, rotation around a fixed pivot point). This has been overcome by full-body models driven by motion capture data from real swings. However, information regarding the validity of these models is still limited. Therefore, the aim of this study was to cr...

In this paper, we study a simple passive dynamic biped robot with point feet and legs without knee. The mathematical model describing the robot is a switched system, which includes an inverted double pendulum. Robot’s gait and its stability depend on parameters such as the slope of the ramp, mass distribution for the robot, and the length of its legs. The main result of the paper is an asymptot...

In this project, we apply reinforcement learning techniques to control an inverted double pendulum on a cart. We successfully learn a controller for balancing in a simulation environment using Qlearning with a linear function approximator, without any prior knowledge of the system at hand. We do however fail to learn a controller for the swingup maneuver, which leads to a discussion on what mig...