Loading...
Please wait, while we are loading the content...
Similar Documents
Simulation of Walking Humanoid Robot Based on Matlab / Simmechanics
| Content Provider | Semantic Scholar |
|---|---|
| Author | Corner, Sébastien |
| Copyright Year | 2011 |
| Abstract | The complexity involved in robotic simulation seems to be an insurmountable obstacle for students to overcome. It usually requires substantial mathematics and programing skills to be master by students. This paper proposes a new approach to humanoid robot simulation that has the potential to reduce the intricacy of creating a biped simulation model. The approach investigates the use of the SimMechanics toolbox to build the computer model of the biped robot. The method operates by assembling body and joints modules to perform the framework of the robot. The control motion of the robot requires neither the mathematical conception of inverse kinematics nor inverse dynamics. It adapts position joints control using PID controller that tracks an angle trajectory file. The methodology of acquiring the corresponding coefficient of the controller, that allows the simulated robot to walk according to the given trajectory file, is described. Another key component of this simulation is a control technique that models the ground based on PID controller. It provides a basis for our analysis of the robot system performance while walking and a better comprehension of the ground reaction forces and impacts. Finally, this paper opens the floor to further projects such as passive dynamic walking. Introduction Simulation becomes a strategic tool in the research field of robotics used by many students, researchers, and developers. It supports the development of advanced designs and investigates a wide range of solutions for complex problems [1]. Design solutions and experimental environments are built according to their imagination and creativity. The exponential increase in hard drive capacity has allowed the development of numerous software packages for modeling multibody systems. Some of them have been specifically developed for robots such as Webots, Adams and RoboWorks. These softwares are really convenient for modeling robots and their dynamic motions, but are viewed as inadequate for modeling actuators, sensors, ground contacts, and impact [2]. SimMechanics extends Simulink/Matlab to provide tools for modeling and simulating mechanical systems [3]. With SimMechanics modeling bodies, joints, actuators, external forces, and the interconnections in the robot drive system is quite simple. Previous papers exhibit the simplicity of this software by providing details and simple examples that allows novice users to be familiar with the software [4][5][6]. Physical modeling and parameter settings are presented in detail. In addition, they develop the approach of modeling the nonlinear system, the inverted pendulum, built through SimMechanics. Their approach is to compare this model with the mathematic model based on the derivation of the forward dynamics and differential equations. Results show that both models are almost identical which confirm and validate the SimMechanics model. The main differences results to the simplicity of implementing this nonlinear system. The mathematical approach is quite cumbersome, as it requires the derivation of the nonlinear motion of that mechanism and linearizes the equations using state variables. The SimMechanics model is much simpler. Bodies and joints are linked together to form the model, while joint sensors and actuators are added for simulation. This paper is about a new approach that simulates biped robotic though SimMechanics that embodies pendulum motions, which has the potential to reduce the intricacy of creating biped simulation model. The approach starts with highlighting the skeleton of the simulated robot made of joints and body links. Once built, the control motion of the robot is explored toward PID controller used to simulate the revolute joints and to simulate ground reactions. This leads the approach to exhibit the results of this research which consists of having the simulated robot accomplish a full step. To validate our approach, screenshot of the simulation will be presented and a website link will be shared to access the video. Architecture of the simulated robot The Frame The structure of the simulated biped robot, presented on figure1a, contains a complete lower body including, two legs and two foot, attached to a torso. The head and arms are not designed to shorten the complexity of the simulation. The hip connects the leg to the torso, using a revolute and weld joint respectively. The knee connects the thighs and the calf which is connected to the foot by the ankle. Both the knee and the ankle are revolute joints. In this research, the robot is free to move in the sagittal plane (y and z axis) and can rotate around the perpendicular axis of the plane (x axis). The overall mass of the robot is 37 kg for a total height of 1.45m. The weight and length are evenly distributed among each link. (The torso weights 10 kg, the hips and each thigh and calf weight 5 kg and each foot weight 1 kg). The moment of inertia of each body is respect to the body’s center of mass. It was assumed that all links were rigid and frictions in the joints were neglected. Finally, the simulated robot is a rigid robot with six degrees of freedom that can move and rotate freely in the sagittal plane. Figure1A: Algorithm structure of the simulation Figure1b: Model of the robot. We note that each revolute joint is controlled by a servo module and the ground contacts as well as the frictions forces are simulated while the foot impacts the ground. SimMechanics does not permit surface contact between objects, therefore, three points of contact are modeled on each foot to simulate the impact. The model of the robot, when the simulation runs, is shown on figure 1b. The green square refers to the ground and has a thickness of 1cm. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | http://www.sebastiencorner.com/app/download/7114460298/FINALPAPER_SIMULATION_SebastienCorner.pdf |
| Alternate Webpage(s) | https://www.sebastiencorner.com/app/download/7114460298/FINALPAPER_SIMULATION_SebastienCorner.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
