Programming with Articulated Objects
Abstract
Computer simulation promises to have a dramatic influence on the study of mechanical systems and the interaction between humans and machines. The central goal of this study methodologies for controlling complex physical objects in simulation. Our work provides a conceptual framework that integrates the mechanical simulator as a multi-level constraint solver. The usefulness of the approach was demonstrated through a series of experiments examining the control of robot locomotion. In our work, simulation is cast as a problem of solving a sequence of time-varying dynamic constraints on the attributes of physical objects. Constraints due to mechanical considerations, such as joints between bodies, and constraints due to control considerations, such as guided movements, are treated in a uniform fashion. At any instant in time, a system of linear equations constrains the instantaneous motion of bodies. These motion equations relate accelerations of bodies to the forces and torques on the bodies. Given the positions and velocities of all bodies at time t, positions and velocities at time t + delta t are found by solving the motion equations for accelerations and integrating from previous states. Given initial positions and velocities, this process is iterated to simulate motion over an interval of time.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 30, 1990
- Accession Number
- ADA231168
Entities
People
- Joseph K. Kearney
Organizations
- University of Iowa