Virtual Manikin Controller - Calculating the movement of a human model

Abstract

Simulation of assembly line industrial robots has long been an area of great research interest. At Fraunhofer-Chalmers Research Centre for Industrial Mathematics (FCC) a main objective has been to automatically calculate valid motions - solving problems of inverse kinematics, path planning and collision avoidance. In order to include human assembly in the simulation environment the manikin project Intelligently Moving Manikin (IMMA) has been developed. The IMMA project is able to calculate valid poses where problems such as balance, collision avoidance and ergonomics has been taken into account. This master thesis improves the manikin motion generation by limiting the velocity and acceleration, using the calculated poses as initial values, and stating the problem as a numerical optimal control problem. The control signal is optimised with respect to time and energy and is at the same time required to fulfill conditions on positions and limits of angular values, angular velocity and angular acceleration. All variables included in the problem formulation are discretised using low order Galerkin approximations and the nonlinear optimisation makes use of the open source interior point optimisation software Ipopt. The result is a smooth and efficient manikin movement which is displayed using the IMMA visualisation environment. The implementation is as general as possible in order to provide an optimisation framework where the objective function, numerical integration method and the constraints are all easily interchangeable.