TRAJECTORY OPTIMIZATION AND REAL-TIME SIMULATION FOR ROBOTICS APPLICATIONS

This paper illustrates the use of the general-purpose multibody free software MBDyn for trajectory optimization in space robotics. Trajectory optimization requires the capability to explore the space of the feasible solutions which implies simultaneously satisfying the dynamics equations of the system and equality and inequality constraints. A well-known and widely adopted technique is based on shooting, i.e. integrating in time the dynamics equations of the system, to compute the objective function and the constraints. When they are both differentiable, their derivatives with respect to the optimization variables can be numerically estimated by difference equations, and Sequential Quadratic Programming (SQP) techniques can be used to search the optimal set of variables. Among the advantages of this technique, any valid means to generate the objective function can be used, including multibody analysis; this allows to introduce a high degree of flexibility and detail in the model of the problem. A distinguishing feature of the presented approach is that a single general-purpose software is used for both the optimization and the real-time hardware-in-the-loop simulation of the prob- lem, with a nearly unparalleled commonality of models and modeling environment.