Assembling of Piecewise Time-Optimal and Smooth Trajectories Along Predefined Paths for Industrial Robots

Description

Time-optimal trajectory planning along predefined geometric paths is an extensively discussed research topic for industrial applications in the field of robotics. The majority of approaches dealing with this problem make use of an explicit parameterisation of the predefined path in terms of the arc length s [1, 2]. These algorithms are applicable to paths with relatively short length. However, the dramatic increasing in calculation time and memory demand renders this approach extremely inefficient or even impossible to be applied to realistic paths. In this paper a solution method is presented where the path is split into subsequent sections that can be handled in an efficient way. The solutions for the individual subproblems are assembled to provide the optimal trajectory of the entire path. It is crucial that this assemblage of optimal solutions of adjacent subsections is C1 compatible at the transition points. Moreover, the process as well as the robot impose jerk constraints. The time-optimal trajectory does a priori not respect such jerk restrictions. This is unacceptable especially for elastic robots [3]. Taking jerk restrictions into account calls for more complex optimization algorithms and hence leads to higher computation time [2]. In practice usually no exact jerk limitations are available but are rather deduced from practical experience. Instead of imposing these restrictions in the optimization, it is proposed in this paper to use a spline approximation of the determined optimal velocity trend in s × (ds/dt)^2 plane. Therewith the torque is guaranteed to be smooth and the algorithm complexity and computation time is reduced.

Period	09 Feb 2015
Event title	EuroCAST 2015
Event type	Conference
Location	SpainShow on map