Minimum-time path tracking for robots with non-convex constraints

The paper examines the time-optimal path tracking problem which belongs to robot motion planning. The goal is to generate a velocity profile for robotic manipulators taking into account kinematic and dynamic constraints. In some cases, profile generation can be translated to a convex optimisation problem through a non-linear change of variables. However, some constraints cannot be taken into account due they non-convex properties. The paper presents a method to solve the time-optimal problem with these non-convex constraints. The method is based on the branch and bound algorithm, and it provides a global solution in contrast to the existing local approaches. Experimental results with a 3-DoF robotic arm are also presented.