VBOC: Learning the Viability Boundary of a Robot Manipulator Using Optimal Control

Safety is often the most important requirement in robotics applications. Nonetheless, control techniques that can provide safety guarantees are still extremely rare for nonlinear systems, such as robot manipulators. A well-known tool to ensure viability kernel, which largest set of states from be ensured. Unfortunately, computing a system challenging general. Several numerical algorithms approximating it have been proposed literature, but they suffer curse dimensionality. This letter presents new approach numerically kernel Our solves optimal problems compute guaranteed on boundary set. allows us learn directly boundary, therefore learning smaller dimensional space. Compared state art systems up dimension 6, our algorithm resulted more than 2 times accurate same computation time, or 6 fast reach accuracy.