A computational Grid can be an interesting solution for distributed remote robot control. It can provide computational resources when the usual ones are too loaded or not powerful enough. It can support fault tolerance allowing redundant computations. It can make possible the robotic system sharing with remote partners. We designed a Grid architecture across Internet, including resources from two laboratories, one in Italy and one in France. It is based on the DIET GridRPC environment and supports distributed remote redundant control of an autonomous robot. This paper focuses on high-level Grid services and on a specific GridRPC library we designed for improving the robotic application development.