The recent penetration of distributed green energy, distribution intelligence, and plug-in electric vehicles (EVs) calls for a better information organization in an Energy Internet. Such an information system usually considers mostly the static grids. In this paper, we propose a test-bed not only for the static grid but also for the dynamic devices such as electric vehicles. To reduce the physical size of such a test-bed, we design the robot to have two wheels. To avoid human interruption to the test, we replace the manual charging procedure with wireless charging. To enable the EV robot for self-balancing with its two wheels, we implanted the EV robot with a control algorithm based on a neural network. To provide the EV robots with charging instructions, a joint optimization is carried out by considering both the economic factor on the customer side and the grid security constraint on the utility side. Numerical results demonstrate the testbed's capability of handling many EV robots.