Cerebral Palsy (CP) is the most common cause of permanent serious physical disability in childhood. New strategies are needed to help promote, maintain, and rehabilitate the functional capacity of children with severe level of impairment. Overground walking rehabilitation devices appear as an alternative treatment for improving gait performance as well as training natural gait patterns among this population. The main objective of this work is to present a Human-Robot interaction strategy for overground rehabilitation to support novel robotic-based therapies for CP rehabilitation. This strategy is implemented in a new robotic platform named CPWalker. In our approach, legs' kinematics information obtained from a Laser Range Finder (LRF) sensor is used to detect the user's locomotion intentions and drive the robotic platform. The controller continuously adjust robot's velocity to human velocity achieving an adequate robot motion that assists the locomotion at each step. During a preliminary validation we observed that our strategy is able to fast adapt to patients and provide them a stable gait pattern at different speeds. As a result, the proposed controller is able to provide a natural interface between the robotic-platform and the patient.