I. Introduction
BRAIN machine interfaces (BMIs) are an innovation in neurotechnology, and have the potential to be a significant aid for those with neuromuscular disabilities. Neural decoding of sensory-motor signals is a necessary step toward this goal. Initial online implementations of BMIs have demonstrated the ability of control for cursor positions or a single robotic movement depending on sensory-motor signal was reported in [2], [3], [4]. Further developments for controlling a real prosthetic arm or sequence of movements are presented in [5], [6], [7], [8], [9]. Instead of focusing on motor movement reconstruction, current research focuses on the development of a system for real-time interaction with the environment and to achieve higher level performance [10], [11], [12].