Inducing Human-like Motion in Robots

Abstract

The paper's objective is to reduce dynamic rigidity and real time computation in robotic systems. Dynamic rigidity has been a key issue in the field of robotic motion. Most methodologies in this field focus on optimal navigation strategies. These strategies are complex and require enormous amount of computation. The proposed system handles these concerns using the following process: first, divide motion into 'n' time steps; then, acquire approximate to precise knowledge of the environment with respect to each time step (t1 to tn) and generate kinematic response to stimuli for respective (tn)s using the concept of muscle memory. Approaches such as Moving Average Smoothing and Motor Learning have been brought to use for the same. Since mobile robots are based on real-time application, response time has also been taken into consideration. Simulations support the proposed theory and show drastic reduction in computation time, response time and dynamic rigidity.