Control of reaching movement in unpredictably changing environment by constraints emergence and satisfaction

Human reaching movement is accomplished even when environmental conditions unexpectedly change in the real world. Since human arm is a redundant system, constraints must be introduced to determine a unique motor command, and simultaneously those should be changed flexibly to adapt to unpredictably changing environment. In this paper, we have implemented this concept for a reaching model of 3 degrees of freedom arm. During movement, each joint evaluates its efficiency for moving the hand in autonomous decentralized way, and interacts with each other so that instantaneously more efficient one works dominantly. Since these interactions work as real-time constraints which change according to environment, various motion patterns can emerge. Adaptability to changes of arm weight, joint viscosity, and malfunction of actuator has been confirmed and its mechanism was qualitatively and mathematically discussed