Evolutionary trajectory planning For intelligent robots

Abstract

Robots had already made considerable impact the way we understand newlineour world and live our lives However their impact and use is limited by the newlineskills they possess Currently deployed autonomous robots lack the newlinemanipulation skills possessed by humans To achieve general autonomy and newlineapplicability in real world robots must possess such skills Autonomous newlinemanipulation requires algorithms that rapidly and reliably compute collisionfree newlinemotion for robotic manipulators with many degrees of freedom newlineUnfortunately adequate algorithms for this task do not currently exist newlineAlthough there are many dimensions of the real world planning task that newlinerequire further research the central problem of reliable real world planning is newlinetrajectory planning Clearly an autonomous robot must be able to plan its own newlinemotions in order to accomplish its task The classic trajectory planning problem newlinecan be described roughly as follows to find a path that connects both initial and newlinefinal configurations of the robot and satisfies some objectives Minimization of newlinetime torque mechanical energy actuator efforts joint accelerations jerks and newlinevibrations singularity avoidance obstacle avoidance criteria etc and robot newlinekinematics constraints dynamics constraints obstacle avoidance constraints newlineand payload constraints etc newlineThe basic theme of this research work is to create a general newlineprocedure to do optimal trajectory planning of robot manipulators under newlineconstraints These constraints are related first to joint limits of the robot newlineend effector and then to the obstacles lying in the robot environment with newlinewhich the robot must not collide The solution to this problem is very newlineimportant in many industrial applications such as material handling arcwelding newlinepainting gluing etc newline newline