Control Strategy Of Biped Robots Based On Underactuation

Bipedal locomotion is an important symbol of human race.Compared with other forms of locomotion,bipedal locomotion has a better adaptation to the environment.Scholars believe that bipedal locomotion should be one of the greatest features of the robots.As the biped robots share similar motion style with human,they are suitable to the environment in which humans live and work.As a result,biped robots are the optimal robots to work instead of human.However,high energy consumption of robot has become an obstacle for further development of biped robots.Un-deractuated biped robots have great advantages over other robots in locomotion efficiency,how-ever,to ensure the walking stability of this kind of robots is always a difficult problem in control area.To overcome the difficulties of efficiency and stability,this paper studies the high efficient walking style design method and proposes a new control strategy based on underactuated robot model.With the proposed methods,the efficiency and stability of bipedal walking are improved.The contributions of this paper are listed in the following aspects.Firstly,aiming at the poor efficiency of bipedal walking,this paper presents a walking style design method based on the compass like robot with feet.Allowing for the foot rotation,constant control inputs are applied to achieve a high efficient walking limit cycle gait with the help of a mechanical latch in the hip joint of the robot and elastic energy storage elements in the ankle joints.By comparing with the walking gaits generated by the method of HZD,the efficiency of the proposed walking gait is verified.Poincare return map is used to check the stability of the limit cycle.For the unstable walking gaits,an event-based control law can be applied to stabilize them.By these works,a stable efficient walking style can be finally achieved.Secondly,for the planar biped robots with one degree of underactuation,this paper studies the control strategy to track the reference limit cycle.The underactuated freedom makes the sys-tem harder to be controlled.Considering the dynamic features of underactuated robots,this paper proposes a transition method to track the angular momentum of the system instead to reduce one dimension of the system helping simplify the problem.For the lower dimensional system,a con-trol law combined approximate linearization and active disturbance rejection control(ADRC)is proposed.When the error is less than the threshold,the error of the system model can be directly ignored,and a linearization is used to design the control law.When the error is bigger than the threshold,the ESO is used to observe the system model error,and compensation is made to reduce the model error and to help design the ADRC.By this control strategy,the limit cycle is stabilized.By comparing with the event-based control law,the advantages in both convergence speed and domain are confirmed.Finally,for the poor adaptability to uneven terrain of underactuated robots,this paper studies the switching control strategy for robots walking efficiently on uneven terrain.In this paper,time scaling method is applied to generate the walking gaits for robots on specific slopes.Based on these efficient walking gaits,controlled symmetries is induced to generate interim walking gaits on other slopes.Switching control law is then proposed using both time scaling control and feedback linerization.By switching the control law and the reference walking gaits on different slopes,a compass-like robot can achieve an efficient walking style on uneven surfaces.Based on NAO robot,walking gaits are designed by time scaling method on level ground,uphill and downhill slopes.Controller has been designed to track these reference walking gaits from a static initial condition.Walking experiments implemented on NAO robot verify the validity of the gait design and control method.