Straight-Line Autonomous Walking for Quadrupedal Robots

The autonomous walking of quadrupedal robots is currently realized mainly through perception, map building, localization, and planning. It also requires expensive hardware such as LIDAR, cameras, Graphics Processing Units(GPU) and so on. The pose of a quadruped robot is mainly estimated from the fusion of an inertial measurement unit(IMU) and kinematics. The integration aspect of the algorithm makes the pose estimation subject to time drift, and the control accuracy of the single-joint motors also affects the accuracy of the pose estimation. In addition, a study shows that the world's top quadrupedal straight-line walking more than 20m or so can be clearly seen lateral offset, its path is not straight, there is a tendency to change to the arc. To address the above issues, we propose a straight-line autonomous walking motion control method based on the position information estimated by IMU and kinematics of a quadruped robotic dog platform, using the coordinate point of the initial position of the autonomous straight-line walking function initiated as the reference system.The quadruped robot developed by Unitree B2 was selected as the experimental platform. The experimental results show that by using a two-dimension plane of joint control and the start point reference system, the method proposed in this paper realizes the autonomous straight-line walking, which enables the quadrupedal robotic dog to ensure that the position is on a straight-line in a Cartesian coordinates coordinate system, and that the heading of the quadruped is in the same direction as that of the straight-line. This improves the system straight-line autonomous walking accuracy, reduces the hardware cost, and reduces the algorithm cost.