Motion Trajectory Perception of an Earthworm-Like Robot Using an Inertial Measurement Unit and Autonomous ZUPT

Taking inspiration from natural organisms, the field of bioinspired robotics is advancing rapidly. Earthworm-like robots, in particular, have undergone significant development, progressing from linear to planar motion. However, challenges arise due to surface-dependent slippage and imprecise actuation, complicating the creation of accurate trajectory perception models for different locomotion patterns of earthworm-like robots on varying terrains. In this study, we introduce a novel model-free method that leverages an inertial measurement unit (IMU) and autonomous zero velocity interval (ZVI) detectors, which greatly enhances the autonomous and precise trajectory estimation capabilities of earthworm-like robots. Our technique features a multicondition ZVI detection mechanism that incorporates proprioceptive sensor information generated during motion while interacting with the surface. This approach addresses oscillations in the zero-speed interval, thereby improving the precision of ZVI detection. Furthermore, by combining a zero velocity update (ZUPT) method with a lightweight error state Kalman filter, our algorithm enables the robot to perceive its motion trajectory with centimeter-level accuracy across various gait patterns. Comparative evaluations with other algorithms underscore the remarkable performance of our method. Notably, our approach can even estimate the slipping behavior of the anchored robot segment, offering valuable insights for future robot control and planning.