Abstract:Algorithms for state estimation of humanoid robots usually assume that the feet remain flat and in a constant position while in contact with the ground. However, this hypothesis is easily violated while walking, especially for human-like gaits with heel-toe motion. This reduces the time during which the contact assumption can be used, or requires higher variances to account for errors. In this paper, we present a novel state estimator based on the extended Kalman filter that can properly handle any contact configuration. We consider multiple inertial measurement units (IMUs) distributed throughout the robot's structure, including on both feet, which are used to track multiple bodies of the robot. This multi-IMU instrumentation setup also has the advantage of allowing the deformations in the robot's structure to be estimated, improving the kinematic model used in the filter. The proposed approach is validated experimentally on the exoskeleton Atalante and is shown to present low drift, performing better than similar single-IMU filters. The obtained trajectory estimates are accurate enough to construct elevation maps that have little distortion with respect to the ground truth.

What problem does this paper attempt to address?

The paper aims to address the problem of pose estimation in bipedal robots (especially humanoid gait robots) during walking. Traditional methods often assume that the robot's feet remain flat and fixed in position when in contact with the ground. However, this assumption is easily broken during actual walking (especially in humanoid gaits that include heel-to-toe motion). This leads to a shortened effective time for the contact assumption or requires higher variance to compensate for errors. The paper proposes a new method based on the Extended Kalman Filter (EKF) that can handle any contact configuration and utilizes multiple Inertial Measurement Units (IMUs) distributed throughout the robot structure, including IMUs on both feet, to track multiple parts of the robot. Additionally, the multi-IMU setup can estimate deformations in the robot structure, thereby improving the dynamic model used for the filter. Experimental results show that the proposed multi-IMU state estimation algorithm exhibits low drift characteristics on the Atalante exoskeleton robot, with trajectory estimation accuracy superior to that of a single IMU filter. Particularly, the drift in the vertical direction is very small, even in the presence of heel-to-toe motion. By attaching a solid-state LiDAR to the exoskeleton, it is also demonstrated that the trajectory estimated using this method is sufficient to construct an elevation map close to the true height map of the ground.

Multi-IMU Proprioceptive State Estimator for Humanoid Robots