Core Concepts
A novel iterative geometric method to accurately predict the 3D pose of mobile ground robots with active flippers on uneven ground, utilizing the sub-voxel accuracy of signed distance fields.
Abstract
The paper presents a novel iterative geometric method to predict the 3D pose of mobile ground robots with active flippers on uneven terrain. The approach utilizes the ability of Euclidean Signed Distance Fields (ESDFs) to represent surfaces with sub-voxel accuracy, enabling accurate prediction of the robot-terrain interaction.
Key highlights:
- The method takes the current joint configuration of the robot into account, allowing it to generalize to different robot platforms.
- It is capable of handling multi-level environments, unlike approaches based on heightmaps.
- Evaluation in simulation and on a real robot platform shows the method outperforms a recent heightmap-based approach, especially in challenging terrain scenarios.
- The implementation is made available as an open-source ROS package.
The algorithm consists of two main stages:
- Falling Stage: The robot is dropped from above the ground until the first contact is found.
- Rotation Stage: The robot is repeatedly rotated around the least stable axis until a stable state is found.
The effectiveness of the approach is demonstrated on two different tracked robots, Asterix and DRZ Telemax, in simulation and on the real Asterix platform. Compared to a tracking system as ground truth, the method achieves an average accuracy of 3.11 cm in position and 3.91° in orientation, outperforming the heightmap-based approach.
Stats
The average position error is 3.11 cm and the average orientation error is 3.91° when evaluated on the real Asterix robot platform.
Quotes
"Compared to a tracking system as ground truth, our method predicts the robot position and orientation with an average accuracy of 3.11 cm and 3.91°, outperforming a recent heightmap-based approach."