This paper introduces a comprehensive classification system to assess various uncrewed ground vehicle (UGV) deployments reported in the literature. The authors categorize motion distortion features into two groups: internal (related to the vehicle properties and commands) and external (related to terrain properties).
The authors present a mapping of UGV deployments based on the maximum kinetic energy of the vehicles and the complexity of the terrains used in the experiments. This mapping highlights the need for a metric that can compare model performances across different motion distortion conditions.
The authors propose a motion distortion metric that quantifies the difference between the observed velocity and the ideal slip-less velocity computed by a nominal model. This metric provides a way to compare the difficulty of motion datasets without explicitly measuring all the motion distortion features.
The authors validate the proposed metric on four datasets collected with two different UGVs (Clearpath Husky and Warthog) navigating on various terrains (tile, snow, gravel, and ice). The results show that the motion distortion metric increases with both the internal (vehicle kinetic energy) and external (terrain complexity) motion distortion features.
The authors also discuss the lessons learned from their previous field deployments, highlighting the importance of considering factors like temperature, ground properties, and energy/time budgets when characterizing UGV motion in remote and off-road environments.
To Another Language
from source content
arxiv.org
Deeper Inquiries