Graph-Based Proprioceptive Localization Using a Discrete Heading-Length Feature Sequence Matching Approach

Proprioceptive localization refers to a new class of robot egocentric methods that do not rely on the perception and recognition external landmarks. These are naturally immune bad weather, poor lighting conditions, or other extreme environmental conditions may hinder exteroceptive sensors such as camera laser ranger finder. depend proprioceptive inertial measurement units and/or wheel encoders. Assisted by magnetoreception, can provide rudimentary estimation vehicle trajectory which is used query prior known map obtain location. Named graph-based localization, we low cost fallback solution for under challenging conditions. As robot/vehicle travels, extract sequence heading-length values straight segments from match with preprocessed graph (HLG) abstracted localize graph-matching approach. Using information HLG, our location alignment verification module compensates drift, slip, tire inflation level. We have implemented algorithm tested it in both simulated physical experiments. The runs successfully finding continuously achieves accurate at level allows (less than 10 m).