Deep Weakly Supervised Positioning

TL;DR

DeepGPS introduces a weakly supervised learning framework for robot positioning that leverages wheel-encoder data as constraints, enabling accurate localization without requiring ground truth positions and reducing calibration effort.

Contribution

The paper presents DeepGPS, a novel framework that trains PoseNet using constraint-based weak supervision from wheel-encoder data, eliminating the need for ground truth positions and manual calibration.

Findings

Achieves less than 2% relative positioning error.

Works effectively on both simulated and real datasets.

Requires minimal human intervention for auto-calibration.

Abstract

PoseNet can map a photo to the position where it is taken, which is appealing in robotics. However, training PoseNet requires full supervision, where ground truth positions are non-trivial to obtain. Can we train PoseNet without knowing the ground truth positions for each observation? We show that this is possible via constraint-based weak-supervision, leading to the proposed framework: DeepGPS. Particularly, using wheel-encoder-estimated distances traveled by a robot along random straight line segments as constraints between PoseNet outputs, DeepGPS can achieve a relative positioning error of less than 2%. Moreover, training DeepGPS can be done as auto-calibration with almost no human attendance, which is more attractive than its competing methods that typically require careful and expert-level manual calibration. We conduct various experiments on simulated and real datasets to demonstrate the general applicability, effectiveness, and accuracy of DeepGPS, and perform a comprehensive analysis of its robustness. Our code is available at https://ai4ce.github.io/DeepGPS/.