A new pose estimation method based on inertial and visual sensors for autonomous robots

Abstract

The inertial and visual sensors are popular sensors on board for robots. The combination of inertial and visual information is helpful to estimate the pose of a robot robustly. In this paper, a new sensing system with inertial and active visual sensors is proposed. A hierarchical fusion strategy is suggested. And a pose estimation method with optimization fusion is presented, which takes image features as constraints to optimize inertial parameters. It has the potential to estimate the pose faster and more accurate. Simulations are also provided in order to assess the proposed integration method conveniently.