Visual–Inertial Fusion With Depth Measuring for Hyper-Redundant Robot’s End Under Constrained Environment

Hyper-redundant robots (HRRs) are used for operations in constrained environment, such as wing rib compartments. Precise perception of the robot end is a fundamental requirement for accurate operations. Conventional sensing schemes cannot be used due to environment constraints. The environment necessitates the implementation of a real-time, self-sensing, multimodal fusion framework. We proposed a visual–inertial sensor fusion method at the robot end for HRRs under constrained environment. It includes a lightweight device and an algorithm based on visual–inertial odometry (VIO), adding a depth-measuring module with laser spots. This module provides real-time scale factors for monocular cameras without adding sensors. It is used to synchronously acquire the trajectory of an HRR end under constrained environment. Experiments on an HRR verified the accuracy and real-time performance. The root-mean-square error (RMSE) is 15.2 mm, and the total processing time is 28.04 ms. This method can be applied to HRRs’ end and enhance the accuracy of robot operations.