Integration of iGPMP2 and iSDF for Manipultor Motion Planning in a Partially-Unknown Semi-Dynamic Environment

Abstract

The use of high degree-of-freedom (DoF) manipulators in industrial manufacturing is well established. Traditionally, manipulators operate in isolated, well-defined work cells with little need for flexibility. Increasingly, however, manipulators are being extended to tasks in unpredictable and, at times, dynamic environments. This thesis presents an approach which enables the Franka Emika (FE) Panda 7-DoF manipulator, to which an Intel RealSense D435i RGB-D camera is mounted, to navigate in partially-unknown semi-dynamic environments. This goal is achieved through the integration of the popular optimization-based motion planner iGPMP2 with a neural network (NN) based environment estimation module iSDF.

The main contributions of this thesis are the proposed modifications to iSDF and iGPMP2 to address their limitations in the application considered in this thesis, the integration of iSDF with iGPMP2, and the experimental validation of the proposed approach. The proposed modification to iSDF extends iSDF for use in non-static environments through the identification ...