Real-time (re-)planning of redundant robots with task-priority in a dynamic and constrained environment

This project proposes a novel planning algorithm based on merging a sampling-based global planner with an inverse kinematics-based controller, defined local planner. The idea is to run off-line the global planner before the motion starts. Then, during the motion, the local planner is executed and simultaneously at lower frequency the global one is performed to capture eventual environmental changes. In this way, the advantages of both the planners are collected. Indeed, the local one ensures the feasibility of trajectories and whereas the global one ensures of avoiding the local minima. Thus, the robot system motion performance does not present the issues that generally result separately applying the two planners.

Faculty Supervisor:

David Meger

Student:

Daniele Di Vito

Partner:

Kinova Robotics

Discipline:

Computer science

Sector:

Information and communications technologies

University:

Program: