Dynamic Motion Planning for Full Body Manipulators

Lead Research Organisation: University of Oxford
Department Name: Autonom Intelligent Machines & Syst CDT

Abstract

This project falls within the EPRSC research areas such as in Artificial Intelligence Technologies, Assistive Technology, and Robotics.

Project Aims and Objectives:
State-of-the art robots still fail in reliably interacting with and manipulating objects in their environment. In particular, there has been limited success in interacting with moving objects, or with static objects while the robot itself is moving. Successful execution of manipulation tasks is a key milestone to improving robot capabilities in dynamic industrial automation processes such as warehouse pick-and-place tasks, as well as moving towards useful assistive robotics technologies such as in a care home or around the common household. My research will address current limitations in motion planning to enable robots to perform manipulation tasks robustly in dynamic environments.

To execute a manipulation task, there is an entire manipulation pipeline, comprising of motion planning, obstacle avoidance, trajectory optimisation and grasp pose synthesis to name a few. In the course of my research I will address the limitations in current manipulation pipelines. To begin this research I will conduct a thorough literature review on motion planning for manipulation tasks. I will look to utilise the latest advances in GPU programming to parallelise the problem and reduce planning times. I will look to use machine learning methods such as neural networks and reinforcement learning to conduct robot learning that can be fused with traditional control methods, as well as be used to develop state-of-the-art grasp pose synthesis (determining how best to grasp an object). Developments and findings in this research will be tested on real robot platforms such as the Toyota Human Support Robot and JACO arm.

Publications

10 25 50

Studentship Projects

Project Reference Relationship Related To Start End Student Name
EP/R512333/1 01/10/2017 30/09/2021
1904254 Studentship EP/R512333/1 01/10/2017 31/12/2021 Mark Finean