Development of a multi-modal manipulator for an urban litter collection robot

Litter is ubiquitous in urban environments, with items such as cans, bottles, food wrappers, cigarette butts and chewing gum being some of the main culprits. Such items are not only detrimental to the appearance of any city, but can also represent a risk to infrastructure (e.g. by blocking drainage), health (e.g. by attracting disease carrying vermin) and the environment (e.g. by finding their way into rivers). Any city that attempts to keep this litter under control will spend a significant amount of money doing so, and will still be fighting an up-hill battle. The starting point for the research to which this PhD will contribute is the hypothesis that autonomous mobile robots could one day solve the problem of urban litter.

Our long-term goal is to develop a heterogeneous robot swarm capable of cleaning up the most common forms of litter from all areas of a typical city, including hard-to-reach locations (e.g. in corners, behind obstacles or in bushes). The first steps towards with goal will be taken by a pair of PhD students working on different aspects of the challenge. The other student (who is not EPSRC-funded) will focus primarily on high-level control of the swarm, addressing issues including resource management, task allocation and navigation, while Nicholas will focus primarily on the mechanical platform as outlined below. The diverse nature of litter, and the unpredictable environment in which it is found, make this a demanding application for a robotic manipulator. As such, the main emphasis of the work is on the design of the manipulator. The mobility system will probably rely on more conventional solutions and may even be an off-the-shelf platform.

Aim:
The aim of this project is to develop a multi-modal robotic manipulator capable of handling a diverse range of litter types and operating in an uncertain real-world environment.

Objectives:
Identify the common types of litter found in urban environments and characterise their mechanical properties as these relate to the manipulation task.
Identify potential strategies for picking up objects, including conventional grasping, suction-based grasping and standard "vacuuming".
Identify the kinematic and control requirements of a manipulator for handling litter in urban environments.
Identify candidate solutions for the manipulator, including conventional "hard robotics", soft robotics and hybrid approaches.
Design a complete manipulator system based that satisfies the kinematic requirements and incorporates multiple modalities for handling a wide range of objects.
Develop a control system for the manipulator that adaptively engages the different handling modalities as needed.
Integrate the manipulator with a suitable mobile platform.
Evaluate the system, first in a controlled lab environment and then in the real world.

Applications and Benefits:
Although the manipulator will be designed for the specific application of litter handling, a robust multi-modal manipulator could be useful in a wide range of alternative applications ranging from domestic service robots to agricultural robots.
The key benefits of this research (as applied to the chosen application) will ultimately be cleaner cities at lower cost.