Location
GROWBOT: A Grower-Reprogrammable Robot for Ornamental Plant Production Tasks (PhD Studentship)

Research

HNS/PO 194 - GROWBOT: A Grower-Reprogrammable Robot for Ornamental Plant Production Tasks (PhD Studentship)

Start Date: 
01/01/2016
Completion Date: 
31/12/2018
Project Leader: 
Dr Matthew Howard, King's College London
Code: 
HNS/PO 194

Industry Representatives: Bruce Harnett (Kernock Park Plants), Jamie Dewhurst (J&A Growers) and Mike Mann (Winchester Growers)

AHDB Horticulture Cost: £68,577

Summary:

This project will explore the use of new, human-robot interactive, soft robotic systems and their application for semi-automated propagation of multiple varieties of ornamental plants.  It will investigate ways in which non-expert users (i.e., those without technical expertise in robot pro- gramming and control), but that are nevertheless skilled in plant processing, can use robots in their work, to relieve them of the more repetitive, labour-intensive tasks encountered.

Focus will be given to improving efficiency and competitiveness in small/medium scale busi- nesses, typically processing relatively small batches of a wide variety of plants, as opposed to the traditional large facilities specialised in processing large volumes of single-varieties. For this reason, the project will investigate ways of automating tasks that are usually difficult to achieve at small scale, such as taking and inserting cuttings, grading and collating plant specimens. 

To achieve  this,  the project  will draw  on advances  in adaptive  robotics,  soft robot  design, and human-robot  interaction.   New technologies  such as robot programming  by demonstra- tion (whereby robots can be ‘trained’ through demonstrations of a task) will be investigated to enable horticultural workers to program soft robotic manipulators to perform repetitions of hor- ticultural tasks, while minimising risk to plant products or personnel through in-built compliance in the mechanical design.

Complex motions of the hand will be recorded through a soft sensor interface (e.g., a data glove) to enable a soft, light-weight and low-cost robotic manipulator to reproduce the movement. The speed and ease with which such automation can be adapted to multiple plant varieties will be assessed with a view to maintaining quick turn-around when processing small batches of plants.