Research

 

Research at Lincoln Agri-Robotics centres around three grand challenges and five core technologies:

 

Grand Challenges:

 

Selective Harvesting 

In the UK alone there are 64,000 seasonal migrants in the fresh produce sector picking fruits and vegetables; similar to the number of tractor drivers deployed across the entire industry. Selective harvesting is a high priority but extremely challenging; picking a strawberry requires novel soft or non-contact grippers, active robotics to target occluded fruit, exceptional and dynamic motion control, robotic picking arms need to be deployed on a novel autonomous vehicle that must operate as a fleet and collaborate with humans. Resolving this single problem requires significant innovation across multiple RAAI domains.

 

Crop Care

Traditional agriculture incorrectly assumes all plants within a field are the same and the crop is uniform. Modern precision agriculture techniques drive productivity by optimising the micro-environment of individual plants within crops rather than assuming the crop is uniform as a whole. However, precision agriculture is limited by the phenomenal complexity and scale of agricultural systems, the lack of platforms to deploy novel sensors and the scale of real time data analysis required to deliver effective decision support and action. Field robotics offer a technology step change, robots can deploy novel sensors to measure soil moisture, physics and chemistry, 3D cameras can measure crop growth and architecture, deep learning techniques can identify weeds, pests and diseases. Integrated Agri-food RAAI offers the platforms, data analysis and actuators to realise the potential of precision agriculture to transform resource productivity of agriculture; increasing yields whilst minimising environmental impact.

 

Phenotyping Robotics 

Crop phenotyping is an indispensable technique to investigate physiological principles involved in the control of basic plant functions as well as for selecting superior genotypes in plant breeding programs. Typically, phenotyping is manual, uses a limited amount of sensors on the ground or deployed on drones with limited fields of view. Field robots now offer a new route to automate and conduct massively parallel phenotyping rapidly speeding up the plant breeding process. They can deploy heavy weight state of the art sensors collecting huge volumes of data. Field robotics for phenotyping could transform plant breeding, but significant advances are required in fleet robot control, robot architecture, active perception and sensing, data fusion and retrieval and alongside direct specialised links to basic plant physiology.

 

Core Technologies:

 

Mobile Autonomy 

  • field navigation
  • crop scouting and harvesting

Manipulation and Soft Robotics 

  • picking ripe fruits

Sensing and Perception

  • Sensing using a range of sensor technologies, such as RGB-D, hyperspectral and/or multispectral cameras, weather factors
  • Perception to recognise ripe fruits, weeds, pests, disease; measure soil moisture, soil physics, soil chemistry; and monitor crop growth

Fleet Management 

  • coordinated field mapping
  • heterogeneous team coordination

Human-Robot Collaboration 

  • avoiding humans in the field
  • cooperating with humans for planning, scouting, harvesting and in situ packing

Systems Integration 

Testing and Evaluation are required to put all the pieces together and ensure:

  • intelligent application/deployment of precision agriculture for a range of crops;
  • field testing; and
  • user evaluation.