Most autonomous vehicles are designed to work best in predictable environments with a defined infrastructure – roads, lanes, crossing points – but this repetitive uniformity rarely exists in farming.
So innovation specialists Cambridge Consultants combined a number of techniques to develop Mamut, a new self-drive solution for agriculture and will be talking about it in Agri-Tech East’s Innovation Insights at Fruit Focus.
Mamut is an AI-powered robot that, equipped with an array of sensors, maps and navigates rows of a field, orchard or vineyard to capture detailed data at plant level, below the canopy, that current drone monitoring methods cannot access.
Mamut robot for data capture
Cambridge Consultants’ Chris Roberts explains: “Agriculture presents very different challenges for autonomous vehicles. Roads are well-defined, whereas fields are irregular, and solutions such as the virtual walls used by robotic lawnmowers won’t work in an agricultural context. Fields are also wet and dirty making visual navigation hard, and many sensors used in autonomous navigation, such as LIDAR, are fragile, expensive bits of kit with lasers and rotating mirrors that don’t work very well in the rain and dust!”
So, Cambridge Consultants are trying out a different approach and is trialling a range of different sensors and cameras, GPS and other navigation units on Mamut.
Chris continues: “Each sensor is used to compensate for the weaknesses of the others. For example, wheel odometers give you a sensible baseline of distance covered but can ‘slip’ through spinning or by hitting bumps. Equally, GPS can ‘jump’ because of branches overhead. So we’ve combined the outputs and employed mathematics to blend the sensors together for more accurate navigation. This approach is called sensor fusion.”
Meet soft hand Hank
As Head of Industrial Robotics at Cambridge Consultants, Chris says that many of the things people find easy, robots struggle to do: “Robots have no sense of touch or feel for things, so they can pick something up and then not realise they have dropped it, blindly holding the arm out and carrying on with its task.”
Cambridge Consultants believes it has the answer with Hank, a soft robotic hand. The silicon gripper inflates like a tentacle without hard ‘bones’ or joints and, using the latest in sensing, they have developed a hand that can ‘feel’.
Chris thinks this development could have applications in fruit production: “There isn’t a blueprint for a bunch of grapes, everyone is different, so you need a robotics system that can deal with the range of variation you find in natural items. Hank has three different sensors that can hold a piece of fruit, which is relatively heavy, squeezing it just enough to pick it up while not gripping it too hard because it’s fragile as well.
“I think Hank’s dexterity and flexibility could be used at any point in fruit picking or production, but, even with a high value crop like fruit, you’d need to consider the commercial benefits of automating specific parts of the process. “
Innovation Insights at Fruit Focus
Chris will be speaking about Mamut and Hank in a lightning presentation as part of Innovation Insights, hosted by Agri-Tech East, at Fruit Focus 2019. This year’s event is held at NIAB EMR, East Malling Trust Estate in Kent on 17 July, 9am – 5pm. Limited free tickets are available for members; please contact us for the discount code.