Honeybees being trained to participate in an experiment by offering them a reward of sugar solution. Individual bees can be identified, and their progress monitored by painting them with variously coloured dots, using non-toxic dyes.
Image: Dr Marie Dacke
First published in the Canberra Times.
Air travel is likely to become a great deal safer, more precise and efficient in future – thanks to the humble honeybee.
From take-off to landing, both piloted and pilotless aircraft as well as ground vehicles are now starting to employ the vision and navigation strategies of these remarkable flying insects.
A wide range of aerial tasks that are repetitive or dangerous for humans, such as checking reservoirs, inspecting power transmission lines, bushfire and weather monitoring, mapping and exploring and crop dusting as well as defence roles may soon be carried out based on what we have learned from bees. One day their abilities may even help us to explore the red planet, Mars.
For 25 years Professor Mandyam ‘Srini’ Srinivasan has pursued the secrets of honeybee vision, flight and navigation with an abiding passion and boundless curiosity, from Yale University in the US, to the Australian National University and, most recently, at The University of Queensland. In the process, he and his colleagues have made discoveries that are today helping to transform travel, transport and robotics as we know them.
“The more we studied these little creatures, the more we discovered how clever and capable they are. They can navigate their way for many kilometres across the countryside, with a brain only the size of a pinhead – but with formidable reliability, as they come and go from the food source. They can control their flight posture, attitude and landings with exquisite precision and grace.”
“We also realised that bees, compared to other animals, are much easier to train and observe,” he says.
In Srinivasan’s spacious new laboratory at The Vision Centre (The ARC Centre of Excellence in Vision Science) and The Queensland Brain Institute, honeybees roam in and out at will.
“The simplest way to lure a bee into the room is to place a sugar water feeder at the window. Once bees notice this, they will come back again and again to the food source. If you want them to join your experiments, you slowly shift the feeder across the room, moving a few feet every hour. The bees, by following the feeder, will continue to move as well. Once the experiments are finished and we stop offering them food, they leave to look somewhere else.”
Bees have two compound eyes, each containing about 5000 tiny lenses called ommatidia. These collect information about the intensity and colour of light arriving into the lens, each ommatidium creating a small image like a pixel in a digital photo. “The compound eyes of a honeybee capture and represent the world as a panoramic pointillistic image, made out of thousands of distinct coloured dots, which give the insect a colourful, nearly all-around view of its surroundings,” Srinivasan explains.
But compound eyes have drawbacks, one being the inability to focus on objects. “They can perceive rapid movement, but they can’t see as much spatial detail in the world as we can,” he says.
Despite these shortcomings, honeybees have surprised the researchers with their intelligence and the ingenious ways they deal with everyday tasks. Along with other flying insects, they possess skills that demand extraordinary coordination and accurate navigation, like regulating flight speed, avoiding objects, orchestrating flawless landings and travelling to and from the hive unerringly.
“For instance, when bees fly through a window or a tunnel, you’ll find that they travel straight down the middle, without veering to either side. They can also weave quickly between tree branches or return safely to the hive from 10 kilometres away.”
“Being tiny has forced these insects to evolve eye designs and nervous systems that perform the required manoeuvres and calculations using strategies that are simple, elegant and often, unexpectedly novel.”
Honeybees rely on several skills to achieve such feats. When a bee flies in a straight line, the images of objects in the environment move past their eyes at speeds that indicate how far away they are. Distant objects travel more slowly, and nearby objects speed by, generating rapid image motion. This is known as optic flow, and insects fly and navigate safely by constantly monitoring optic flow.
Experiments run by Srinivasan’s research team confirm that honeybees depend on optic flow to avoid obstacles, weave through narrow gaps, regulate flight speed, execute smooth landings and gauge how far they have travelled to reach a food source.
For example, they trained honeybees to fly along a narrow tunnel with vertical stripes on both walls. The bees aligned themselves in the tunnel in such a way as to ensure that the images passed at the same speed in both eyes, achieving a balanced optic flow. This ensured that the bee flew safely down the middle of the tunnel, without bumping into either wall. When the researchers moved the vertical stripes on one wall against the bee’s direction of flight, the bees veered away from that particular side, as the movement had created a stronger image flow. This demonstrated how bees sense and avoid collisions with dangerously close objects.
The same concept is used to regulate flight speed, Srinivasan says. Bees were found to fly faster when the vertical stripes on two tunnel walls were moving in the same direction as the bee’s flight, and slower when the stripes moved in the opposite direction. “This means that when bees fly through a narrow passage, the strong image flow warns them that they’re close to other objects, and this causes them to slow down,” he says.
Another important feature of using optic flow as a guide is that it enables safe landings. By measuring the optic flow generated by the ground as they fly over it and adjusting their flight speed to hold the optic flow constant as they approach their target, landing bees slow down progressively as they near the ground, reaching hover just prior to touchdown.
“The simplicity and elegance of this biological autopilot is that they do not require knowledge of the height above the ground, or the speed of flight – all they need to measure is the optic flow generated by the ground,” Srinivasan says.
Bees use a similar tactic to estimate the distance that they have travelled even over many kilometres, Srini says. “Instead of measuring energy consumption or time of flight, or counting wing beats, they only need to sum up the movement of the image of the environment through they have flown.”
Apart from using optic flow to deal with their everyday tasks, another important strategy that bees adopt is using the sky as a celestial compass – based on patterns of polarised light which they can see, but humans can’t. The ommatidia located in the upper parts of their compound eyes are highly sensitive to the orientation of this polarised light, endowing the bees with a ‘sky compass’ that they can use for their navigation.
Besides a compass, having a fixed image of the sky and horizon is equally important in navigation, he says. Three simple eyes known as ocelli, located in the crown of a bee’s head, work like camera lenses to show the bee its current orientation relative to the horizon. Honeybees combine all three of these navigational aids - optic flow, polarised light and orientation - to calculate where they are, how they are oriented, and where they are going. “To find their way home safely, they have to know not only how far they have travelled, but also the direction in which they have moved. Thus, they estimate their heading and set their course by using the sun or the polarised pattern in the sky as a compass, along with the location of the sky and the horizon.”
Thanks to Srinivasan and his team, many of the remarkable strategies used by honeybees to navigate are now being introduced to both land vehicles and aircraft, using computers. This has not only enabled researchers to test how insects see and navigate under real world conditions, it also has also provided new solutions for the design of navigation systems for robotic and piloted vehicles.
His team at The Vision Centre has produced several prototype pilotless aircraft (known as Unmanned Aerial Vehicles or UAVs) that can not only travel collision-free, but also autonomously control their speed, their height above the ground, and their direction of flight autonomously. They can also perform extreme aerobatic manoeuvres, as well as land safely, without the aid of a pilot.
The idea of combining Srinivasan’s findings on bee vision with robotics was originally proposed by robotics scientists in Italy, the US and Australia, who suggested that a robot might be able to steer its way along corridors using optic flow to guide it, he recounts.
“They showed that by balancing the speed of the images on both walls, as bees do, robots could actually progress along the middle of the corridor without bumping into the walls,” he says. Srinivasan’s laboratory then used a similar approach to train a terrestrial robot to compute where it was at all times in relation to its starting point, and return to its starting point successfully without being able to recognise any landmarks in the environment – simply by playing back the optical memory of its outward course.
The use of insect vision and navigation received a major boost when defence organisations around the world became interested in its possibilities for designing the next generation of aerial drones and unmanned battlefield vehicles – and at this point serious funds began to flow for what had, until then, been a fairly obscure branch of entomology.
The researchers used these resources to pioneer the science of bio-engineering and solve the robotics navigation challenge. This led to the successful launch of the first “bee robot”, a pilotless aircraft built to fly at a constant, low height above the ground which was calculated and controlled by the optic flow that it sensed using a fisheye camera lens as an eyeball.
While it is difficult to imitate the exact way that a honeybee lands with a fixed-wing aircraft, the researchers have designed a similar system that holds the optic flow from the image of the approaching ground constant as the aircraft’s altitude drops. This in turn is used to throttle back and position the plane, ensuring a smooth touchdown. In time such systems will probably be used to improve the precision and safety of landings in commercial aircraft, and even to prevent collisions between motor vehicles.
The holy grail for an unmanned aircraft is the ability to roam around autonomously, Srinivasan says. “The main difference between the kind of UAVs that we work on, and the well-known “drones” is that our UAVs are designed to work without any GPS navigational information, that is, in the absence of any external assistance. GPS information on its own will only tell a drone where it is located in the world, but it will not prevent the drone from colliding with a mountain or other obstacle. This is why it’s much more reliable to use an on-board visual navigational system so the aircraft can see what is going on around it. This is especially important when the UAV has to explore locations that are completely new, for which no detailed maps are available.” This will particularly be the case in exploring the surface of the planet Mars, which US space agency NASA plans to do using fleets of small unmanned flyers that may well be equipped with bee-like vision.
“Insects are surprisingly unhampered by their miniscule brains and limited processing capacity, when it comes to tasks that require intricate visual-motor coordination,” Srinivasan says. “Instead, they have evolved simple, elegant and alternative solutions, and these insights are now being used to develop new, biologically inspired strategies for the guidance of autonomous, airborne vehicles.”
“With the correct technology, one can imagine a whole range of other applications that are based on how these tiny creatures cope. These might include steering a car autonomously through a tunnel, sensing other moving objects on the road and avoiding imminent collisions.”
For Srinivasan, the path from honeybee research to the building of sophisticated airborne robots has been a case of serendipity. “When I embarked on this research, I only wanted to find out how bees work. I had no intention of applying the results to robotics. I just wanted to discover how a honeybee lands so delicately on a flower. Or, when it flies in through a window, how it manages to position itself exactly in the middle. These seemed to me wonderful questions to answer.”
Like insulin, penicillin and so many of the greatest discoveries of science, the inspiration arose out of simple curiosity about how the world works. And that is what drives all great science.