Move like an ant
We know that insects aren’t particularly smart by human standards, but despite their limitations, they are capable of incredible feats of organization and direction, making them a source of inspiration for researchers working on microrobots and lightweight drones that grapple with similar constraints.
For example, the desert ant Cataglyphis can forage long distances and then walk straight back to its nest – up to 1 km away.
The low payload and available power preclude solutions used in heavier autonomous systems such as self-driving cars. For example, LIDAR (“laser radar”) is great for creating 3D maps of the environment, but it is too heavy and power hungry. It also requires a lot of computing, which itself requires a lot of memory and processing, is power hungry, and is heavy.
Beacons and GPS signals are alternatives, but they require expensive setups, can be unreliable or simply not possible. Therefore, understanding how insects such as ants and bees navigate the world with only a limited “hardware” and energy supply could help us replicate that in our own creations for robots and drones.
This is the general idea of using biologically inspired robots, a topic we explored further in our article How Robotics Takes Inspiration from Nature, with robots inspired by octopuses, salamanders, snakes and dogs.
Why use microrobots?
Small robots and drones are cheaper to manufacture and can cover larger surfaces at once for the same cost. Their small size also allows them to observe in greater detail without the risk of crashing into their surroundings, for example flying inside a greenhouse to scan plants for early signs of disease or pests.
Or they could be deployed on search-and-rescue missions, exploring ruins and wilderness areas to find people in need. A swarm of such robotic “birds/ants/dragonflies” could quickly locate survivors after an earthquake, for example.
Source: Delft University of Technology
How ants move around the world
One way is through vision, which insects are particularly good at because they have a nearly omnidirectional visual system (they see in all directions at once), but the resolution of this vision is relatively low.
Some of the oldest established theories on how insects use vision to orient themselves are found in the “snapshot model.”
The idea is that an insect’s brain periodically takes a snapshot of its surrounding environment, and when it needs to return “home”, it compares its current environment with the most recently saved snapshot.
This concept is now well understood down to the neuron level, so it could potentially be relatively easy to replicate in a robot.
In theory, this approach may be sufficient, but in practice it has some limitations.
To work well, a very close succession of snapshots is required, and missing even one piece of data can lead to disorientation and cause the robot to become completely lost. The large number of snapshots required would be too much for both the ant’s brain and the robot’s memory.
Adding odometer readings
Another method used by ants, and insects in general, is to track their movements, a method called odometry, which is also used in robotics, but the problem is that it lacks precision: each step is estimated from motion sensors (or subjective perception in the case of ants), but it never fully reflects the actual movement.
As a result, the accuracy of odometry-based location estimates gradually degraded and became inaccurate over time.
For researchers from Delft University of Technology in the Netherlands, combining the two methods was the key insight: In the scientific paper “Visual route tracking for small autonomous robots”, the researchers combine visual snapshots and odometry to increase the autonomy of microrobots.
Performance improvements
This allows the robot to periodically reset the drift of its odometry whenever it finds one of the landmark snapshots.
Source: Science Robotics
At the same time, relying primarily on odometry reduces the need for ultra-close snapshots, allowing microrobots to move faster between points without having to constantly check for visual cues for their trajectory.
“The main insight underlying our strategy is that if the robot moves between snapshots based on odometry, the interval between snapshots can be much greater.
Homing works as long as the robot gets close enough to the snapshot location, i.e., as long as the robot’s odometry drift falls within the snapshot’s catchment area.”
Professor Guido de Kroon.
Using new orientation software that combines snapshot and odometry, the research team tested how much data could be used to orient a robot weighing just 56g over a distance of 100m.
Source: Science Robotics
It’s incredibly small, at just 1.16 kilobytes in size. For reference, the average image taken with a smartphone is several thousand kilobytes each, and most images online are tens or hundreds of kilobytes.
Even better, all the image processing can be done in lightweight minicomputers called “microcontrollers,” which are found in many inexpensive electronic devices.
application
industry
These microrobots and drones have very limited data processing capabilities, with most of the processing power of their on-board microcontrollers being dedicated to managing navigation and data collection.
But such drones can be used to track inventory in a warehouse or monitor crops in a greenhouse. Drones walk or fly to collect data such as photos, bar codes, and RFID tags. These data points can then be stored on a small SD card.
These recordings are then transferred to a larger computer or server where they can be post-processed and converted into useful data.
army
Another potential application area is military technology, especially given the growing importance of drones on the modern battlefield, as the war in Ukraine has demonstrated.
Small flying drones, light enough to fit in a soldier’s pack, could be sent ahead to scout, bringing back pictures of enemy positions to sheltering soldiers.
This region is subject to high levels of electronic warfare (EW) jamming and a high probability of constantly changing conditions, making autonomous drone navigation a must. Light weight and low power consumption will also be key features. In the study discussed here, the drone was able to navigate a 300-meter trajectory in a simulated forested environment.
Source: FLIR
Further research
The strategy of combining odometer measurements with snapshots is very efficient and can be made even more efficient by improving the accuracy of the odometer. The algorithm used can also be tuned to be even more memory power efficient.
Another improvement would be to add collision avoidance capabilities, as the robot already has omnidirectional vision.
A solution still needs to be found if the robot gets lost for some reason. For example, the researchers suggest that the robot could estimate the size of its catchment area online and provide search instructions when it gets lost on its route.
The procedure is particularly well suited to small robots that often struggle to navigate otherwise, but it could start to be applied to larger robots in the future to reduce the need for expensive equipment like LIDAR and lighten computing and power requirements.
Drone and robotics companies
1. Auto Store Holdings (AUTO.OL)
Autonomous vehicles, like self-driving cars, may be on the way, but they’ve been a tricky technology to develop, even for tech leaders like Google and Tesla. But one field is already being revolutionized by self-driving cars and robotics: logistics.
Norway’s AutoStore provides automated warehouses for a variety of industries, including pharmaceuticals, clothing, grocery, aviation, logistics and industrial manufacturers. AutoStore’s three main segments of business are apparel, industrial and third-party logistics companies.
The company’s warehouse uses autonomous robots that can autonomously identify, pick up, and deliver packages and products to their designated location. See them in action in this video.
The company is expanding rapidly after the pandemic as more large companies realize the benefits of building more efficient, resilient, and faster logistics systems. On average, upgrading to an autonomous warehouse takes just one to three years to recoup the initial investment.
AutoStore operates in 50 countries with 58,500 robots in operation for 900 customers, and has seen revenue grow at a compound annual growth rate of 50% since 2017, two to three times faster than the annual growth rate of the automated warehouse market (estimated at 15%).
Source: AutoStore
Like many European technology companies, AutoStore offers highly advanced solutions that are little known to the general public.
Most warehouses will move towards automation, and leaders in this space will likely outperform the sector’s growth as it makes sense to turn to providers who can deploy these solutions at scale and cheaply.
Robots that are more autonomous and more efficient at finding their way around could be both an opportunity and a threat for AutoStore, which currently requires a complete redesign of its warehouses to use its robotics solution.
In the future, robots will be able to find their way without needing the grid that we currently use, making them much easier to deploy, less disruptive to ongoing operations, and requiring significantly less upfront investment, solving what remains a major obstacle to mass adoption of this technology.
Source: AutoStore
2. Zebra Technologies Corporation (ZBRA)
Zebra Technologies makes tracking labels and scanners that can monitor every component in a “smart” factory, including mobile computers, barcode scanners, machine vision, location technology, tags and RFID (radio frequency identification).
This level of data collection and analysis will be a key component in implementing robots outside the assembly line, especially as they become more mobile and flexible.
The company is at the origin of the popularization of CodeBar and since 2018 has been making acquisitions to integrate all the technologies necessary for the “roboticization” and digitalization of modern warehouses and factories.
Photo from Zebra
Currently, the company’s main business areas are e-commerce/retail, transportation/logistics, followed by manufacturing.
Photo from Zebra
With robots becoming central to e-commerce and logistics, the demand for Zebra tracking systems is growing.
For now, there remains a need to make room for larger robots.
Once microrobots weighing just a few tens of grams can roam around and scan RFID tags, we may soon see bee-like drone hives continuously monitor all activity in factories and warehouses, handling the process autonomously.
