Researchers are creating an insect-inspired autonomous navigation strategy for small, lightweight robots

Have you ever wondered how insects are able to go so far beyond their home and still find their way? The answer to this question is not only relevant to biology, but also to the creation of artificial intelligence for small, autonomous robots.

Drone researchers from TU Delft felt inspired by biological insights into how ants visually recognize their environment and combine this with counting their steps to get back home safely. They used these findings to develop an insect-inspired autonomous navigation strategy for small, lightweight robots.

The strategy allows such robots to return home after long trajectories, while requiring extremely little computation and memory (0.65 kilobytes per 100 m). In the future, tiny autonomous robots could find a wide range of uses, from monitoring inventory in warehouses to finding gas leaks in industrial sites.

The researchers published their findings in Scientific roboticson July 17, 2024.

Holding on to the little guy

Tiny robots, from tens to several hundred grams, have the potential for interesting real-world applications. Due to their light weight, they are extremely safe, even if they accidentally bump into someone.

Since they are small, they can move in tight areas. And if they can be produced cheaply, they can be deployed in large numbers so they can quickly cover a large area, such as in greenhouses for early detection of pests or diseases.

However, getting such small robots to operate on their own is difficult because they have extremely limited resources compared to larger robots. The main obstacle is that they must be able to navigate on their own. For this purpose, robots can get help from external infrastructure. They can use position estimates from GPS satellites outdoors or wireless beacons indoors.

However, it is often not desirable to rely on such an infrastructure. GPS is not available indoors and can be very inaccurate in cluttered environments such as urban canyons. And installing and maintaining beacons indoors is relatively expensive or simply not possible, such as in search and rescue scenarios.

The artificial intelligence necessary for autonomous navigation with only onboard means was built with large robots like self-driving cars in mind. Some approaches rely on heavy, power-hungry sensors, such as LiDAR laser rangers, that simply cannot be carried or powered by small robots.

Other approaches use the sense of vision, which is a very energy-efficient sensor that provides rich information about the environment. However, these approaches typically attempt to create highly detailed 3D maps of the environment. This requires large amounts of processing and memory that can only be provided by computers that are too large and power-hungry for small robots.

Step counting and visual crumbs

This is why some researchers have turned to nature for inspiration. Insects are of particular interest because they operate over distances that could be relevant to many real-world applications while using very limited sensing and computing resources.

Biologists are increasingly understanding the basic strategies used by insects. Specifically, insects combine tracking their own movement (so-called “odometry”) with visually guided behavior based on their low-resolution but nearly omnidirectional visual system (so-called “gaze memory”).

While odometry is increasingly understood even at the neuronal level, the precise mechanisms underlying gaze memory are still less well understood.

One of the first theories about how this works proposes a “snapshot” model. In it, an insect such as an ant is designed to occasionally take pictures of its environment.

Later, when approaching the image, the insect can compare its current visual perception with the image and move to minimize the differences. This allows the insect to navigate or “home” to the image location and remove any drift that inevitably accumulates just by performing odometry.

“Snapshot-based navigation can be compared to how Hansel and Gretel tried not to get lost in the fairy tale of Hansel and Gretel. When Hans threw stones on the ground, he was able to return home. But when he threw bread crumbs that were eaten by the birds, Hans and Gretel got lost in in our case, images are images,” says Tom van Dijk, first author of the study.

“Just like with a rock, for the image to work, the robot must be close enough to the location of the image. If the visual environment is too different from that at the location of the image, the robot can move in the wrong direction and never come back.” That’s why you need to use enough images – or in the case of Jeniček, throw a sufficient number of stones.

“On the other hand, throwing stones at each other would exhaust Hans’s stones too quickly. In the case of the robot, using too many frames leads to a lot of memory consumption. Previous work in this area usually had the frames very close to each other so that the robot could visually return first to one frame and then to the next.”

“The main insight underlying our strategy is that you can place images much further apart if the robot travels between images based on odometry,” says Guido de Croon, professor of bio-inspired drones and co-author of the paper.

“The guidance will work as long as the robot ends up close enough to the image location, i.e. if the robot’s odometry deviation falls within the ‘capture area’ of the image. This also allows the robot to travel much further, because the robot flies much slower when it is moving into a frame than when it is flying from one frame to another based on odometry.”

The proposed insect-inspired navigation strategy enabled the 56-gram “CrazyFlie” drone equipped with an omnidirectional camera to cover distances of up to 100 meters with just 0.65 kilobytes. All visual processing took place on a small computer called a “microcontroller” that can be found in many inexpensive electronic devices.

Putting robotic technology to work

“The proposed insect-inspired navigation strategy is an important step towards real-world application of small autonomous robots,” says Guido de Croon.

“The functionality of the proposed strategy is more limited than the functionality provided by state-of-the-art navigation methods. It does not create a map and only allows the robot to return to the starting point.”

“However, for many applications, this may be more than sufficient. For example, to track inventory in warehouses or monitor crops in greenhouses, drones could fly up, collect data, and then return to a base station. They could store mission-related images on a small SD card for subsequent processing by the server, but they would not need them for the navigation itself.”