Autonomous Robots' AI Should Take Inspiration From the World of Insects, Researchers Argue

Tiny autonomous creatures serve as the perfect model for tiny autonomous robots, the researchers suggest.

A team of researchers from Delft University of Technology, the University of Washington, and the University of Sheffield have put forward a convincing argument that those looking to design the future of small, autonomous robots should take a leaf from nature when developing their artificial intelligences — specifically, insects.

"We argue that inspiration from insect intelligence is a promising alternative to classic methods in robotics for the artificial intelligence (AI) needed for the autonomy of small, mobile robots," the researchers explain. "The advantage of insect intelligence stems from its resource efficiency (or parsimony) especially in terms of power and mass."

The creator of RoboFly is among the researchers arguing for insect inspiration in robotic AI. (📹: University of Washington)

Arguing that traditional approaches are hitting hard limits — in particularly the difficulties in scaling down an AI that works perfectly well in something like a large-scale drone or autonomous vehicle to run on something the size of your finger, and the semiconductor industry's troubles in keeping the performance and efficiency gains coming year-on-year — the team claims that the insect world already has many of the challenges solved.

"We adopt the view that AI is the 'pursuit of intelligent behavior by artificial methods,'" the team explains, referencing a 1993 article in Artificial Life by Luc Steels, "explicitly acknowledging that insect behaviors are intelligent. If we succeed in harnessing insect-inspired AI, small robots will be able to tackle difficult tasks while staying within their limited computational and memory budget."

Guido de Croon's Sniffy Bug swarms to locate gas leak sources and alert users to the danger. (📹: MAVLab TU Delft)

Those behind the paper certainly have experience to back up their claims: First author Guido de Croon recently developed an insect-inspired drone swarm dubbed Sniffy Bug and designed to locate the source of gas leaks; Julien Dupeyroux's AntBot is self-descriptive, using sky polarization for navigation like its six-legged precursors; Sawyer Fuller's RoboFly is likewise obvious in its inspiration, using a fly-like flapping motion; and James Marshall's experiments with swarm-based foraging are directly mimicking insect food-seeking behaviors.

"The right approach is not to implement existing autonomy algorithms in novel processors," the researchers argue. "Instead, the robot engineer will have to strive for the same kind of parsimony that is found in insect intelligence.

Julien Dupeyroux's AntBot navigates like the insect from which its name is taken. (📹: ScienceVio)

"This will be vital for small robots with limited resources, like tiny insect-like flying drones," the team continues, "but it will also be important for larger robots when they have to execute many complex tasks, when their bodies are covered with tiny sensors, and when energy efficiency is an overriding concern. Indeed, in nature, parsimony is not reserved for insects alone; it is a governing principle for all animals."

The full review article is available in the journal Science Robotics under open-access terms.

Gareth Halfacree
Freelance journalist, technical author, hacker, tinkerer, erstwhile sysadmin. For hire:
Latest articles
Sponsored articles
Related articles
Latest articles
Read more
Related articles