Biomimetic robots

A tool to understand collective behavior

Since the 90’s, biologists and computer scientists have collaborated to develop a new scientific discipline: swarm robotics. Computer simulations and mathematical modeling are not sufficient to fully understand the collective behavior of animals. More specific models were thus developed, leading to bioinspired robots that provide biologists with some real answers. They have now revealed, for example, the importance of the geometry of a network for the foraging performance of the Argentine ant colony. In the future, such robots could even have medical applications.

Since the 90’s, biologists and computer scientists have collaborated to develop a new scientific discipline: swarm robotics. Computer simulations and mathematical modeling are not sufficient to fully understand the collective behavior of animals. More specific models were thus developed, leading to bioinspired robots that provide biologists with some real answers. They have now revealed, for example, the importance of the geometry of a network for the foraging performance of the Argentine ant colony. In the future, such robots could even have medical applications.

 

This article also exists in French ("Les robots biomimétiques"), translated by Timothée Froelich.

 

Sources: n0madi, FlickR, CC licence ; © Simon garnier, CRCA, CNRS Toulouse

 

In a recent paper published in PLOS Computational Biology, Guy Theraulaz et al. used micro-robots to reproduce as closely as possible the logic of ants, to understand how they move collectively from their nest to their food source.

 

How robots can mimic ants’ behavior

 

When looking for food, ants explore their environment and create an original trail. Various physical obstacles lead to the creation of bifurcations, or forks in their path, resulting in a real labyrinth. To orient themselves and keep track of their nest, they use different kinds of information: visual, proprioceptive, social and structural. Previous studies with Argentine ants showed that they develop a network with symmetrical and asymmetrical bifurcations, and that they are able to collectively select the shortest route based on pheromones and the branching angle of the trail.

To understand the cognitive process and the role of the angles in route selection, these scientists from the Paul Sabatier University in Toulouse adapted the autonomous miniature robot Alice, developed by Gilles Caprari at the Swiss Federal Institute of Technology Lausanne (EPFL). To study and mimic the ants’ movements, some equipment was added to Alice robots like two light sensors mimicking the antennae and four infrared sensors and transmitters to detect obstacles. The ants’ pheromone trail was replaced by a light trail capable of evaporation if the path was not used. The results suggest that ants do not use a complex cognitive process to travel across bifurcations when following a pheromone trail. In addition, an asymmetrical network favors the choice of the shortest path. For Guy Theraulaz, “the use of these robots to understand the biological phenomenon reveals the role of physics that could not have been detected with computer simulation.”

 

 

Swarm robotics: collaboration between biologists and computer scientist

 

 

Swarm robotics appeared in the 90s as the result of applying the coordination principles of swarm intelligence to collective robotics. Robots and algorithms were thus developed to study the organization of animal groups, particularly birds, fish, and insects like bees, ants and cockroaches. Simon Garnier, a biologist working closely with computer scientists, explained in a review published in 2011 how robotics can contribute to the study of collective animal behavior.

Even if computer simulations and mathematical modeling establish a link between the activities of individual animals and the group behavior, they provide only a general indication. The preparation of robots requires a better knowledge of individual skills and thus, gives a more detailed relationship between these skills and social behaviors. Furthermore, as physical entities, robots can directly interact with their environment (both the physical environment and other animals) and even adapt their behavior to the one of the species being studied. These robots can aggregate or sort some objects, localize a target or collectively transport objects, for example.

The new tools of animal tracking, behavior labeling and analyses developed by computer scientists allow biologists to observe and describe nature in great detail. The results obtained may lead to new ideas and concepts in biology, such as the role of asymmetry in ants’ choice of the shortest path.

The collaboration between biologists and computer scientists will continue to exist but will surely evolve with new findings. Furthermore, some direct applications of their work like nanorobotics will likely be developed. Indeed, based on the various interactions observed, some autonomous groups of nanorobots could be used for surgeries like the removal of blood clots.

 

To know more:

Artificial Intelligence and home automation