Couzin, Iain D.
0000-0001-8556-4558 Berufsbeschreibung
Nachname
Vorname
Name
Suchergebnisse Publikationen
3D-POP : An Automated Annotation Approach to Facilitate Markerless 2D-3D Tracking of Freely Moving Birds with Marker-Based Motion Capture
Recent advances in machine learning and computer vision are revolutionizing the field of animal behavior by enabling researchers to track the poses and locations of freely moving animals without any marker attachment. However, large datasets of annotated images of animals for markerless pose tracking, especially high-resolution images taken from multiple angles with accurate 3D annotations, are still scant. Here, we propose a method that uses a motion capture (mo-cap) system to obtain a large amount of annotated data on animal movement and posture (2D and 3D) in a semi-automatic manner. Our method is novel in that it extracts the 3D positions of morphological keypoints (e.g eyes, beak, tail) in reference to the positions of markers attached to the animals. Using this method, we obtained, and offer here, a new dataset - 3D-POP with approximately 300k annotated frames (4 million instances) in the form of videos having groups of one to ten freely moving birds from 4 different camera views in a 3.6m x 4.2m area. 3D-POP is the first dataset of flocking birds with accurate keypoint annotations in 2D and 3D along with bounding box and individual identities and will facilitate the development of solutions for problems of 2D to 3D markerless pose, trajectory tracking, and identification in birds.
Formation of efficient transportation networks in the Argentine ant
Transportation networks play a crucial role in the success of human societies and ant colonies. Their topology and morphology can dramatically affect the distribution of individuals, materials and information, and the overall productivity of a population. In a recent study, we discovered that the asymmetrical organization of bifurcations in ant pheromone trail networks was responsible for a 3-fold increase in the amount of food transported to the nest compared to a symmetrically organized network. Such asymmetrically organized networks have been found in several ant species but the mechanisms responsible for their formation remain unknown. Here we present for the first time experimental and theoretical evidence that the formation of asymmetrical bifurcations in ant trail networks is caused by a simple interaction between the trail-following behavior of ants and properties of their walking behavior. We measured the trajectories of ants following a pheromone trail that bifurcated with different angles into two trails. For narrow angles, ants tended to alter their course after crossing the point at which the trail bifurcated. For wide angles, they tended to alter their course before the bifurcation point. For intermediate angles similar to the angles found in natural networks, ants altered their course around the trail bifurcation point. We used computer simulations to show that the repetition of this behavior at a bifurcation point leads to a stable, asymmetrical organization of the bifurcation and we explored the impact of the linear and angular speeds of ants on the final shape of the bifurcations. Our results demonstrate that efficient transportation networks can emerge solely from the activity of network users, without prior planning. Uncovering simple mechanisms that lead to the formation of robust and efficient transportation networks is critical to understanding the ecological success of ants and to proposing new solutions for man-made systems.
An Efficient GPU Implementation for Large Scale Individual-Based Simulation of Collective Behavior
In this work we describe a GPU implementation for an individual-based model for fish schooling. In this model each fish aligns its position and orientation with an appropriate average of its neighbors' positions and orientations. This carries a very high computational cost in the so-called nearest neighbors search. By leveraging the GPU processing power and the new programming model called CUDA we implement an efficient framework which permits to simulate the collective motion of high-density individual groups. In particular we ...
Alternating spatial patterns for coordinated group motion
Motivated by recent observations of fish schools, we study coordinated group motion for individuals with oscillatory speed. Neighbors that have speed oscillations with common frequency, amplitude and average but different phases, move together in alternating spatial patterns, taking turns being towards the front, sides and back of the group. We propose a model and control laws to investigate the connections between these spatial dynamics, communication when sensing is range or direction limited, and convergence of coordinated group motions.
I-MuPPET : Interactive Multi-Pigeon Pose Estimation and Tracking
Most tracking data encompasses humans, the availability of annotated tracking data for animals is limited, especially for multiple objects. To overcome this obstacle, we present I-MuPPET, a system to estimate and track 2D keypoints of multiple pigeons at interactive speed. We train a Keypoint R-CNN on single pigeons in a fully supervised manner and infer keypoints and bounding boxes of multiple pigeons with that neural network. We use a state of the art tracker to track the individual pigeons in video sequences. I-MuPPET is tested quantitatively on single pigeon motion capture data, and we achieve comparable accuracy to state of the art 2D animal pose estimation methods in terms of Root Mean Square Error (RMSE). Additionally, we test I-MuPPET to estimate and track poses of multiple pigeons in video sequences with up to four pigeons and obtain stable and accurate results with up to 17 fps. To establish a baseline for future research, we perform a detailed quantitative tracking evaluation, which yields encouraging results.
Sensory networks and distributed cognition in animal groups
Understanding how social influence shapes biological processes is a central challenge in contemporary science, essential for achieving progress in a variety of fields ranging from the organization and evolution of coordinated collective action among cells, or animals, to the dynamics of information exchange in human societies. Using an integrated experimental and theoretical approach, I will address how, and why, animals coordinate behavior. In many schooling fish and flocking birds, decision-making by individuals is so integrated that it has been associated with the concept of a "collective mind". As each organism has relatively local sensing ability, coordinated animal groups have evolved collective strategies that allow individuals, through the dynamical properties of social transmission, to access higher-order capabilities at the group level. However we know very little about the relationship between individual and collective cognition. A major limitation is that it has not been possible to observe directly the pathways of communication, and social networks are typically based on proxies such as spatial proximity among organisms. I will demonstrate new imaging technology that allows us to reconstruct (automatically) the dynamic, time-varying networks that correspond to the visual cues employed by organisms when making movement decisions. Sensory networks are shown to provide a much more accurate representation of how social influence propagates in groups, and one that cannot be captured correctly by social networks based on spatial proximity (regardless of how they are parameterized). I investigate the coupling between spatial and information dynamics in groups and reveal that emergent problem solving is the predominant mechanism by which mobile groups sense, and respond to complex environmental gradients. This distributed sensing requires rudimentary cognition and is shown to be highly robust to noise. I will also demonstrate the critical role uninformed individuals (those who have no information about the feature upon which a collective decision is being made) play in fast, and effective, democratic consensus decision making in collectives.
Emerging collective behaviors of animal groups
Many animal groups routinely make consensus decisions jointly with all group members. This paper builds a novel model merging the locally neighboring reciprocal action and alignment together to investigate the mechanisms of consensus decision-making and its robustness. Our model reveals that the shapes of the coherent flocks are limited in a common narrow interval for different group sizes and information structures. Moreover, the coherent groups display a surprising degree of tolerance against errors, however, they simultaneously show an extremely fragile to attacks. Our model and approach discover some novel phenomena and also reveal some underlying mechanisms of the consensus decision-making and its robustness in biological systems.
An Agent-Based Model to Simulate the Formation and Dynamics of Self-Assembled Structures in Army Ants
Army ants of the genus Eciton are capable of joining their bodies together into temporary self-assembled structures. These structures are highly responsive to traffic conditions and environmental geometry, functioning as a kind of living infrastructure for the colony. With a series of previous field experiments we have examined two kinds of these structures, described in both published and as-yet-unpublished work. I present here an individual-based, spatially explicit model that qualitatively reproduces the formation and dynamics of both of these types of self-assembled structures. The model simulates the emergence of structures from a set of simple individual rules that rely only on local sensing.
Bio-inspired Source Seeking with no Explicit Gradient Estimation
Inspired by behaviors of fish groups seeking darker (shaded) regions in environments with complex lighting variations, we develop distributed source-seeking algorithms for a group of sensing agents with no explicit gradient estimation. We choose a baseline for agent groups and decompose the velocity of each agent into two parts. The first part, which is perpendicular to the baseline, is chosen to be proportional to the measurements, agreeing with observations from fish groups. The second part, which is parallel to the baseline, can be designed to control the relative distances among the agents. This decomposition is leveraged to implement formation-maintaining strategies and source seeking behaviors for the entire group. We prove that the moving direction of a group will converge towards the gradient direction while the formation is maintained.
Spatial models of bistability in biological collectives
We explore collective behavior in biological systems using a cooperative control framework. In particular, we study a hysteresis phenomenon in which a collective switches from circular to parallel motion under slow variation of the neighborhood size in which individuals tend to align with one another. In the case that the neighborhood radius is less than the circular motion radius, both circular and parallel motion can occur. We provide Lyapunov-based analysis of bistability of circular and parallel motion in a closed-loop system of self-propelled particles with coupled-oscillator dynamics.