Optimizing self-rotating bristle-bots for active matter implementation with robotic swarms

Robotic swarms have been recently explored as a versatile and scalable alternative to traditional microscale platforms for experimental studies of active matter. These robotic setups consist of either self-propelled or self-rotating particles. In the present paper, we develop and experimentally realize a swarm of self-rotating bristle-bots suitable for a wide range of active-matter experiments. We focus on optimizing the bristle-bot design and controlling the sliding friction between individual robots.

Spatial Structure and Information Transfer in Visual Networks

In human and animal groups, social interactions often rely on the transmission of information via visual observation of the behavior of others. These visual interactions are governed by the laws of physics and sensory limits. Individuals appear smaller when far away and thus become harder to detect visually, while close by neighbors tend to occlude large areas of the visual field and block out interactions with individuals behind them. Here, we systematically study the effect of a group’s spatial structure, its density as well as polarization and aspect ratio of the physical bodies, on the properties of static visual interaction networks. In such a network individuals are connected if they can see each other as opposed to other interaction models such as metric or topological networks that omit these limitations due to the individual’s physical bodies. We find that structural parameters of the visual networks and especially their dependence on spatial group density are fundamentally different from the two other types. This results in characteristic deviations in information spreading which we study via the dynamics of two generic SIR-type models of social contagion on static visual and metric networks. We expect our work to have implications for the study of animal groups, where it could inform the study of functional benefits of different macroscopic states. It may also be applicable to the construction of robotic swarms communicating via vision or for understanding the spread of panics in human crowds.

Human vs. Deep Neural Network Performance at a Leader Identification Task

Control of robotic swarms through control over a leader(s) has become the dominant approach to supervisory control over these largely autonomous systems. Resilience in the face of attrition is one of the primary advantages attributed to swarms yet the presence of leader(s) makes them vulnerable to decapitation. Algorithms which allow a swarm to hide its leader are a promising solution. We present a novel approach in which neural networks, NNs, trained in a graph neural network, GNN, replace conventional controllers making them more amenable to training. Swarms and an adversary intent of finding the leader were trained and tested in 4 phases: 1-swarm to follow leader, 2-adversary to recognize leader, 3-swarm to hide leader from adversary, and 4-swarm and adversary compete to hide and recognize the leader. While the NN adversary was more successful in identifying leaders without deception, humans did better in conditions in which the swarm was trained to hide its leader from the NN adversary. The study illustrates difficulties likely to emerge in arms races between machine learners and the potential role humans may play in moderating them.

Exploring the Effects of Swarm Degradations on Trustworthiness Perceptions, Reliance Intentions, and Reliance Behaviors

Swarms comprise robotic assets operating autonomously through local control laws. Research on human-swarm interaction (HSwI) investigates how human operators collaborate with swarms to accomplish shared goals. Researchers have begun to investigate the role of trust in HSwI, specifically which aspects of robotic swarms affect human trust. Through a human factors lens, the present research builds on earlier HSwI work and investigates the effect of swarm asset degradations on trustworthiness perceptions, reliance intentions, and reliance behaviors. Results showed that trustworthiness perceptions of and intentions to rely on swarms (but not reliance behaviors) were correlated, demonstrating the relation between theoretically relevant antecedents to trust in HSwI contexts. Contrary to past work, the results showed no statistical evidence that asset degradations differentially affect trustworthiness perceptions, reliance intentions, or reliance behaviors. Limitations of the current work (e.g., heterogeneity of post-intervention foraging behavior, sample size) are discussed and followed with future research suggestions.

Convexified Graph Neural Networks for Distributed Control in Robotic Swarms

A network of robots can be viewed as a signal graph, describing the underlying network topology with naturally distributed architectures, whose nodes are assigned to data values associated with each robot. Graph neural networks (GNNs) learn representations from signal graphs, thus making them well-suited candidates for learning distributed controllers. Oftentimes, existing GNN architectures assume ideal scenarios, while ignoring the possibility that this distributed graph may change along time due to link failures or topology variations, which can be found in dynamic settings. A mismatch between the graphs on which GNNs were trained and the ones on which they are tested is thus formed. Utilizing online learning, GNNs can be retrained at testing time, overcoming this issue. However, most online algorithms are centralized and work on convex problems (which GNNs scarcely lead to). This paper introduces novel architectures which solve the convexity restriction and can be easily updated in a distributed, online manner. Finally, we provide experiments, showing how these models can be applied to optimizing formation control in a swarm of flocking robots.

Population model of Temnothorax albipennis as a distributed dynamical system: I. self-consistent threshold is an emergent property in combination of quorum sensing and chemical perception of limited resource

House hunting of ant, such as Temnothorax albipennis, has been shown to be a distributed dynamical system. Such a system includes agent-based algorithm [1], with agents in different roles including nest exploration, nest assessment, quorum sensing, and brood item transportation. Such an algorithm, if used properly, can be applied on artificial intelligent system, like robotic swarms. Despite of its complexity, we are focusing on the quorum sensing mechanism, which is also observed in bacteria model. In bacterial model, multiple biochemical networks co-exist within each cell, including binding of autoinducer and cognate receptors, and phosphorylation-dephosphorylation cycle. In ant hunting, we also have ant commitment to the nest, mimicking binding between autoinducer and cognate receptors. We also have assessment ant specific to one nest and information exchange between two assessment ants corresponding to different nests, which is similar process to the phosphorylation-dephosphorylation cycle in bacteria quorum sensing network. Due to the similarity between the two models, we borrow the idea from bacteria quorum sensing to clarify the definition of quorum threshold through biological plausible mechanism related to limited resource model. We further made use of the contraction analysis to explore the trade-off between decision split and decision consensus within ant population. Our work provides new generation model for understanding how ant adapt to the changing environment during quorum sensing.