Article
Biochemistry & Molecular Biology
Leo Clement, Sebastian Schwarz, Antoine Wystrach
Summary: We found that visually navigating ants display regular lateral oscillations coupled with variations of forward speed to optimize distance covered and scan left and right directions. This pattern of movement is produced endogenously and conserved across different ant species. The amplitude of the oscillations can be modulated by visual cues to adjust exploration/exploitation balance.
Article
Behavioral Sciences
Sonja Eckel, Martin Egelhaaf, Charlotte Doussot
Summary: In the search for their nest, bumblebees rely not only on visual cues, but also on natural scent marks to pinpoint potential locations. This finding highlights the crucial role of odor in helping bees find their way back to their nest.
FRONTIERS IN BEHAVIORAL NEUROSCIENCE
(2023)
Review
Biology
Ayse Yilmaz, Johannes Spaethe
Summary: Ants are ecologically important insects with impressive capabilities for visual learning and orientation, able to discriminate between different colours irrespective of light intensity. However, findings across species are variable and inconsistent, indicating that our understanding of ant color vision and the role of ecological and phylogenetic factors is still in its early stages. Through compiling studies on molecular, physiological, and behavioral aspects of ant color vision, this review aims to deepen our understanding of color vision and gain insights into the colorful world of ants.
PHILOSOPHICAL TRANSACTIONS OF THE ROYAL SOCIETY B-BIOLOGICAL SCIENCES
(2022)
Article
Physiology
Martin Strube-Bloss, Patrick Guenzel, Carmen A. Nebauer, Johannes Spaethe
Summary: Animals often combine information from different sensory pathways to accurately perceive their environment. The process of multimodal integration is not well understood, but it is believed that the neural representation of individual elements in a multimodal stimulus can enhance or suppress each other's perception. In a study on Western honeybees, it was found that the integration of olfactory and visual stimuli accelerated visual responses but slowed down olfactory response time.
FRONTIERS IN PHYSIOLOGY
(2023)
Article
Computer Science, Artificial Intelligence
G. C. H. E. de Croon, C. De Wagter, T. Seidl
Summary: Researchers have proposed a new method for robots to estimate distances between objects by their visual appearance, which has been successfully implemented on a small flying robot. This approach results in improved performance in tasks such as landing and obstacle avoidance.
NATURE MACHINE INTELLIGENCE
(2021)
Article
Biochemistry & Molecular Biology
Marilia Freire, Antonio Bollig, Markus Knaden
Summary: The elements in the drawings are recyclable and reusable materials, which can be used by artificial intelligence to generate new drawing designs.
Article
Biology
David Baracchi, Martin Giurfa, Patrizia d'Ettorre
Summary: The study found that contrary to the prediction of the speed-accuracy trade-off theory, ants did not trade off speed against accuracy in the process of nestmate recognition.
JOURNAL OF EXPERIMENTAL BIOLOGY
(2021)
Review
Entomology
Devasena Thiagarajan, Silke Sachse
Summary: Insect behaviors provide valuable insights into evolution and organism complexity, and the study of sensory integration that leads to behavior is highly relevant. This review summarizes recent findings on multimodal sensory integration and the resulting behaviors in insects. The sensory systems in insects, including olfaction, vision, and gustation, have a long history of research, and advancements are continuously being made in understanding thermo- and mechano-sensory systems. Specialized senses like hygro- and magneto-reception have also been identified. The combinatorial approach to studying sensory behaviors mimics the real-life environment of insects, and recent advancements have highlighted the importance of multimodal integration in understanding complex insect behaviors. This review focuses on three insect models (honeybees, ants, and flies) and their behaviors, as well as the underlying neuronal mechanisms of sensory integration.
Review
Ecology
Ian W. Keesey
Summary: This article reviews and discusses the neurobiological and genetic research on Drosophila species, exploring the functional variations in sensory system evolution, the roles of neuroecology and neuroevolution in speciation events, and the pressures, mechanisms, and constraints in the natural environments.
FRONTIERS IN ECOLOGY AND EVOLUTION
(2022)
Article
Neurosciences
Kazuaki Ikeda, Masaki Kataoka, Nobuaki K. Tanaka
Summary: Recent studies have revealed the significant role of ephaptic transmission, which does not depend on synapses, in neural information relay and processing. Experiments on female fruit flies show that ephaptic transmission can reach the depth of the brain and mediate light-induced firing rate increases in olfactory sensory neurons.
JOURNAL OF NEUROSCIENCE
(2022)
Article
Computer Science, Artificial Intelligence
Federico Landi, Lorenzo Baraldi, Marcella Cornia, Massimiliano Corsini, Rita Cucchiara
Summary: This paper introduces a novel VLN architecture, PTA, which addresses the issues of multi-modality, long-term dependencies, and adaptability to different locomotive settings in VLN. Through early and late fusion strategies, PTA achieves promising results in low-level VLN.
COMPUTER VISION AND IMAGE UNDERSTANDING
(2021)
Article
Computer Science, Artificial Intelligence
Zhenyu Li, Aiguo Zhou
Summary: Existing deep reinforcement learning-based mobile robot navigation relies on single-modal visual perception for local navigation, while the exploration of multimodal visual fusion-based global navigation is still underway. In this study, we propose a recurrent deduction deep learning model (RDDRL) for multimodal vision-robot navigation to address these challenges. By incorporating a recurrent reasoning mechanism (RRM) into the reinforcement learning model, the agent is able to store memory, predict future transitions, and improve decision-making. The fusion of scene recognition, semantic segmentation, and pose estimation provides partial observations for RDDRL, enabling global-scale behavioral decision-making. Simulation experiments in CARLA scenarios and real-world tests demonstrate that RDDRL outperforms state-of-the-art RL methods in terms of driving stability and safety, with a collision rate decreasing from 0.2% in training state to 0.0% in test state.
APPLIED INTELLIGENCE
(2023)
Article
Biology
Xuelong Sun, Shigang Yue, Michael Mangan
Summary: The research assessed the function of the central complex in insect navigation, identifying a biologically plausible neural transfer mechanism that helps insects to robustly recover and adapt to different environmental disturbances. Additionally, it was proposed that these circuits can be flexibly repurposed by different insect navigators to address their unique ecological needs.
Article
Biology
Thomas S. Collett, Andrew O. Philippides
Summary: Wood ants can learn routes using magnetic directional cues, but their performance can be affected by competing directional cues. Additionally, there is evidence that ants develop magnetically directed home and food vectors based on path integration. However, ants perform worse than honeybees in utilizing magnetic information contextually.
JOURNAL OF EXPERIMENTAL BIOLOGY
(2022)
Article
Computer Science, Artificial Intelligence
Adam Niewola
Summary: This article presents an improved method for mobile robot localization in GPS-denied scenarios. By using Gaussian mixture models and point-by-point extended Kalman filter, high-precision localization can be achieved in rough terrain and outdoor nonurbanized environment.
IEEE INTELLIGENT SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Phil Husbands, Yoonsik Shim, Michael Garvie, Alex Dewar, Norbert Domcsek, Paul Graham, James Knight, Thomas Nowotny, Andrew Philippides
Summary: This paper explores the current developments in evolutionary and bio-inspired approaches to autonomous robotics, with a focus on research conducted at the University of Sussex. It discusses how embodied dynamics can be exploited to create behaviors in robotic systems. Case studies include evolving electronic circuits for robot controllers, navigation methods inspired by insects, neuromechanical chaos for motor behaviors, and brain-body-environment interactions for efficient flapping winged flight.
APPLIED INTELLIGENCE
(2021)
Article
Biology
Xuelong Sun, Shigang Yue, Michael Mangan
Summary: The research assessed the function of the central complex in insect navigation, identifying a biologically plausible neural transfer mechanism that helps insects to robustly recover and adapt to different environmental disturbances. Additionally, it was proposed that these circuits can be flexibly repurposed by different insect navigators to address their unique ecological needs.
Article
Robotics
Anil Ozdemir, Mark Scerri, Andrew B. Barron, Andrew Philippides, Michael Mangan, Eleni Vasilaki, Luca Manneschi
Summary: Recognizing previously visited locations is an important but unsolved task in autonomous navigation. Recent research has shown that Echo State Networks (ESNs) can effectively solve this task and improve performance and generalization abilities.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2022)
Article
Entomology
C. Buehlmann, P. Graham
Summary: This study investigated the impact of different motivational states on the innate visual response of wood ant foragers. The results showed that regardless of the motivational state, wood ant foragers orient towards prominent visual cues, indicating a hardwired behavior that exists across different motivational and ecological contexts.
Article
Biology
Blayze Millward, Steve Maddock, Michael Mangan
Summary: Understanding the functioning of compound eyes is important for biologists and engineers to develop artificial sensory systems and understand visually complex behaviors in nature. Previous research on compound eye models has focused on features like field of view and resolution, overlooking the role of shape and overall structure. This study introduces a new open-source software, CompoundRay, that accurately renders the visual perspective of bees and enables investigation of research questions using modern methods like machine learning.
Article
Mathematical & Computational Biology
Fabian Steinbeck, Thomas Nowotny, Andy Philippides, Paul Graham
Summary: Effective motor behaviors are crucial for navigation in ever-changing environments. Insects have developed adaptive movement patterns to increase their success in achieving navigational goals. A conserved brain area in the insect brain, known as the Lateral Accessory Lobe, is involved in generating small scale search movements to enhance sensory sampling. Spiking Neural Network models inspired by insects can generate a variety of adaptive movement patterns, such as zig-zagging, and are robust to noise.
FRONTIERS IN COMPUTATIONAL NEUROSCIENCE
(2022)
Article
Engineering, Multidisciplinary
Michael Mangan, Dario Floreano, Kotaro Yasui, Barry A. Trimmer, Nick Gravish, Sabine Hauert, Barbara Webb, Poramate Manoonpong, Nicholas Szczecinski
Summary: Invertebrates, with their smaller nervous systems, serve as ideal model systems for robot design principles. They have inspired new materials, geometries and motion control systems for building softer, smaller and lighter robots. Research on their brain circuits has also revealed the structure and function behind their navigation and swarming capabilities. These principles and biomimetic robots have greatly advanced our understanding of animal function.
BIOINSPIRATION & BIOMIMETICS
(2023)
Article
Behavioral Sciences
Sudhakar Deeti, Ken Cheng, Paul Graham, Antoine Wystrach
Summary: Ants engage in scanning bouts to gather visual information for navigation. The start of these scanning bouts and the dynamics of body turns and fixations were examined in Australian desert ants. The findings suggest a random-rate or Poisson process in the timing of these behaviors.
JOURNAL OF COMPARATIVE PHYSIOLOGY A-NEUROETHOLOGY SENSORY NEURAL AND BEHAVIORAL PHYSIOLOGY
(2023)
Article
Behavioral Sciences
Cornelia Buehlmann, Scarlett Dell-Cronin, Angela Diyalagoda Pathirannahelage, Roman Goulard, Barbara Webb, Jeremy E. Niven, Paul Graham
Summary: Wood ants use a combination of innate and learnt navigational strategies to travel between their nest and feeding sites. Mechanical lesions in the central complex of wood ants affect their learned visual guidance and turning control.
JOURNAL OF COMPARATIVE PHYSIOLOGY A-NEUROETHOLOGY SENSORY NEURAL AND BEHAVIORAL PHYSIOLOGY
(2023)
Article
Multidisciplinary Sciences
Lars Haalck, Michael Mangan, Antoine Wystrach, Leo Clement, Barbara Webb, Benjamin Risse
Summary: Tracking the behavior of small animals in complex environments is challenging for computer vision. The CATER methodology combines unsupervised probabilistic detection with global environment reconstruction, allowing for precise quantification of animal behavior in natural settings. Through experiments with desert ants, this method provides insights into navigation and behavior in realistic conditions, bridging the gap between lab and field experiments.
Article
Robotics
Le Zhu, Michael Mangan, Barbara Webb
Summary: For robotics applications, low-power and efficient onboard solutions are highly desirable. This study took inspiration from insects, specifically ants, and developed a neural algorithm for spatiotemporal memory based on a known neural circuit in the insect brain. By combining this algorithm with a bioinspired event camera, the researchers were able to achieve route recognition for visual navigation, which proved to be more robust than traditional SeqSLAM. Additionally, the model was able to evaluate visual familiarity in real time from event camera footage by encoding memory in a spiking neural network running on a neuromorphic computer.
Article
Biology
Franziska Feist, Kiera Smith, Paul Graham
Summary: Herring gulls, one of few species thriving in human-dominated landscapes, were found to pay attention to and make foraging decisions based on human cues in a food-related context. When a demonstrator was seen eating from a specific food item, gulls were significantly more likely to peck on the presented items, particularly the one matching the demonstrator's food. This cross-species social information transfer could be a result of the cognitive flexibility in kleptoparasitic species like herring gulls.
Article
Psychology, Biological
Robert Barrie, Lars Haalck, Benjamin Risse, Thomas Nowotny, Paul Graham, Cornelia Buehlmann
Summary: Research shows that ants rely on individual memory for habitual route guidance even in complex environments, using visual cues as the dominant sensory modality. These findings shed new light on how ants, or insects in general, navigate through complex multimodal environments.
LEARNING & BEHAVIOR
(2023)