Abstract:Motion detectors in the locust optic lobe and brain fall into two categories: neurones that respond selectively to approaching vs. receding objects and neurones that respond selectively to a particular pattern of image motion over a substantial part of the eye, generated by the locust's own movements through its environment. Neurones from the two categories can be differentiated on the basis of their response to motion at a constant velocity at a fixed distance from the locust: neurones of the first category respond equally well to motion in any direction whereas neurones in the second category respond selectively to one preferred direction of motion. Several of the motion detectors of the first category, responding to approaching objects, share the same input organisation, suggesting that it is important in generating a tuning for approaching objects. Anatomical, physiological, and modelling studies have revealed how the selectivity of the response is generated. The selectivity arises as a result of a critical race between excitation, generated when image edges move out over the eye and delayed inhibition, generated by the same edge movements. For excitation to build up, the velocity and extent of edge motion over the eye must increase rapidly. The ultrastructure of the afferent inputs onto the dendrites of collision sensitive neurones reveals a possible substrate for the interaction between excitation and inhibition. This interpretation is supported by both physiological and immunocytochemical evidence. The input organisation of these neurones has been incorporated into the control structure of a small mobile robot, which successfully avoids collisions with looming objects. The ecological role of motion detectors of the second category that respond to image motion over a substantial part of the visual field, is discussed as is the input organisation that generates this selective response. The broad tuning of these neurones, particularly at low velocities (<0.02 degree/s), suggests they may have a role in navigation during migratory flights at altitude. By contrast, their optimum tuning to high-image velocities suggests these motion detectors are adapted for use in a fast flying insect, which does not spend significant time hovering.

Robot-Locust Social Information Transfer Occurs in Predator Avoidance Contexts

Cooperative Flocking And Learning In Multi-Robot Systems For Predator Avoidance

Why Robots Should Be Social: Enhancing Machine Learning through Social Human-Robot Interaction

Beyond Bio-Inspired Robotics: How Multi-Robot Systems Can Support Research on Collective Animal Behavior

Automated optimisation of multi-level models of collective behaviour in a mixed society of animals and robots

Synergistic olfactory processing for social plasticity in desert locusts

Shaping Multi-Robot Patrol Performance with Heterogeneity in Individual Learning Behavior

Translating Virtual Prey-Predator Interaction to Real-World Robotic Environments: Enabling Multimodal Sensing and Evolutionary Dynamics

Motion detectors in the locust visual system: From biology to robot sensors

Memetic robots

Modulating emotional states of rats through a rat-like robot with learned interaction patterns

Evolutionary optimisation of neural network models for fish collective behaviours in mixed groups of robots and zebrafish

Learning from humans to build social cognition among robots

Testing the reality gap with kilobots performing two ant-inspired foraging behaviours

A Study on Social Inclusion of Humanoid Robots: A Novel Embodied Adaptation of the Cyberball Paradigm

Impact of Aging and Cognitive Mechanisms on High-Speed Motor Activation Patterns: Evidence From an Orthoptera-Robot Interaction

A look into the cockpit of the developing locust: Looming detectors and predator avoidance

iCub Knows Where You Look: Exploiting Social Cues for Interactive Object Detection Learning

Quantifying the biomimicry gap in biohybrid robot-fish pairs